I have two hard drives and two cd drives and would like to know the best way to connect them to my mobo. I have both HDDs as primary and secondary masters with the cd's as slaves. Since connecting like this my pc seems to be running really slow and I get 'Pri master capable but disabled' message during start up. Am I doing something wrong?
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations IamaSherpa on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Multiple HDD & CD ROM connection 1
- Thread starter smuuge
- Start date
jeremybarker
Technical User
You should connect both the hds to one channel, master and slave. Also have the cd-roms on a separate channel. The reason your pc is running slow is because having a cd-rom attached to the same channel as a hd slows the hd down.
Hope this helps.
Hope this helps.
-
1
- #3
Jeremybarker is correct, assuming your machine is relatively new (hard drives ATA100 and the CDroms only ATA33)
This unfortunately is a very commonly asked question with no real answer, the problem is idealy all the IDE devices would be better off on there own channels but this is not possible on a normal two channel motherboard unless you have raid or an additional PCI to IDE controller card,you could even go the scisi card route with scisi CDroms.
Best compromise though (as long as the hard drives are of similar spec, both ATA100) would be hard drives together, CDroms together. Martin
Vote if you found this post helpful please!!
This unfortunately is a very commonly asked question with no real answer, the problem is idealy all the IDE devices would be better off on there own channels but this is not possible on a normal two channel motherboard unless you have raid or an additional PCI to IDE controller card,you could even go the scisi card route with scisi CDroms.
Best compromise though (as long as the hard drives are of similar spec, both ATA100) would be hard drives together, CDroms together. Martin
Vote if you found this post helpful please!!
It did just ocur to me to just point out that if your hard drives are ATA 66 or faster they need to be connected to the motherboard with the newer/finer 80 core 40pin ultra cable, these have been around for sometime now and are supplied with all new motherboards just don't get it mixed up with the slower 40 core cable the CDroms devices use. Martin Vote if you found this post helpful please!!
I have to disagree with the other posters on device setup,
I run HD primary with CD slave with no problems, however that is assuming that your cd is ata 33. I have 2 hd, cdrw, dvd-rom, they are crossed matched as well. paparazi is correct in that you need the ATA66/100/133 cables to get full data xfer speeds, you also need to insure that you have the current drivers, also make sure that you have both IDE channels enabled in the bios, and windows (I'm assuming you are running windows) and that your drives are set up correctly in the bios (ata100 should be set to 5), if VIA make sure the 4-1 drivers are the latest, if SIS you might be better with MS drivers, I'm not familiar with the other chipsets, also check device manager and check that there are no yellow or red markers (resolve them), click on each device and make sure that dma is checked, drop down to system devices and enable dma and enter reserve 64k. That should clear up most of performance problems. (w2k is slightly different)
There are several reasons why the other posters recommend what they said, first is that PIO will run at the slowest device speed, a lot of cd-roms are pio 3 or 4 (old hd are pio or pre ata with same problem), most current cd/dvd/cdrw are ata33, as long as both devices are ata enabled then there is no hit on performance for single devices. However if you use both devices on the same channel at the same time there will be a hit on performance, which is why you were set up correctly with both hard drives as primary (at least the way I use dual hd). Look at it this way, when you have both hard drive on the primary channel as master and slave, you can only access one device at a time, so if you are xfering data from hd1 to hd2, then the channel can only accomodate one drive at a time, hd1 is sending data thru the cable into the ide bus, then hd1 waits while the data turns around and goes to hd2 (twice as long), if you had both hard drives as primary, then both drives can work concurrently and performance will be optimal. The same is true of the other devices, ex: you burn a lot from cd's then you are better off with cdrw and cd/dvd devices on separate channels, just think about how you will use devices and set it up accordingly.
One other fact, if you have seagate barracuda IV on windows 2000 it will be slow, according to seagate, the drive is so fast that it has to double check everything which is why its slow on windows 2000 (lol, sounds like a bunch of crock), btw I dual boot as well and its one of the fastest drives under win98.
Still confused? Sorry if the explanations weren't as clear, I know I sometimes lose people on this, just let me know.
Good Luck!!
I run HD primary with CD slave with no problems, however that is assuming that your cd is ata 33. I have 2 hd, cdrw, dvd-rom, they are crossed matched as well. paparazi is correct in that you need the ATA66/100/133 cables to get full data xfer speeds, you also need to insure that you have the current drivers, also make sure that you have both IDE channels enabled in the bios, and windows (I'm assuming you are running windows) and that your drives are set up correctly in the bios (ata100 should be set to 5), if VIA make sure the 4-1 drivers are the latest, if SIS you might be better with MS drivers, I'm not familiar with the other chipsets, also check device manager and check that there are no yellow or red markers (resolve them), click on each device and make sure that dma is checked, drop down to system devices and enable dma and enter reserve 64k. That should clear up most of performance problems. (w2k is slightly different)
There are several reasons why the other posters recommend what they said, first is that PIO will run at the slowest device speed, a lot of cd-roms are pio 3 or 4 (old hd are pio or pre ata with same problem), most current cd/dvd/cdrw are ata33, as long as both devices are ata enabled then there is no hit on performance for single devices. However if you use both devices on the same channel at the same time there will be a hit on performance, which is why you were set up correctly with both hard drives as primary (at least the way I use dual hd). Look at it this way, when you have both hard drive on the primary channel as master and slave, you can only access one device at a time, so if you are xfering data from hd1 to hd2, then the channel can only accomodate one drive at a time, hd1 is sending data thru the cable into the ide bus, then hd1 waits while the data turns around and goes to hd2 (twice as long), if you had both hard drives as primary, then both drives can work concurrently and performance will be optimal. The same is true of the other devices, ex: you burn a lot from cd's then you are better off with cdrw and cd/dvd devices on separate channels, just think about how you will use devices and set it up accordingly.
One other fact, if you have seagate barracuda IV on windows 2000 it will be slow, according to seagate, the drive is so fast that it has to double check everything which is why its slow on windows 2000 (lol, sounds like a bunch of crock), btw I dual boot as well and its one of the fastest drives under win98.
Still confused? Sorry if the explanations weren't as clear, I know I sometimes lose people on this, just let me know.
Good Luck!!
CitrixEngineer
Technical User
fastoon:
You offer some good advice, especially about making sure the disk controller and chipset drivers are up to date, but I think there are holes in your IDE theories - please feel free to correct me with links to authoritative sites.
Quote:/ "as long as both devices are ata enabled then there is no hit on performance for single devices." /:Quote
It is my understanding (and I could well be wrong!) that the different ATA protocols only communicate at the lowest common denominator; ie if you stick an ATA33 and an ATA133 device on the same physical cable, they will both use ATA33, so the faster drive will be effectively slowed down.
So a performance hit is expected on the faster device.
Quote:/"Look at it this way, when you have both hard drive on the primary channel as master and slave, you can only access one device at a time, so if you are xfering data from hd1 to hd2, then the channel can only accomodate one drive at a time, hd1 is sending data thru the cable into the ide bus, then hd1 waits while the data turns around and goes to hd2 (twice as long), if you had both hard drives as primary, then both drives can work concurrently and performance will be optimal." /:Quote
Not true; the 80-channel IDE cable is perfectly capable of shifting data from both devices at once without noticeable degradation, as is the old 40-pin cable. There is a wait state in the electronics, it's true, but that is in terms of milliseconds, and is because true multitasking does not exist in a Windows environment. It is not true to say that putting ATA/ATAPI devices onto separate IDE cables will speed up data transfer.
What will slow down data transfer is 2 things;
1. Putting a slower AT protocol device onto the same channel as a fast one.
2. Putting a Packet Interface (as in ATA_PI = CD/DVD) device onto the same channel as an LBA device (any ATA HDD, ie any Hard Disk 540Mb or greater).
To smuuge, I would do as Paparazi and jeremybarker suggest; put the 2 Hdds on the 1st IDE channel, as master and slave, and put the 2 opticals on the 2nd IDE. Next, autodetect everything in the BIOS to ensure that the system can recognise the drives. Then do as fastoon suggests = upgrade to latest driver revisions.
I hope this helps CitrixEngineer@yahoo.co.uk
You offer some good advice, especially about making sure the disk controller and chipset drivers are up to date, but I think there are holes in your IDE theories - please feel free to correct me with links to authoritative sites.
Quote:/ "as long as both devices are ata enabled then there is no hit on performance for single devices." /:Quote
It is my understanding (and I could well be wrong!) that the different ATA protocols only communicate at the lowest common denominator; ie if you stick an ATA33 and an ATA133 device on the same physical cable, they will both use ATA33, so the faster drive will be effectively slowed down.
So a performance hit is expected on the faster device.
Quote:/"Look at it this way, when you have both hard drive on the primary channel as master and slave, you can only access one device at a time, so if you are xfering data from hd1 to hd2, then the channel can only accomodate one drive at a time, hd1 is sending data thru the cable into the ide bus, then hd1 waits while the data turns around and goes to hd2 (twice as long), if you had both hard drives as primary, then both drives can work concurrently and performance will be optimal." /:Quote
Not true; the 80-channel IDE cable is perfectly capable of shifting data from both devices at once without noticeable degradation, as is the old 40-pin cable. There is a wait state in the electronics, it's true, but that is in terms of milliseconds, and is because true multitasking does not exist in a Windows environment. It is not true to say that putting ATA/ATAPI devices onto separate IDE cables will speed up data transfer.
What will slow down data transfer is 2 things;
1. Putting a slower AT protocol device onto the same channel as a fast one.
2. Putting a Packet Interface (as in ATA_PI = CD/DVD) device onto the same channel as an LBA device (any ATA HDD, ie any Hard Disk 540Mb or greater).
To smuuge, I would do as Paparazi and jeremybarker suggest; put the 2 Hdds on the 1st IDE channel, as master and slave, and put the 2 opticals on the 2nd IDE. Next, autodetect everything in the BIOS to ensure that the system can recognise the drives. Then do as fastoon suggests = upgrade to latest driver revisions.
I hope this helps CitrixEngineer@yahoo.co.uk
Hi CitrixEngineer,
Thanks for your input, based on my knowledge of performance with multiple devices on the same IDE channel I have only read about reference to PIO devices and performance problems, all material detailing ATA suggest they are compatible and do not suffer a perfomance degradation unless set up in a manner different from what I have suggested.
I would like to ask you to point me in the right direction to verify your statements or your dispute of my experience/advice, of course on these particular topic it has been very difficult for me to find any relevant information. So if you can help in that regard I would appreciate it.
Most of my opinion/observation has to do with personal experience, I've built multiple computers usually using 4 or more devices, I prefer an ATA or RAID card to get all devices on their own channel to avoid this very problem. In my experience using CD/DVD player and CDRW player on the same channel regardless of which order its in will result in coasters or with burn proof technology will take approx 3-6 times as long to complete. I also regulary transfer large files between hard drives, if both hard drives are on the same channel it has taken up to 5-10 times longer than if both drives are on separate channels, this experience points to the pci bus as the bottleneck.
------------------------------------------------------
Since you seem to be an authority - I decided to do a quick search on the web for any information, still difficult, but I'm no longer looking for merely technical information for the enjoyment and advancement of my own knowledge. I'm looking for specifics on your debate (if you've been there than you know about discrediting the idea based on other facts that are distorted regardless if the overall issue has merit or not)
Please review and comment on the following statement.
1. "3.0 PCI Bus Protocol PCI is a synchronous bus architecture with all data transfers being performed relative to a system clock (CLK). The initial PCI specification permitted a maximum clock rate of 33 MHz...
>>>allowing one bus transfer to be performed every 30 nanoseconds<<<<
...Later, Revision 2.1 of the PCI specification extended the bus definition to support operation at 66 MHz, but the vast majority of today's personal computers continue to implement a PCI bus that runs at a maximum speed of 33 MHz."
The preceding quote was taken from TechFest
As you can see only one bus transfer can be performed at a time, even if it is very fast in this case 30 nanoseconds, it still is only one way, add up all the data and that 30 nanosecond hit starts to add up to milliseconds, then seconds, then minutes, etc., but you get the picture.
----
2. "Something important to keep in mind is the possible conflict that attaching devices of differing interface characteristics has on overall system performance...
(my comments - the above is what you , paparazi, and jeremybarker are saying and it is true up to a point.)
... At one time, attaching a fast disk drive as a master and a slow CD ROM drive as a slave on the same IDE channel would pull that channel's overall performance down to the level of the slower device. ...
(my comments - again isn't this what you're saying and recommending today?, please reread statement that starts "At one time..." hmmmmm)
Early PIIX controllers put severe limitations on the configuration of IDE modes. There could be only two modes available. So, if both devices couldn't operate at the fastest mode supported by the controller, either both would run at the speed of the slowest, or one of them would run at PIO mode 0 regardless of its capabilities. The first PIIX imposed this limitation even across channels. The PIIX3 removed the limitation across channels, but not across devices in the same channel. The latest controllers with UDMA support (PIIX4 and PIIX4E) have removed these limits completely, and the mode can be configured independently for each device.
>>This should not be an issue any longer.<<
(my comments - exactly what I'm saying, say I wish I had come across this info earlier instead of doing it the hard way of building, rebuilding, stripping down, upgrading, playing/hacking drivers, adding/removing drives, etc. hmmmmm, oh well, lol, in any case I'm the experimental type and enjoy the process and the results.)
However, the software drivers may not all be taking advantage of the hardware improvement (We can't confirm if the drivers shipping with Win98SE take advantage of this improvement in channel mode selection, but Win2K does for sure). Even though this may be a dead issue, it doesn't hurt to avoid connecting a hard drive and a CD ROM on the same IDE channel unless the CD ROM does support UDMA (has a DMA checkbox in Device Manager like the hard drive) and it is enabled."
(my comment - WOW! exactly what I said. Well not exactly but the same conclusion and recommendation! Who would've thunk!)
The preceding quote was taken from: SCSI vs. IDE Bus Mastering for DAWs, Part 4 by D. Glen Cardenas and Jose M. Catena.
----
I don't know how autoritative these sources are for your approval, of course I'm not looking for your approval.
Please inform me about any inconsistancy/errors in latest post from an authoritative source. (I thought I'd throw that back at you, lol)
BTW I'm not in the tech field, this is just a hobby, however I do know what I'm speaking of from experience.
I have taken the A+ cert cold and aced the first part (of which the bus, pci, ide, and others is included although not this particular issue), failed the second part (os), in reviewing current A+ cert books I have concluded based on the information provided that my position is correct. Keep in mind that the older A+ cert books that I've read does have outmoded information. Again this is just a hobby and for entertainment purposes, I do build and troubleshoot computer systems for friends and relatives. Could it be that you in the field are not keeping up to date? Please keep in mind that there have been many changes and while it may not be pertinent to your job/career, it does show how committed/dedicated you are to your job/career. I spend a considerable amount of time just keeping current in my career, I spend at least 100 hours if not more per year in keeping up with all the new regulations, in my industry the regulations mandate a minimum of 30 hours continuing education per year, however that is just for the minimum, people who have no ambition, who don't want to provide that little bit more to differentiate themselves from others in their field will only do the minimum to get by and keep their jobs. I hope that is not you. I would not think to get into a debate regarding your specialization nor imply that you are not doing your job. And while this issue is one that most people seem to misunderstand, for someone that is in the industry and does make a living at this, you seem to be way out of touch, or am I misunderstanding what you are saying?
I wish you well in your endeavers.
Thanks for your input, based on my knowledge of performance with multiple devices on the same IDE channel I have only read about reference to PIO devices and performance problems, all material detailing ATA suggest they are compatible and do not suffer a perfomance degradation unless set up in a manner different from what I have suggested.
I would like to ask you to point me in the right direction to verify your statements or your dispute of my experience/advice, of course on these particular topic it has been very difficult for me to find any relevant information. So if you can help in that regard I would appreciate it.
Most of my opinion/observation has to do with personal experience, I've built multiple computers usually using 4 or more devices, I prefer an ATA or RAID card to get all devices on their own channel to avoid this very problem. In my experience using CD/DVD player and CDRW player on the same channel regardless of which order its in will result in coasters or with burn proof technology will take approx 3-6 times as long to complete. I also regulary transfer large files between hard drives, if both hard drives are on the same channel it has taken up to 5-10 times longer than if both drives are on separate channels, this experience points to the pci bus as the bottleneck.
------------------------------------------------------
Since you seem to be an authority - I decided to do a quick search on the web for any information, still difficult, but I'm no longer looking for merely technical information for the enjoyment and advancement of my own knowledge. I'm looking for specifics on your debate (if you've been there than you know about discrediting the idea based on other facts that are distorted regardless if the overall issue has merit or not)
Please review and comment on the following statement.
1. "3.0 PCI Bus Protocol PCI is a synchronous bus architecture with all data transfers being performed relative to a system clock (CLK). The initial PCI specification permitted a maximum clock rate of 33 MHz...
>>>allowing one bus transfer to be performed every 30 nanoseconds<<<<
...Later, Revision 2.1 of the PCI specification extended the bus definition to support operation at 66 MHz, but the vast majority of today's personal computers continue to implement a PCI bus that runs at a maximum speed of 33 MHz."
The preceding quote was taken from TechFest
As you can see only one bus transfer can be performed at a time, even if it is very fast in this case 30 nanoseconds, it still is only one way, add up all the data and that 30 nanosecond hit starts to add up to milliseconds, then seconds, then minutes, etc., but you get the picture.
----
2. "Something important to keep in mind is the possible conflict that attaching devices of differing interface characteristics has on overall system performance...
(my comments - the above is what you , paparazi, and jeremybarker are saying and it is true up to a point.)
... At one time, attaching a fast disk drive as a master and a slow CD ROM drive as a slave on the same IDE channel would pull that channel's overall performance down to the level of the slower device. ...
(my comments - again isn't this what you're saying and recommending today?, please reread statement that starts "At one time..." hmmmmm)
Early PIIX controllers put severe limitations on the configuration of IDE modes. There could be only two modes available. So, if both devices couldn't operate at the fastest mode supported by the controller, either both would run at the speed of the slowest, or one of them would run at PIO mode 0 regardless of its capabilities. The first PIIX imposed this limitation even across channels. The PIIX3 removed the limitation across channels, but not across devices in the same channel. The latest controllers with UDMA support (PIIX4 and PIIX4E) have removed these limits completely, and the mode can be configured independently for each device.
>>This should not be an issue any longer.<<
(my comments - exactly what I'm saying, say I wish I had come across this info earlier instead of doing it the hard way of building, rebuilding, stripping down, upgrading, playing/hacking drivers, adding/removing drives, etc. hmmmmm, oh well, lol, in any case I'm the experimental type and enjoy the process and the results.)
However, the software drivers may not all be taking advantage of the hardware improvement (We can't confirm if the drivers shipping with Win98SE take advantage of this improvement in channel mode selection, but Win2K does for sure). Even though this may be a dead issue, it doesn't hurt to avoid connecting a hard drive and a CD ROM on the same IDE channel unless the CD ROM does support UDMA (has a DMA checkbox in Device Manager like the hard drive) and it is enabled."
(my comment - WOW! exactly what I said. Well not exactly but the same conclusion and recommendation! Who would've thunk!)
The preceding quote was taken from: SCSI vs. IDE Bus Mastering for DAWs, Part 4 by D. Glen Cardenas and Jose M. Catena.
----
I don't know how autoritative these sources are for your approval, of course I'm not looking for your approval.
Please inform me about any inconsistancy/errors in latest post from an authoritative source. (I thought I'd throw that back at you, lol)
BTW I'm not in the tech field, this is just a hobby, however I do know what I'm speaking of from experience.
I have taken the A+ cert cold and aced the first part (of which the bus, pci, ide, and others is included although not this particular issue), failed the second part (os), in reviewing current A+ cert books I have concluded based on the information provided that my position is correct. Keep in mind that the older A+ cert books that I've read does have outmoded information. Again this is just a hobby and for entertainment purposes, I do build and troubleshoot computer systems for friends and relatives. Could it be that you in the field are not keeping up to date? Please keep in mind that there have been many changes and while it may not be pertinent to your job/career, it does show how committed/dedicated you are to your job/career. I spend a considerable amount of time just keeping current in my career, I spend at least 100 hours if not more per year in keeping up with all the new regulations, in my industry the regulations mandate a minimum of 30 hours continuing education per year, however that is just for the minimum, people who have no ambition, who don't want to provide that little bit more to differentiate themselves from others in their field will only do the minimum to get by and keep their jobs. I hope that is not you. I would not think to get into a debate regarding your specialization nor imply that you are not doing your job. And while this issue is one that most people seem to misunderstand, for someone that is in the industry and does make a living at this, you seem to be way out of touch, or am I misunderstanding what you are saying?
I wish you well in your endeavers.
CitrixEngineer
Technical User
Wow, what a long post! I will certainly address every single technical issue you raise at a point where I have a bit more time.
I'd just like to make a few observations for now;
1. Technical debate is healthy - and expected on a forum like this. I make no personal attacks, but am quick to question anything which conflicts with my understanding. I'm glad you have decided to challenge my ideas!
2. I do not consider myself an authority, but am aware that my style of writing comes across as authoritative. I have been working with computers for a very long time, but am always ready to learn something new. I may well be out of touch - technology moves quickly.
3. I am not always right, and welcome corrections in my observations of technology on my path to greater understanding of how it all works.
Here's a really good technical and authoritative link on the subject in hand;
It confirms, to me, at least, that putting CD devices and HDDs on the same channel is becoming less of an issue - but is an issue I've "grown up with".
Here's the quote from this page, for anyone who doesn't want to trawl the whole article;
"The lack of independent device timing on many older systems is one reason why placing ATAPI devices like CD-ROMs on the same channel as a fast hard disk is usually not recommended."
I'll read your posting more carefully later, and give you the fuller reply that such a detailed response deserves
CitrixEngineer@yahoo.co.uk
I'd just like to make a few observations for now;
1. Technical debate is healthy - and expected on a forum like this. I make no personal attacks, but am quick to question anything which conflicts with my understanding. I'm glad you have decided to challenge my ideas!
2. I do not consider myself an authority, but am aware that my style of writing comes across as authoritative. I have been working with computers for a very long time, but am always ready to learn something new. I may well be out of touch - technology moves quickly.
3. I am not always right, and welcome corrections in my observations of technology on my path to greater understanding of how it all works.
Here's a really good technical and authoritative link on the subject in hand;
It confirms, to me, at least, that putting CD devices and HDDs on the same channel is becoming less of an issue - but is an issue I've "grown up with".
Here's the quote from this page, for anyone who doesn't want to trawl the whole article;
"The lack of independent device timing on many older systems is one reason why placing ATAPI devices like CD-ROMs on the same channel as a fast hard disk is usually not recommended."
I'll read your posting more carefully later, and give you the fuller reply that such a detailed response deserves
CitrixEngineer@yahoo.co.uk
CitrixEngineer
Technical User
With apologies for the length of this reply – I thought it would be easier to copy the original posting and make comments >indented, than attempt to answer in a referential fashion. I have attempted to make as full a response as time permitted, and welcome any comments on technical inaccuracies.
Thanks for your input, based on my knowledge of performance with multiple devices on the same IDE channel I have only read about reference to PIO devices and performance problems, all material detailing ATA suggest they are compatible and do not suffer a perfomance degradation unless set up in a manner different from what I have suggested.
>We all draw on what we have seen and experienced
I would like to ask you to point me in the right direction to verify your statements or your dispute of my experience/advice, of course on these particular topic it has been very difficult for me to find any relevant information. So if you can help in that regard I would appreciate it.
>I have offered the link to pcguide.com as an authoritative reference, but can find others at an equal level, if desired.
Most of my opinion/observation has to do with personal experience, I've built multiple computers usually using 4 or more devices, I prefer an ATA or RAID card to get all devices on their own channel to avoid this very problem. In my experience using CD/DVD player and CDRW player on the same channel regardless of which order its in will result in coasters or with burn proof technology will take approx 3-6 times as long to complete. I also regulary transfer large files between hard drives, if both hard drives are on the same channel it has taken up to 5-10 times longer than if both drives are on separate channels, this experience points to the pci bus as the bottleneck.
>I have not found this, with a well defragmented hard disk. On the contrary, putting the two ATAPI devices on the same channel enhances CD copying, IME. We will have to agree to disagree on this
------------------------------------------------------
Since you seem to be an authority - I decided to do a quick search on the web for any information, still difficult, but I'm no longer looking for merely technical information for the enjoyment and advancement of my own knowledge. I'm looking for specifics on your debate (if you've been there than you know about discrediting the idea based on other facts that are distorted regardless if the overall issue has merit or not)
Please review and comment on the following statement.
1. "3.0 PCI Bus Protocol PCI is a synchronous bus architecture with all data transfers being performed relative to a system clock (CLK). The initial PCI specification permitted a maximum clock rate of 33 MHz...
>>>allowing one bus transfer to be performed every 30 nanoseconds<<<<
...Later, Revision 2.1 of the PCI specification extended the bus definition to support operation at 66 MHz, but the vast majority of today's personal computers continue to implement a PCI bus that runs at a maximum speed of 33 MHz."
The preceding quote was taken from TechFest
As you can see only one bus transfer can be performed at a time, even if it is very fast in this case 30 nanoseconds, it still is only one way, add up all the data and that 30 nanosecond hit starts to add up to milliseconds, then seconds, then minutes, etc., but you get the picture.
>PCI 3.0 is still in the pipeline, ie it has not yet been implemented. The rest of what you say does not contradict me, so does not constitute argument; Yes, only one bus transfer can be performed at a time – this is what interrupts and latency timings are for. I do not dispute this.
I note that, although techfest is a good reference to be used among others, this article is over two years old, and contains no reference to the current PCI 2.3 standard.
Please review this site, which is the authority on the PCI bus;
And
----
2. "Something important to keep in mind is the possible conflict that attaching devices of differing interface characteristics has on overall system performance...
(my comments - the above is what you , paparazi, and jeremybarker are saying and it is true up to a point.)
... At one time, attaching a fast disk drive as a master and a slow CD ROM drive as a slave on the same IDE channel would pull that channel's overall performance down to the level of the slower device. ...
(my comments - again isn't this what you're saying and recommending today?, please reread statement that starts "At one time..." hmmmmm)
>Agreed that this is a historical thing – I’ve been working with computers for nearly 20 years now, so old ways do tend to become ingrained. However, if you went to my pcguide link in my earlier posting, you will note that it is “becoming less of an issue”. Ie it is still an issue to some extent. Where did your quote come from?
Early PIIX controllers put severe limitations on the configuration of IDE modes. There could be only two modes available. So, if both devices couldn't operate at the fastest mode supported by the controller, either both would run at the speed of the slowest, or one of them would run at PIO mode 0 regardless of its capabilities. The first PIIX imposed this limitation even across channels. The PIIX3 removed the limitation across channels, but not across devices in the same channel. The latest controllers with UDMA support (PIIX4 and PIIX4E) have removed these limits completely, and the mode can be configured independently for each device.
>>This should not be an issue any longer.<<
>Maybe…I note your use of the word “should”. Where does this information come from? I refer again to pcguide.com
Your references are exclusively to the Intel PIIX series, and do not take into account Via, SIS, Ali, AMD, etc. PCI to IDE is just one function of the South Bridge chipset. This function does not take into account optimisations of the Disk controllers themselves, particularly add-ons like the Highpoint RAID cards, or the on-board IDE controller itself. Please make the distinction between the disk controller and the South Bridge when making these sorts of references.
(my comments - exactly what I'm saying, say I wish I had come across this info earlier instead of doing it the hard way of building, rebuilding, stripping down, upgrading, playing/hacking drivers, adding/removing drives, etc. hmmmmm, oh well, lol, in any case I'm the experimental type and enjoy the process and the results.)
However, the software drivers may not all be taking advantage of the hardware improvement (We can't confirm if the drivers shipping with Win98SE take advantage of this improvement in channel mode selection, but Win2K does for sure). Even though this may be a dead issue, it doesn't hurt to avoid connecting a hard drive and a CD ROM on the same IDE channel unless the CD ROM does support UDMA (has a DMA checkbox in Device Manager like the hard drive) and it is enabled."
(my comment - WOW! exactly what I said. Well not exactly but the same conclusion and recommendation! Who would've thunk!)
The preceding quote was taken from: SCSI vs. IDE Bus Mastering for DAWs, Part 4 by D. Glen Cardenas and Jose M. Catena.
>Your quote and ensuing comment seems a little confused to me – this is hardly a drawn conclusion.
Your source, digitalprosound is not one I would generally recognise as authoritative for computer-specific, much less hard-disk specific information, and I am not familiar with the names of the two authors. FYI I have been creating music on computers since the BBC B days, and work part-time in a Virgin-sponsored professional recording studio, so have a keen interest in digital sound recording.
The article, although long and generally at a reasonable technical level, is flawed in many places. For example on the first page, second paragraph “Well, even though there may be little advantage to either EIDE or SCSI in most system configurations…” I think most participants in this forum can see the flaw in this sentence. I found more flaws as the article progressed, and note that many links go to the author’s own pages rather than technical sources.
I notice that even their own definition of IDE is flawed; “Stands for Integrated Drive Electronics. Common name given to the ATA disk drive format popularly used in PCs today. Usually connects directly to the Mother Board. EIDE is Enhanced IDE. Drives capable of bus mastering are EIDE drives.” IDE is not exclusive to ATA, as this implies.
---I don't know how autoritative these sources are for your approval, of course I'm not looking for your approval.
Please inform me about any inconsistancy/errors in latest post from an authoritative source. (I thought I'd throw that back at you, lol)
>Is this enough information?
BTW I'm not in the tech field, this is just a hobby, however I do know what I'm speaking of from experience.
>Good for you – I started out as a hobbyist. Keep it up – and don’t mind pedantic old techs like me ;-)
I have taken the A+ cert cold and aced the first part (of which the bus, pci, ide, and others is included although not this particular issue), failed the second part (os), in reviewing current A+ cert books I have concluded based on the information provided that my position is correct. Keep in mind that the older A+ cert books that I've read does have outmoded information. Again this is just a hobby and for entertainment purposes, I do build and troubleshoot computer systems for friends and relatives. Could it be that you in the field are not keeping up to date? Please keep in mind that there have been many changes and while it may not be pertinent to your job/career, it does show how committed/dedicated you are to your job/career. I spend a considerable amount of time just keeping current in my career, I spend at least 100 hours if not more per year in keeping up with all the new regulations, in my industry the regulations mandate a minimum of 30 hours continuing education per year, however that is just for the minimum, people who have no ambition, who don't want to provide that little bit more to differentiate themselves from others in their field will only do the minimum to get by and keep their jobs. I hope that is not you. I would not think to get into a debate regarding your specialization nor imply that you are not doing your job. And while this issue is one that most people seem to misunderstand, for someone that is in the industry and does make a living at this, you seem to be way out of touch, or am I misunderstanding what you are saying?
>A+ is a recognised qualification, and a very good base for further technical development, and the information is getting better. However, industry certs are just that – a base to build upon with real-world experience, which can often prove the theories drastically flawed. I think my information is reasonably up-to-date (ref the two links I have included in this critique).
I wish you well in your endeavers.
>And the same to you
CitrixEngineer@yahoo.co.uk
Thanks for your input, based on my knowledge of performance with multiple devices on the same IDE channel I have only read about reference to PIO devices and performance problems, all material detailing ATA suggest they are compatible and do not suffer a perfomance degradation unless set up in a manner different from what I have suggested.
>We all draw on what we have seen and experienced
I would like to ask you to point me in the right direction to verify your statements or your dispute of my experience/advice, of course on these particular topic it has been very difficult for me to find any relevant information. So if you can help in that regard I would appreciate it.
>I have offered the link to pcguide.com as an authoritative reference, but can find others at an equal level, if desired.
Most of my opinion/observation has to do with personal experience, I've built multiple computers usually using 4 or more devices, I prefer an ATA or RAID card to get all devices on their own channel to avoid this very problem. In my experience using CD/DVD player and CDRW player on the same channel regardless of which order its in will result in coasters or with burn proof technology will take approx 3-6 times as long to complete. I also regulary transfer large files between hard drives, if both hard drives are on the same channel it has taken up to 5-10 times longer than if both drives are on separate channels, this experience points to the pci bus as the bottleneck.
>I have not found this, with a well defragmented hard disk. On the contrary, putting the two ATAPI devices on the same channel enhances CD copying, IME. We will have to agree to disagree on this
------------------------------------------------------
Since you seem to be an authority - I decided to do a quick search on the web for any information, still difficult, but I'm no longer looking for merely technical information for the enjoyment and advancement of my own knowledge. I'm looking for specifics on your debate (if you've been there than you know about discrediting the idea based on other facts that are distorted regardless if the overall issue has merit or not)
Please review and comment on the following statement.
1. "3.0 PCI Bus Protocol PCI is a synchronous bus architecture with all data transfers being performed relative to a system clock (CLK). The initial PCI specification permitted a maximum clock rate of 33 MHz...
>>>allowing one bus transfer to be performed every 30 nanoseconds<<<<
...Later, Revision 2.1 of the PCI specification extended the bus definition to support operation at 66 MHz, but the vast majority of today's personal computers continue to implement a PCI bus that runs at a maximum speed of 33 MHz."
The preceding quote was taken from TechFest
As you can see only one bus transfer can be performed at a time, even if it is very fast in this case 30 nanoseconds, it still is only one way, add up all the data and that 30 nanosecond hit starts to add up to milliseconds, then seconds, then minutes, etc., but you get the picture.
>PCI 3.0 is still in the pipeline, ie it has not yet been implemented. The rest of what you say does not contradict me, so does not constitute argument; Yes, only one bus transfer can be performed at a time – this is what interrupts and latency timings are for. I do not dispute this.
I note that, although techfest is a good reference to be used among others, this article is over two years old, and contains no reference to the current PCI 2.3 standard.
Please review this site, which is the authority on the PCI bus;
And
----
2. "Something important to keep in mind is the possible conflict that attaching devices of differing interface characteristics has on overall system performance...
(my comments - the above is what you , paparazi, and jeremybarker are saying and it is true up to a point.)
... At one time, attaching a fast disk drive as a master and a slow CD ROM drive as a slave on the same IDE channel would pull that channel's overall performance down to the level of the slower device. ...
(my comments - again isn't this what you're saying and recommending today?, please reread statement that starts "At one time..." hmmmmm)
>Agreed that this is a historical thing – I’ve been working with computers for nearly 20 years now, so old ways do tend to become ingrained. However, if you went to my pcguide link in my earlier posting, you will note that it is “becoming less of an issue”. Ie it is still an issue to some extent. Where did your quote come from?
Early PIIX controllers put severe limitations on the configuration of IDE modes. There could be only two modes available. So, if both devices couldn't operate at the fastest mode supported by the controller, either both would run at the speed of the slowest, or one of them would run at PIO mode 0 regardless of its capabilities. The first PIIX imposed this limitation even across channels. The PIIX3 removed the limitation across channels, but not across devices in the same channel. The latest controllers with UDMA support (PIIX4 and PIIX4E) have removed these limits completely, and the mode can be configured independently for each device.
>>This should not be an issue any longer.<<
>Maybe…I note your use of the word “should”. Where does this information come from? I refer again to pcguide.com
Your references are exclusively to the Intel PIIX series, and do not take into account Via, SIS, Ali, AMD, etc. PCI to IDE is just one function of the South Bridge chipset. This function does not take into account optimisations of the Disk controllers themselves, particularly add-ons like the Highpoint RAID cards, or the on-board IDE controller itself. Please make the distinction between the disk controller and the South Bridge when making these sorts of references.
(my comments - exactly what I'm saying, say I wish I had come across this info earlier instead of doing it the hard way of building, rebuilding, stripping down, upgrading, playing/hacking drivers, adding/removing drives, etc. hmmmmm, oh well, lol, in any case I'm the experimental type and enjoy the process and the results.)
However, the software drivers may not all be taking advantage of the hardware improvement (We can't confirm if the drivers shipping with Win98SE take advantage of this improvement in channel mode selection, but Win2K does for sure). Even though this may be a dead issue, it doesn't hurt to avoid connecting a hard drive and a CD ROM on the same IDE channel unless the CD ROM does support UDMA (has a DMA checkbox in Device Manager like the hard drive) and it is enabled."
(my comment - WOW! exactly what I said. Well not exactly but the same conclusion and recommendation! Who would've thunk!)
The preceding quote was taken from: SCSI vs. IDE Bus Mastering for DAWs, Part 4 by D. Glen Cardenas and Jose M. Catena.
>Your quote and ensuing comment seems a little confused to me – this is hardly a drawn conclusion.
Your source, digitalprosound is not one I would generally recognise as authoritative for computer-specific, much less hard-disk specific information, and I am not familiar with the names of the two authors. FYI I have been creating music on computers since the BBC B days, and work part-time in a Virgin-sponsored professional recording studio, so have a keen interest in digital sound recording.
The article, although long and generally at a reasonable technical level, is flawed in many places. For example on the first page, second paragraph “Well, even though there may be little advantage to either EIDE or SCSI in most system configurations…” I think most participants in this forum can see the flaw in this sentence. I found more flaws as the article progressed, and note that many links go to the author’s own pages rather than technical sources.
I notice that even their own definition of IDE is flawed; “Stands for Integrated Drive Electronics. Common name given to the ATA disk drive format popularly used in PCs today. Usually connects directly to the Mother Board. EIDE is Enhanced IDE. Drives capable of bus mastering are EIDE drives.” IDE is not exclusive to ATA, as this implies.
---I don't know how autoritative these sources are for your approval, of course I'm not looking for your approval.
Please inform me about any inconsistancy/errors in latest post from an authoritative source. (I thought I'd throw that back at you, lol)
>Is this enough information?
BTW I'm not in the tech field, this is just a hobby, however I do know what I'm speaking of from experience.
>Good for you – I started out as a hobbyist. Keep it up – and don’t mind pedantic old techs like me ;-)
I have taken the A+ cert cold and aced the first part (of which the bus, pci, ide, and others is included although not this particular issue), failed the second part (os), in reviewing current A+ cert books I have concluded based on the information provided that my position is correct. Keep in mind that the older A+ cert books that I've read does have outmoded information. Again this is just a hobby and for entertainment purposes, I do build and troubleshoot computer systems for friends and relatives. Could it be that you in the field are not keeping up to date? Please keep in mind that there have been many changes and while it may not be pertinent to your job/career, it does show how committed/dedicated you are to your job/career. I spend a considerable amount of time just keeping current in my career, I spend at least 100 hours if not more per year in keeping up with all the new regulations, in my industry the regulations mandate a minimum of 30 hours continuing education per year, however that is just for the minimum, people who have no ambition, who don't want to provide that little bit more to differentiate themselves from others in their field will only do the minimum to get by and keep their jobs. I hope that is not you. I would not think to get into a debate regarding your specialization nor imply that you are not doing your job. And while this issue is one that most people seem to misunderstand, for someone that is in the industry and does make a living at this, you seem to be way out of touch, or am I misunderstanding what you are saying?
>A+ is a recognised qualification, and a very good base for further technical development, and the information is getting better. However, industry certs are just that – a base to build upon with real-world experience, which can often prove the theories drastically flawed. I think my information is reasonably up-to-date (ref the two links I have included in this critique).
I wish you well in your endeavers.
>And the same to you
CitrixEngineer@yahoo.co.uk
I would just like to say I have found your debate/debates most entertaining and very informative, in a confusing sort of way!!
You are both obviously very technically savi and I applaud the brinkmanship and posture jousting but!!!!!!!
LETS KISS AND MAKE UP NOW LADS!
Before it gets BLOODY!
Because I am not holding anyones coat if it goes to punches. lololol
PS what did all that stuff mean????????????? lolol Just trying to help, sometimes falling short, I am only human after all.
You are both obviously very technically savi and I applaud the brinkmanship and posture jousting but!!!!!!!
LETS KISS AND MAKE UP NOW LADS!
Before it gets BLOODY!
Because I am not holding anyones coat if it goes to punches. lololol
PS what did all that stuff mean????????????? lolol Just trying to help, sometimes falling short, I am only human after all.
CitrixEngineer
Technical User
Hopefully this is a technical debate, with no personal animosities intended or taken. In such a debate, the only side that should be taken is that of the truth and hard facts. It should not be mistaken for a battle of wills or the desire to simply prove oneself right. That sort of thing happens on 3dFiles.com, not Tek-Tips. ;-)
It is clear that fasttoon disagrees with the core point of this thread, and I am interested in resolving it so that all who are following this thread may learn - including myself. Maybe my sources are wrong. Whatever, the logic that has been presented so far seems to me to be fundamentally flawed, but I'd like to see good reasoning behind this theory before I dismiss it or replace my own concepts.
As long as this keeps On Topic, I see no reason why the debate should not continue. If more links are provided to sources, then that is more knowledge that is addded to the melting pot.
I think the points should be made a little shorter and to the point, however...
Guilty :~/
CitrixEngineer@yahoo.co.uk
It is clear that fasttoon disagrees with the core point of this thread, and I am interested in resolving it so that all who are following this thread may learn - including myself. Maybe my sources are wrong. Whatever, the logic that has been presented so far seems to me to be fundamentally flawed, but I'd like to see good reasoning behind this theory before I dismiss it or replace my own concepts.
As long as this keeps On Topic, I see no reason why the debate should not continue. If more links are provided to sources, then that is more knowledge that is addded to the melting pot.
I think the points should be made a little shorter and to the point, however...
Guilty :~/
CitrixEngineer@yahoo.co.uk
Hey CitrixEngineer,
I guess I didn't have the email notifications turned on, so it will be a while for me to read your posts and understand your points, agree, disagree, and reply.
In any case just by serendipity a newsletter I subscribe to came up with a remark that touches upon our debate.
---
"Striped sets certainly aren't anything new. They've been in Microsoft's higher-end operating systems since the early NT days (NT 3.1 - maybe even LAN Manager). There is NO way that it will triple the performance of a system, so please don't believe that, no matter who says they saw that sort of improvement. It's simply not possible for several reasons - most notably because software stripes can't hold a candle to hardware implementations, and for the most part, such a thing will be employed using IDE drives...
>>As such, it cannot perform two I/O requests simultaneously on the same controller, so if data needs to be read from or written to both disks on a channel, they have to line up and wait their turn.<<
(my comments - exactly my line of reasoning)
...It's really no different than using a single IDE drive in that case...
>>>If multiple IDE controllers are used, then there may be some measure of performance improvement,<<<
(my comment - Again my point of view, keep in mind that he is refering to software solution, however with hardware there will be little or no noticiable loss of performance)
...but again, there is overhead in the software management of the I/O above and beyond what the hardware has to do. IDE, by its very nature, uses the system CPU to manage I/O requests. Brien is a good guy, but stating "three times faster" is simply not correct..."
Above preceding excerpt is from: 03.19.2002 GnomeREPORT
Lockernome - Chris Pirello
---
As you can see there are a lot of other more knowledgeable persons out there who have done more research then I (I think this is the guy that organizes Gnomedex), of course as a hobby its more of experience and learning at the same time. So my original conclusion and recommendation still stand, however that may be modified or changed depending on reviewing what you have wrote as well as other sources of information that I may come across, expect a reply in a few days.
I guess I didn't have the email notifications turned on, so it will be a while for me to read your posts and understand your points, agree, disagree, and reply.
In any case just by serendipity a newsletter I subscribe to came up with a remark that touches upon our debate.
---
"Striped sets certainly aren't anything new. They've been in Microsoft's higher-end operating systems since the early NT days (NT 3.1 - maybe even LAN Manager). There is NO way that it will triple the performance of a system, so please don't believe that, no matter who says they saw that sort of improvement. It's simply not possible for several reasons - most notably because software stripes can't hold a candle to hardware implementations, and for the most part, such a thing will be employed using IDE drives...
>>As such, it cannot perform two I/O requests simultaneously on the same controller, so if data needs to be read from or written to both disks on a channel, they have to line up and wait their turn.<<
(my comments - exactly my line of reasoning)
...It's really no different than using a single IDE drive in that case...
>>>If multiple IDE controllers are used, then there may be some measure of performance improvement,<<<
(my comment - Again my point of view, keep in mind that he is refering to software solution, however with hardware there will be little or no noticiable loss of performance)
...but again, there is overhead in the software management of the I/O above and beyond what the hardware has to do. IDE, by its very nature, uses the system CPU to manage I/O requests. Brien is a good guy, but stating "three times faster" is simply not correct..."
Above preceding excerpt is from: 03.19.2002 GnomeREPORT
Lockernome - Chris Pirello
---
As you can see there are a lot of other more knowledgeable persons out there who have done more research then I (I think this is the guy that organizes Gnomedex), of course as a hobby its more of experience and learning at the same time. So my original conclusion and recommendation still stand, however that may be modified or changed depending on reviewing what you have wrote as well as other sources of information that I may come across, expect a reply in a few days.
CitrixEngineer
Technical User
There is much in what you are saying, but if you refer to the link to PCGuide, you will see that Packet devices don't work the same way as Block mode devices, are incapable of transferring data at anything like the same speed, have slower access times, and, as yet, do not subscribe to ATA100 or 133. In fact most are still UDMA33 devices.
It's logical, even preferable to have HDDs on separate IDE channels, and it _may_ improve performance to have the packet drives on separate channels, although given their access times, I doubt it.
The biggest drawback to this configuration is that you are putting different PIO mode devices onto the same channel, so that the entire channel will operate at the same speed as the slowest device.
That is what the argument hinges on, and is what most PC technicians will tell you is the case. Whenever someone installs a new, fast HDD into a machine on the same IDE cable as a ROM, they complain of slowness. The fix is invariably to move the ROM.
Again, referring to the PCGuide article(s), the solution recommended for a 2HDD/2ATAPI setup is HDD0 as Master on IDE1, HDD1 as Master on IDE2, and the two ATAPI devices on the same channel on a separate IDE controller (eg a soundcard with IDE support).
It would be an interesting excercise to benchmark the different setups and prove the theories.
Thank you for your interesting and thoughtful contributions. CitrixEngineer@yahoo.co.uk
It's logical, even preferable to have HDDs on separate IDE channels, and it _may_ improve performance to have the packet drives on separate channels, although given their access times, I doubt it.
The biggest drawback to this configuration is that you are putting different PIO mode devices onto the same channel, so that the entire channel will operate at the same speed as the slowest device.
That is what the argument hinges on, and is what most PC technicians will tell you is the case. Whenever someone installs a new, fast HDD into a machine on the same IDE cable as a ROM, they complain of slowness. The fix is invariably to move the ROM.
Again, referring to the PCGuide article(s), the solution recommended for a 2HDD/2ATAPI setup is HDD0 as Master on IDE1, HDD1 as Master on IDE2, and the two ATAPI devices on the same channel on a separate IDE controller (eg a soundcard with IDE support).
It would be an interesting excercise to benchmark the different setups and prove the theories.
Thank you for your interesting and thoughtful contributions. CitrixEngineer@yahoo.co.uk
hey CitrixEngineer,
Whooooweeeee.
What a read, well if you can't buried them with obfuscation then bury them in misdirection or in your case with both. LOL.
For anyone following this thread you might want to print it out. lol.
-------
ok starting with your 1st post dated 3-19-2002
the url you're pointing to
while correct according to my understanding it is dealing with PIO devices (pre ata) and is correct in that regard.
And while pcguide may be an authority I do not consider an article dealing with PIO as to be authority on current devices and configuration. However if we were considering the best way to configure PIO devices then yes this would be a good article to follow.
--------
Ok your second post dated 3-19-2002
(out of order)
stating that one of my url is not current is correct however on your previous post you use something even more ancient (pio) to prove your point and the quotation "..CD-ROMs on the same channel as a fast hard disk is usually not recommended.." which again refers to the pio standard, please make up your mind as to what constitute current information and your downplaying of that information if you are going to refer back to even older materials.
Again keep in mind that this would not have been my top pick it was just a quick pick that I was able to do in about 10 min just to see if I could find information specific to my position, also I see that you did not refute the second hyperlink as that was the more meaningful piece
I agree with you on experience, it's hard to refute something that you have experienced personally regardless of what others are experiencing.
Your next comment was answered above. (outdated material that covers PIO only) and answered, well not answered but brought up as an explanation of why most people think this is true in my original post of 3-18-2002
"..There are several reasons why the other posters recommend what they said, first is that PIO will run at the slowest device speed, a lot of cd-roms are pio 3 or 4 (old hd are pio or pre ata with same problem), most current cd/dvd/cdrw are ata33, as long as both devices are ata enabled then there is no hit on performance for single devices..." Again exactly true according to the url you posted.
You are correct that with a well defragged disk the performance hit will be less noticable, yet it would still be there.
Your reference to PCI 3.0 is correct however invalid as this is not what this is about (at least not yet). Your agreement on latency and timing is funny as that is what I'm basing my entire supposition on and does refute what you are trying to prove. That there is a lag on devices using the same channel. (and that is what this original issue is all about)
your url is good however in repeating my comment "What a read, well if you can buried them with obfuscation then bury them in misdirection or in your case with both. LOL." Why do I say that? because you must have a password and in order to get a password you must be an industry insider which I'm not and of which I assumed you were by your user name, and also why I was don't think you kept up with the technology when you stated your position. In any case in those sections in which I was able to access, specifically:
PCI local bus
PCI Compliance Checklist
revision 2.2 9908.15
Addendum A: Master Protocol test Scenarios for Components.
this part seems to indicate that each I/O is separate and distinct and not concurrent. (this seems to support my point rather than yours).
Please email a password and I will review the additional information that is not available to the public and comment on that as well.
Your next comment refers back to article which you agree with me yet this is the outdated information only dealing with PIO of which we both agree is correct but is not applicable to ATA devices.
In your next comment you agree with me (huh? I thought you had the opposing view point) and state where did that quote come from, again I'll put the url in just like in my original post of 3-18-2002
The preceding quote was taken from: SCSI vs. IDE Bus Mastering for DAWs, Part 4 by D. Glen Cardenas and Jose M. Catena.
Keep in mind that if you want to dispute/refute this quote or more precisely the reasoning behind this quote you must go back to the authors of the piece, Again these pieces were found with a search engine in less than 15 min to bolster my position based on your request for authoritative views of which, I again mention, I don't need nor care for your approval which I feel are not up to date. (is this a flame, yes in a way, I will comment fully at the end.) Would I have chosen these articles as my first pick, I don't know I was just looking on the web for quick information. However they do point out that my experience is not out of the blue and does documented that other people have come to the same conclusions.
your next comment again refers to the outdated article from pcguide.com and again ask where did my quote come from, see preceding.
your next comment is on the chipset, while the article uses Intel as the chipset, this applies to all current chipsets for the micro so your insinuation that chipsets will affect the configuration of the ata devices is designed to lead astray the non-technically astute. In the same statement you are also alluding that the disk controllers and ata cards have something to do with the configuration of the devices. This is also incorrect as the IDE and PCI bus will still control the information flow and that is the point of my whole position that this is the bottleneck. It doesn't matter how much more efficient a particular controller, driver, ata card, etc is you will still encounter this bottleneck as this is inherent in all micro designs. (micro=micro computer or pc for those that are confused)
Your next comment stated you are confused (I make a comment to this at the end as well) you may have missed what I was referring to, here it is:
"The first PIIX imposed this limitation even across channels. The PIIX3 removed the limitation across channels, but not across devices in the same channel. The latest controllers with UDMA support (PIIX4 and PIIX4E) have removed these limits completely, and the mode can be configured independently for each device."
Thats why I commented "(my comments - exactly what I'm saying,... ...enjoy the process and the results.)"
to further clarify, in this article, they are referring to the limitation inherent in the earlier design of putting different interface devices on the same channel, and how this also affected devices on different channels (sounds like they are refering to PIO standards and what that article you keep referring to at is referring to.) The standards have improved (UDMA which is the current standard)to the point that this is no longer the case and does refute what you are recommending for the configuration devices. As they state "...each device can be configured independently for each device..." Again this is my understanding of the current standards and why I state the configuration of my devices are superior at least in my situation. If you have a problem with this quote, please take it up with the authors of this article, to make sure you don't get confused again here it is:
The preceding quote was taken from: SCSI vs. IDE Bus Mastering for DAWs, Part 4 by D. Glen Cardenas and Jose M. Catena.
Your next point is correct on the naming definition, most users both technical and non-technical use IDE, EIDE, ATA, UATA, UDMA interchangably, just look at the ads for hard drive and other periphials, although you are correct and I agree that it is not all the same. Still if that is the only error you can point out then your refutation of my comments/views/position is less than tenuous.
Your next to last comment is something I can agree with as all the certification process is just a competence exam and like all exams is designed to give you the basic theories which can be applied to real world situation and which may not work properly, hence the phrase "it was a text book situation". Again the 2 urls you are refering to (really its 3)are not applicable as the pcguide.com is out of date dealing with PIO devices, the PCI SIG site is for industry insiders which require a password (will you be kind enough to give it to me?) and of which the guides and articles that are available to the public does seem to bolster my position, the other url you have listed in your post networkcomputing.com (see there are 3 not 2), while the url pointed to the new PCI 3.0 standard which is not even out yet, I could say that if you are refering to future standards why not go with SATA which from what I understand will make configuration of devices moot, in any case the site seems to be a good site to browse. I thank you for that as I do like to see what is going on out there, who knows it may even bolster your point someplace in that website (lol).
--------
ok your third post dated 3-19-2002
you state I disagree with the core point of this thread, the core point of this thread is "Multiple HDD & CD ROM connection" by smuuge who goes on to ask "I have two hard drives and two cd drives and would like to know the best way to connect them to my mobo". of which I have not agreed or disagreed, I have given my advice and opinion based on experience.
the rest of your post of making it shorter and to the point I agree with and will comment at the end of my post.
----------
ok your post dated 3-21-2002
your reference to pcguide.com is refering to PIO devices and not ata devices, so is not applicable as mentioned above.
your next comment even states that my conclusion is logical
"It's logical, even preferable to have HDDs on separate IDE channels, and it _may_ improve performance to have the packet drives on separate channels, although given their access times, I doubt it"
your next reference proves my point.
"The biggest drawback to this configuration is that you are putting different PIO mode devices onto the same channel, so that the entire channel will operate at the same speed as the slowest device."
Again we are not debating PIO devices, were are speaking of ATA devices, (wait!! its IDE, or maybe EIDE, or maybe its UATA, nope wait we were talking about UDMA devices, lol)
your next comment of complaints when mixing hd and rom is correct when one device is a PIO device or when utilizing both drives at the same time, which is what I was stating. (of course your position is this is not true for similar devices. so we partially agree here)
Your next comment "Again, referring to the PCGuide article(s), the solution recommended for a 2HDD/2ATAPI setup is HDD0 as Master on IDE1, HDD1 as Master on IDE2, and the two ATAPI devices on the same channel on a separate IDE controller (eg a soundcard with IDE support)." I did not see this in the article you hyperlinked, please post the hyperlink as I believe you are referring to a different article, however this is exactly what I advocate as far as the hard drives are concerned. Thank you for finally agreeing with me!! (does this mean you are refuting your own argument? lol)
this is my original post dated 3-18-2002 and what started this debate.
"I run HD primary with CD slave with no problems, however that is assuming that your cd is ata 33. I have 2 hd, cdrw, dvd-rom, they are crossed matched as well."
and
"which is why you were set up correctly with both hard drives as primary "
Exactly what I have been saying hard drives on separate channels, you also agree by quoting that pcguide.com article. (again please post the correct url for reference so I can review and gloat. LOL)
Now there is a difference here on the connections of the "roms" as you call it. I get around that potential problem by using ata/raid cards.
second post 3-18-2002
"I prefer an ATA or RAID card to get all devices on their own channel to avoid this very problem."
however not knowing if the original poster "smuuge" has access or is willing to get an add in card, the next best solution is to cross configure the devices, ex: Primary IDE Channel HD0 primary, rom secondary, Secondary IDE Channel HD1 primary, rom secondary, again depending on what you use it for decides which "rom" goes on which channel.
This seems like the solution you have not advocated yet, although you are now advocating having multiple hard drives on separate channels based on the pcguide.com article. So it seems based on outside authority rather than personal experience you are slowly coming around to my positions, we are halfway there.
Your last comment
"It would be an interesting excercise to benchmark the different setups and prove the theories."
I have as mentioned in second post on 3-18-2002
"In my experience using CD/DVD player and CDRW player on the same channel regardless of which order its in will result in coasters or with burn proof technology will take approx 3-6 times as long to complete. I also regulary transfer large files between hard drives, if both hard drives are on the same channel it has taken up to 5-10 times longer than if both drives are on separate channels, this experience points to the pci bus as the bottleneck."
There was no reason to benchmark this for my benefit as the performance was very noticable.
It is really best for people to use various configurations and come to their own conclusion. If they want to benchmark the various configurations and post the results I'm all for that.
-----------
Conclusion of this post,
Damn!! this is too long, and based on some of the comments of CitrixEngineer he infers that it got to be really confusing, I had to open 3 browsers and print this out to follow what was going on. I can imagine the confusion of others trying to follow along. Everyone that wants to follow along debate, please to print it out and ask questions if you are still confused.
A good debater will obfuscate, misdirect, mislead, point out errors large and small in the details that have no bearing on the point in question. While the larger issue is lost along the way. A good debater will bring the issue back into light as I have done in this latest post and which CitrixEngineer has alluded to in his last post of 3-19-2002.
To this I salute you CitrixEngineer, a good debater
I do take issue with the "Authoritative" tone in CitrixEngineer's post, requesting authoritative sources for my opinion and advice, however that may be just my misunderstanding of his style as he is in the industry and engineers of all types will request fact's and just the fact's when in reality real life sometimes does not play that way and he does agree partitially with his comment on A+ certs.
While I have no desire to debate someone in their own field while having no knowledge in that field (I take it that you are an engineer in networking/lan/mainframe or something like that.) I do understand that sometimes you can be so focused on your specialty that you give just lip service to what the other fields that are on the periphery, and that it can be difficult to keep up. Which is why I posted the hours of continuing education required for my field and why I go out of my way to exceed the minimum to understand what the outlying specialties are and how they can possibly impact my job.
I have no respect in someone that espouse a point without the background to fully understand the particular issue (maybe he does but it doesn't show) and instead of proving their point goes to lengths to break down opposing views, had he just stated he disagreed instead of asking for "authoritative sites" I would have demurred and given my experience and reasonings why, rather than at times sounding like a flame. (of which this paragraph is moving to)
To tell you the truth, I was upset at his first reply to my post and based on his user name I assumed that he was in the "biz" and should know better. I certainly felt bad after a harsh posting without thinking, maybe he wasn't in the "biz" and I just thinking of apologizing for my style of reply. If it means anything you have my apology on my style of reply, currently I'm no longer upset nor apologetic to CitrixEngineer. It certainly has been fun and educational in a way, take my other riposte as sarcasm for that is what its meant to be. I have taken some of your comments in the same vein, hope you meant it as such as well.
I do feel good because you are coming around to my point of view, maybe a few more posts will take you over the edge and firmly into my camp. That in itself is not the issue nor is what this debate should be about. It should be about the quest for knowledge and how you or anyone can use that for their own benefit.
Since this post has started to touch on subjects that have little or no bearing on the original questions it is becoming wearing and if this was your intent or debating style then I surrender and you win.
For everyone out there I suggest taking time to try different configuration and come to your own conclusion. I have come to my conclusion based on experience over the last 4 years, so while not a long experience it has been very compressed (if I had to estimate what I spent on just this issue, configurations of drives, it would be something like 7 days straight or to put another way approx 168 hours, ask me about fat32 vs ntfs, thats another issue I've spent too much time on as well, lol, What!! don't I have more important things in my life? I need to get a life, lol)
I apologize for such a long post on such an esoteric subject (or at least thats what I think it was starting to become, the competitive spirit in me wanted to reply point for point) and wasting all this space, for those that do learn something, I congratulate you on being able to follow this tread and learning something whatever it was, lol even if it was learning that I waste too much time on something like this and need to get a life, lol.
I don't think I'll reply to any other posts in this tread, of course if its interesting? hmmm, maybe, lol.
Wish you all well if you got to the end of this post.
Whooooweeeee.
What a read, well if you can't buried them with obfuscation then bury them in misdirection or in your case with both. LOL.
For anyone following this thread you might want to print it out. lol.
-------
ok starting with your 1st post dated 3-19-2002
the url you're pointing to
while correct according to my understanding it is dealing with PIO devices (pre ata) and is correct in that regard.
And while pcguide may be an authority I do not consider an article dealing with PIO as to be authority on current devices and configuration. However if we were considering the best way to configure PIO devices then yes this would be a good article to follow.
--------
Ok your second post dated 3-19-2002
(out of order)
stating that one of my url is not current is correct however on your previous post you use something even more ancient (pio) to prove your point and the quotation "..CD-ROMs on the same channel as a fast hard disk is usually not recommended.." which again refers to the pio standard, please make up your mind as to what constitute current information and your downplaying of that information if you are going to refer back to even older materials.
Again keep in mind that this would not have been my top pick it was just a quick pick that I was able to do in about 10 min just to see if I could find information specific to my position, also I see that you did not refute the second hyperlink as that was the more meaningful piece
I agree with you on experience, it's hard to refute something that you have experienced personally regardless of what others are experiencing.
Your next comment was answered above. (outdated material that covers PIO only) and answered, well not answered but brought up as an explanation of why most people think this is true in my original post of 3-18-2002
"..There are several reasons why the other posters recommend what they said, first is that PIO will run at the slowest device speed, a lot of cd-roms are pio 3 or 4 (old hd are pio or pre ata with same problem), most current cd/dvd/cdrw are ata33, as long as both devices are ata enabled then there is no hit on performance for single devices..." Again exactly true according to the url you posted.
You are correct that with a well defragged disk the performance hit will be less noticable, yet it would still be there.
Your reference to PCI 3.0 is correct however invalid as this is not what this is about (at least not yet). Your agreement on latency and timing is funny as that is what I'm basing my entire supposition on and does refute what you are trying to prove. That there is a lag on devices using the same channel. (and that is what this original issue is all about)
your url is good however in repeating my comment "What a read, well if you can buried them with obfuscation then bury them in misdirection or in your case with both. LOL." Why do I say that? because you must have a password and in order to get a password you must be an industry insider which I'm not and of which I assumed you were by your user name, and also why I was don't think you kept up with the technology when you stated your position. In any case in those sections in which I was able to access, specifically:
PCI local bus
PCI Compliance Checklist
revision 2.2 9908.15
Addendum A: Master Protocol test Scenarios for Components.
this part seems to indicate that each I/O is separate and distinct and not concurrent. (this seems to support my point rather than yours).
Please email a password and I will review the additional information that is not available to the public and comment on that as well.
Your next comment refers back to article which you agree with me yet this is the outdated information only dealing with PIO of which we both agree is correct but is not applicable to ATA devices.
In your next comment you agree with me (huh? I thought you had the opposing view point) and state where did that quote come from, again I'll put the url in just like in my original post of 3-18-2002
The preceding quote was taken from: SCSI vs. IDE Bus Mastering for DAWs, Part 4 by D. Glen Cardenas and Jose M. Catena.
Keep in mind that if you want to dispute/refute this quote or more precisely the reasoning behind this quote you must go back to the authors of the piece, Again these pieces were found with a search engine in less than 15 min to bolster my position based on your request for authoritative views of which, I again mention, I don't need nor care for your approval which I feel are not up to date. (is this a flame, yes in a way, I will comment fully at the end.) Would I have chosen these articles as my first pick, I don't know I was just looking on the web for quick information. However they do point out that my experience is not out of the blue and does documented that other people have come to the same conclusions.
your next comment again refers to the outdated article from pcguide.com and again ask where did my quote come from, see preceding.
your next comment is on the chipset, while the article uses Intel as the chipset, this applies to all current chipsets for the micro so your insinuation that chipsets will affect the configuration of the ata devices is designed to lead astray the non-technically astute. In the same statement you are also alluding that the disk controllers and ata cards have something to do with the configuration of the devices. This is also incorrect as the IDE and PCI bus will still control the information flow and that is the point of my whole position that this is the bottleneck. It doesn't matter how much more efficient a particular controller, driver, ata card, etc is you will still encounter this bottleneck as this is inherent in all micro designs. (micro=micro computer or pc for those that are confused)
Your next comment stated you are confused (I make a comment to this at the end as well) you may have missed what I was referring to, here it is:
"The first PIIX imposed this limitation even across channels. The PIIX3 removed the limitation across channels, but not across devices in the same channel. The latest controllers with UDMA support (PIIX4 and PIIX4E) have removed these limits completely, and the mode can be configured independently for each device."
Thats why I commented "(my comments - exactly what I'm saying,... ...enjoy the process and the results.)"
to further clarify, in this article, they are referring to the limitation inherent in the earlier design of putting different interface devices on the same channel, and how this also affected devices on different channels (sounds like they are refering to PIO standards and what that article you keep referring to at is referring to.) The standards have improved (UDMA which is the current standard)to the point that this is no longer the case and does refute what you are recommending for the configuration devices. As they state "...each device can be configured independently for each device..." Again this is my understanding of the current standards and why I state the configuration of my devices are superior at least in my situation. If you have a problem with this quote, please take it up with the authors of this article, to make sure you don't get confused again here it is:
The preceding quote was taken from: SCSI vs. IDE Bus Mastering for DAWs, Part 4 by D. Glen Cardenas and Jose M. Catena.
Your next point is correct on the naming definition, most users both technical and non-technical use IDE, EIDE, ATA, UATA, UDMA interchangably, just look at the ads for hard drive and other periphials, although you are correct and I agree that it is not all the same. Still if that is the only error you can point out then your refutation of my comments/views/position is less than tenuous.
Your next to last comment is something I can agree with as all the certification process is just a competence exam and like all exams is designed to give you the basic theories which can be applied to real world situation and which may not work properly, hence the phrase "it was a text book situation". Again the 2 urls you are refering to (really its 3)are not applicable as the pcguide.com is out of date dealing with PIO devices, the PCI SIG site is for industry insiders which require a password (will you be kind enough to give it to me?) and of which the guides and articles that are available to the public does seem to bolster my position, the other url you have listed in your post networkcomputing.com (see there are 3 not 2), while the url pointed to the new PCI 3.0 standard which is not even out yet, I could say that if you are refering to future standards why not go with SATA which from what I understand will make configuration of devices moot, in any case the site seems to be a good site to browse. I thank you for that as I do like to see what is going on out there, who knows it may even bolster your point someplace in that website (lol).
--------
ok your third post dated 3-19-2002
you state I disagree with the core point of this thread, the core point of this thread is "Multiple HDD & CD ROM connection" by smuuge who goes on to ask "I have two hard drives and two cd drives and would like to know the best way to connect them to my mobo". of which I have not agreed or disagreed, I have given my advice and opinion based on experience.
the rest of your post of making it shorter and to the point I agree with and will comment at the end of my post.
----------
ok your post dated 3-21-2002
your reference to pcguide.com is refering to PIO devices and not ata devices, so is not applicable as mentioned above.
your next comment even states that my conclusion is logical
"It's logical, even preferable to have HDDs on separate IDE channels, and it _may_ improve performance to have the packet drives on separate channels, although given their access times, I doubt it"
your next reference proves my point.
"The biggest drawback to this configuration is that you are putting different PIO mode devices onto the same channel, so that the entire channel will operate at the same speed as the slowest device."
Again we are not debating PIO devices, were are speaking of ATA devices, (wait!! its IDE, or maybe EIDE, or maybe its UATA, nope wait we were talking about UDMA devices, lol)
your next comment of complaints when mixing hd and rom is correct when one device is a PIO device or when utilizing both drives at the same time, which is what I was stating. (of course your position is this is not true for similar devices. so we partially agree here)
Your next comment "Again, referring to the PCGuide article(s), the solution recommended for a 2HDD/2ATAPI setup is HDD0 as Master on IDE1, HDD1 as Master on IDE2, and the two ATAPI devices on the same channel on a separate IDE controller (eg a soundcard with IDE support)." I did not see this in the article you hyperlinked, please post the hyperlink as I believe you are referring to a different article, however this is exactly what I advocate as far as the hard drives are concerned. Thank you for finally agreeing with me!!
(does this mean you are refuting your own argument? lol)this is my original post dated 3-18-2002 and what started this debate.
"I run HD primary with CD slave with no problems, however that is assuming that your cd is ata 33. I have 2 hd, cdrw, dvd-rom, they are crossed matched as well."
and
"which is why you were set up correctly with both hard drives as primary "
Exactly what I have been saying hard drives on separate channels, you also agree by quoting that pcguide.com article. (again please post the correct url for reference so I can review and gloat.
LOL)Now there is a difference here on the connections of the "roms" as you call it. I get around that potential problem by using ata/raid cards.
second post 3-18-2002
"I prefer an ATA or RAID card to get all devices on their own channel to avoid this very problem."
however not knowing if the original poster "smuuge" has access or is willing to get an add in card, the next best solution is to cross configure the devices, ex: Primary IDE Channel HD0 primary, rom secondary, Secondary IDE Channel HD1 primary, rom secondary, again depending on what you use it for decides which "rom" goes on which channel.
This seems like the solution you have not advocated yet, although you are now advocating having multiple hard drives on separate channels based on the pcguide.com article. So it seems based on outside authority rather than personal experience you are slowly coming around to my positions, we are halfway there.
Your last comment
"It would be an interesting excercise to benchmark the different setups and prove the theories."
I have as mentioned in second post on 3-18-2002
"In my experience using CD/DVD player and CDRW player on the same channel regardless of which order its in will result in coasters or with burn proof technology will take approx 3-6 times as long to complete. I also regulary transfer large files between hard drives, if both hard drives are on the same channel it has taken up to 5-10 times longer than if both drives are on separate channels, this experience points to the pci bus as the bottleneck."
There was no reason to benchmark this for my benefit as the performance was very noticable.
It is really best for people to use various configurations and come to their own conclusion. If they want to benchmark the various configurations and post the results I'm all for that.
-----------
Conclusion of this post,
Damn!! this is too long, and based on some of the comments of CitrixEngineer he infers that it got to be really confusing, I had to open 3 browsers and print this out to follow what was going on. I can imagine the confusion of others trying to follow along. Everyone that wants to follow along debate, please to print it out and ask questions if you are still confused.
A good debater will obfuscate, misdirect, mislead, point out errors large and small in the details that have no bearing on the point in question. While the larger issue is lost along the way. A good debater will bring the issue back into light as I have done in this latest post and which CitrixEngineer has alluded to in his last post of 3-19-2002.
To this I salute you CitrixEngineer, a good debater
I do take issue with the "Authoritative" tone in CitrixEngineer's post, requesting authoritative sources for my opinion and advice, however that may be just my misunderstanding of his style as he is in the industry and engineers of all types will request fact's and just the fact's when in reality real life sometimes does not play that way and he does agree partitially with his comment on A+ certs.
While I have no desire to debate someone in their own field while having no knowledge in that field (I take it that you are an engineer in networking/lan/mainframe or something like that.) I do understand that sometimes you can be so focused on your specialty that you give just lip service to what the other fields that are on the periphery, and that it can be difficult to keep up. Which is why I posted the hours of continuing education required for my field and why I go out of my way to exceed the minimum to understand what the outlying specialties are and how they can possibly impact my job.
I have no respect in someone that espouse a point without the background to fully understand the particular issue (maybe he does but it doesn't show) and instead of proving their point goes to lengths to break down opposing views, had he just stated he disagreed instead of asking for "authoritative sites" I would have demurred and given my experience and reasonings why, rather than at times sounding like a flame. (of which this paragraph is moving to)
To tell you the truth, I was upset at his first reply to my post and based on his user name I assumed that he was in the "biz" and should know better. I certainly felt bad after a harsh posting without thinking, maybe he wasn't in the "biz" and I just thinking of apologizing for my style of reply. If it means anything you have my apology on my style of reply, currently I'm no longer upset nor apologetic to CitrixEngineer. It certainly has been fun and educational in a way, take my other riposte as sarcasm for that is what its meant to be. I have taken some of your comments in the same vein, hope you meant it as such as well.
I do feel good because you are coming around to my point of view, maybe a few more posts will take you over the edge and firmly into my camp. That in itself is not the issue nor is what this debate should be about. It should be about the quest for knowledge and how you or anyone can use that for their own benefit.
Since this post has started to touch on subjects that have little or no bearing on the original questions it is becoming wearing and if this was your intent or debating style then I surrender and you win.
For everyone out there I suggest taking time to try different configuration and come to your own conclusion. I have come to my conclusion based on experience over the last 4 years, so while not a long experience it has been very compressed (if I had to estimate what I spent on just this issue, configurations of drives, it would be something like 7 days straight or to put another way approx 168 hours, ask me about fat32 vs ntfs, thats another issue I've spent too much time on as well, lol, What!! don't I have more important things in my life? I need to get a life, lol)
I apologize for such a long post on such an esoteric subject (or at least thats what I think it was starting to become, the competitive spirit in me wanted to reply point for point) and wasting all this space, for those that do learn something, I congratulate you on being able to follow this tread and learning something whatever it was, lol even if it was learning that I waste too much time on something like this and need to get a life, lol.
I don't think I'll reply to any other posts in this tread, of course if its interesting? hmmm, maybe, lol.
Wish you all well if you got to the end of this post.
Hey CitrixEngineer,
I want to clarify a point. I was thinking about what I posted, reviewing it in my mind and this thought came out of the blue.
"Your next comment "Again, referring to the PCGuide article(s), the solution recommended for a 2HDD/2ATAPI setup is HDD0 as Master on IDE1, HDD1 as Master on IDE2, and the two ATAPI devices on the same channel on a separate IDE controller (eg a soundcard with IDE support)." I did not see this in the article you hyperlinked, please post the hyperlink as I believe you are referring to a different article, however this is exactly what I advocate as far as the hard drives are concerned. Thank you for finally agreeing with me!! (does this mean you are refuting your own argument? lol)"
I think I may have misunderstood your position, even if it does sound like you agree with me, I think you were again referring to PIO devices in which case the problem of the devices defaulting to the lowest setting is what you are saying and it just seems like you are agreeing with my point which you have inadvertantly done and agree to my view by default, in other words the reasoning is different but the end result is the same, separate channels for the hard drive.
I also want to say that I have looked up some of your posts, your post on vmm was very eloquent, precise, concise, (unlike this post) and I agree with you totally, you know in my industry ghost writing is very popular although not generally known by the consuming public, is this the same in your industry? If it is it's a good pick. In any case I'm impressed (is this an indirect jibe or what?, lol) and if not I'm still impressed.
Thank you for your time and replies, as I mentioned before I'm ready to give up just because of the time it takes to read/review/research/respond.
Shake and move on? "shake" "shake" lol.
ps. I still would like a password for the pci sig site.
I want to clarify a point. I was thinking about what I posted, reviewing it in my mind and this thought came out of the blue.
"Your next comment "Again, referring to the PCGuide article(s), the solution recommended for a 2HDD/2ATAPI setup is HDD0 as Master on IDE1, HDD1 as Master on IDE2, and the two ATAPI devices on the same channel on a separate IDE controller (eg a soundcard with IDE support)." I did not see this in the article you hyperlinked, please post the hyperlink as I believe you are referring to a different article, however this is exactly what I advocate as far as the hard drives are concerned. Thank you for finally agreeing with me!!
(does this mean you are refuting your own argument? lol)"I think I may have misunderstood your position, even if it does sound like you agree with me, I think you were again referring to PIO devices in which case the problem of the devices defaulting to the lowest setting is what you are saying and it just seems like you are agreeing with my point which you have inadvertantly done and agree to my view by default, in other words the reasoning is different but the end result is the same, separate channels for the hard drive.
I also want to say that I have looked up some of your posts, your post on vmm was very eloquent, precise, concise, (unlike this post) and I agree with you totally, you know in my industry ghost writing is very popular although not generally known by the consuming public, is this the same in your industry? If it is it's a good pick. In any case I'm impressed (is this an indirect jibe or what?, lol) and if not I'm still impressed.
Thank you for your time and replies, as I mentioned before I'm ready to give up just because of the time it takes to read/review/research/respond.
Shake and move on? "shake" "shake" lol.
ps. I still would like a password for the pci sig site.
hey CitrixEngineer,
you know in reviewing my post during my drive home (its 3:09a now) I realize how tired I was in replying to your comments as I'm finding holes in my comments/opinions/replies, if you do point them out (am I testing you?, lol) I will reply more succinctly.
Before I left the office I had to research another matter and used mskb for the answer, as I was driving home I though hey who better to turn to than ol' 'ncle bill, after all most people refer to mskb as the bible of windows and usually of the pc world.
So now that I'm home I did search mskb and lo and behold, this is what I found:
INF: Max Async IO Configuration Parameter (Q112539)
"...First, consider how a data transfer occurs with a single controller and four drives. In the outbound transfer sequence, the device driver transfers a buffer of data to the controller's on-board buffer. This takes place very rapidly via DMA, shared memory, or programmed I/O, typically in a few hundred microseconds at typical bus rates. Then the controller (under varying amounts of device driver assistance) must command the necessary seek operations from the drive that can take up to 50 milliseconds, which is hundreds of times longer than the bus-to-controller transfer. Following this, the actual data is transferred from the controller buffer to the disk drive at the transfer rate determined by the drive type. There may also be rotational latency involved prior to starting the transfer. During this interval, in many systems the device driver and the task which called it must simply wait for the hard drive. Operations cannot be performed on the second and subsequent drives until the first drive finishes, because the controller does not have the necessary logic to keep track of multiple pending operations...."
(my comment - While this is article is specific to NT and Microsoft SQL Server version 4.2x, my understanding is that this is inherent in all micros. Also - this is really succinct, damn!!, I could have used this and kept my posts to a minimum, lol, thank you 'ncle bill.)
---
and get this, the article further states:
"...In the case of four controllers (each attached to its own drive) if Windows NT or Windows NT Server striping is used, a transfer sequence can immediately begin on the second or subsequent controller/drive. In this case the four drives can independently be in different phases of the transfer since each has its own controller to keep track of this..."
(my comment - can we say raid?, oops ol' 'ncle bill did, or ata cards? lol)
Even though I'm very tired I see at least one thing here that can be put to use against a specifec part of my assumptions/theories/experience although the main parts compare very well with this article and explanation. (another test to see if you have the knack? lol, are you up to the task? apparently I no longer am as I conceded defeat to you in the last post. lol)
In any case this definitely blows your argument/theories totally away with no recourse.
If you give up I will email you the holes, if you promise to email me a password.
Again thanks for your time and opinions.
Boy!! am I glad its a friday and I can sleep in tomorrow, lol.
G'night one and all!!
you know in reviewing my post during my drive home (its 3:09a now) I realize how tired I was in replying to your comments as I'm finding holes in my comments/opinions/replies, if you do point them out (am I testing you?, lol) I will reply more succinctly.
Before I left the office I had to research another matter and used mskb for the answer, as I was driving home I though hey who better to turn to than ol' 'ncle bill, after all most people refer to mskb as the bible of windows and usually of the pc world.
So now that I'm home I did search mskb and lo and behold, this is what I found:
INF: Max Async IO Configuration Parameter (Q112539)
"...First, consider how a data transfer occurs with a single controller and four drives. In the outbound transfer sequence, the device driver transfers a buffer of data to the controller's on-board buffer. This takes place very rapidly via DMA, shared memory, or programmed I/O, typically in a few hundred microseconds at typical bus rates. Then the controller (under varying amounts of device driver assistance) must command the necessary seek operations from the drive that can take up to 50 milliseconds, which is hundreds of times longer than the bus-to-controller transfer. Following this, the actual data is transferred from the controller buffer to the disk drive at the transfer rate determined by the drive type. There may also be rotational latency involved prior to starting the transfer. During this interval, in many systems the device driver and the task which called it must simply wait for the hard drive. Operations cannot be performed on the second and subsequent drives until the first drive finishes, because the controller does not have the necessary logic to keep track of multiple pending operations...."
(my comment - While this is article is specific to NT and Microsoft SQL Server version 4.2x, my understanding is that this is inherent in all micros. Also - this is really succinct, damn!!, I could have used this and kept my posts to a minimum, lol
, thank you 'ncle bill.)---
and get this, the article further states:
"...In the case of four controllers (each attached to its own drive) if Windows NT or Windows NT Server striping is used, a transfer sequence can immediately begin on the second or subsequent controller/drive. In this case the four drives can independently be in different phases of the transfer since each has its own controller to keep track of this..."
(my comment - can we say raid?, oops ol' 'ncle bill did, or ata cards? lol)
Even though I'm very tired I see at least one thing here that can be put to use against a specifec part of my assumptions/theories/experience although the main parts compare very well with this article and explanation. (another test to see if you have the knack? lol, are you up to the task? apparently I no longer am as I conceded defeat to you in the last post. lol)
In any case this definitely blows your argument/theories totally away with no recourse.
If you give up I will email you the holes, if you promise to email me a password.
Again thanks for your time and opinions.
Boy!! am I glad its a friday and I can sleep in tomorrow, lol.
G'night one and all!!
CitrixEngineer
Technical User
I'm not going to try to pick holes in this, since I feel that there is too much wandering off topic going on, but I'll condense all my comments to a few brief points;
1. The core of this thread is "What is the best way to connect 2 HDDs and 2 CD devices".
2. The answer, from all of the links, is clearly to give each device its own channel. But this is not practical, given that most PCs have 2 IDE channels.
3. Fastoon, you have suggested making the HDDs the Masters and the CDs the slaves on each IDE channel, and everyone else has said that splitting HDDs and CDs is the way to go. This is the central argument, and the one I shall stick to.
4. All the evidence suggests that the best performance will be obtained by such a split, because CD devices are slower than HDDs and will force the HDDs to operate at the slower speed. What we are talking about here is a compromise, not the ideal *(see 2.)
5. So what we have here is a difference of opinion over _part_ of what was, after all, a very long post covering a multitude of details. Naturally no-one could refute all of it. This is why I had to agree with parts of what you were saying - because there was so much detail in your post, that much of it was correct. It was your central postulation that Roms and HDDs are happiest sharing channels that I was refuting.
6. Since you seem interested in my background, yes, I am currently involved in networking very large numbers of servers and users. My roots are in PCs, however. I'm talking about the original IBM PCs, with 256k RAM and twin floppy drives. I still remember buying my first PC with a hard disk - 10Mb in size. It was an IBM PC XT. For 10 years I worked as a hardware engineer, using the original IBM manuals - which were very detailed. PC hardware is the base on which I built my experience - I don't claim to know everything, since technology moves too quickly and I am but human, but experience is a fine teacher. I later read for a degree in computer studies to further build on my experience and break into new areas. Education of any kind is a great complement to hands on experience.
7. Thanks for your comments on my Virtual Memory piece - I plan to write a few more articles on the subject, since that one is very general, and tends to dwell on the technical side of the NT architecture a bit too much, according to other comments I've had. I enjoy getting comments on my articles, since they help me to improve future ones. Feel free to e-mail me.
8. Finally, you seem confused over the difference between PIO modes and ATA. Webopedia has succinct definitions; The fact that these are out of date is irrelevant, because they are merely definitions of terms. Again, PCGuide is best for a full technical discussion of this As you will see, ATA is a specification which defines the PIO modes. For more details on ATA, follow this link The link I posted above to the PCI Special Interest Group does require a password for the really technical stuff, but there is more than enough freely available material in the downloadable PDFs to cover the side issue of PCI standards. If you want to get at the rest, you'll have to subscribe like the rest of us ;-) I'm still confused (but happy to stay that way) as to why you think that the PIIx chipset is at all relevant in itself.
As a quick aside, your link to Max Async IO Configuration Parameter and subsequent comments seems to back up my comments...(ducks to avoid flames LOL)
I think that there are enough links and comments in this thread for anyone still interested enough in this topic to make their own minds up. I really don't see that there is an argument here any more;
To split or not to split. THAT is the question.
CE out. CitrixEngineer@yahoo.co.uk
1. The core of this thread is "What is the best way to connect 2 HDDs and 2 CD devices".
2. The answer, from all of the links, is clearly to give each device its own channel. But this is not practical, given that most PCs have 2 IDE channels.
3. Fastoon, you have suggested making the HDDs the Masters and the CDs the slaves on each IDE channel, and everyone else has said that splitting HDDs and CDs is the way to go. This is the central argument, and the one I shall stick to.
4. All the evidence suggests that the best performance will be obtained by such a split, because CD devices are slower than HDDs and will force the HDDs to operate at the slower speed. What we are talking about here is a compromise, not the ideal *(see 2.)
5. So what we have here is a difference of opinion over _part_ of what was, after all, a very long post covering a multitude of details. Naturally no-one could refute all of it. This is why I had to agree with parts of what you were saying - because there was so much detail in your post, that much of it was correct. It was your central postulation that Roms and HDDs are happiest sharing channels that I was refuting.
6. Since you seem interested in my background, yes, I am currently involved in networking very large numbers of servers and users. My roots are in PCs, however. I'm talking about the original IBM PCs, with 256k RAM and twin floppy drives. I still remember buying my first PC with a hard disk - 10Mb in size. It was an IBM PC XT. For 10 years I worked as a hardware engineer, using the original IBM manuals - which were very detailed. PC hardware is the base on which I built my experience - I don't claim to know everything, since technology moves too quickly and I am but human, but experience is a fine teacher. I later read for a degree in computer studies to further build on my experience and break into new areas. Education of any kind is a great complement to hands on experience.
7. Thanks for your comments on my Virtual Memory piece - I plan to write a few more articles on the subject, since that one is very general, and tends to dwell on the technical side of the NT architecture a bit too much, according to other comments I've had. I enjoy getting comments on my articles, since they help me to improve future ones. Feel free to e-mail me.
8. Finally, you seem confused over the difference between PIO modes and ATA. Webopedia has succinct definitions; The fact that these are out of date is irrelevant, because they are merely definitions of terms. Again, PCGuide is best for a full technical discussion of this As you will see, ATA is a specification which defines the PIO modes. For more details on ATA, follow this link The link I posted above to the PCI Special Interest Group does require a password for the really technical stuff, but there is more than enough freely available material in the downloadable PDFs to cover the side issue of PCI standards. If you want to get at the rest, you'll have to subscribe like the rest of us ;-) I'm still confused (but happy to stay that way) as to why you think that the PIIx chipset is at all relevant in itself.
As a quick aside, your link to Max Async IO Configuration Parameter and subsequent comments seems to back up my comments...(ducks to avoid flames LOL)
I think that there are enough links and comments in this thread for anyone still interested enough in this topic to make their own minds up. I really don't see that there is an argument here any more;
To split or not to split. THAT is the question.
CE out.
CitrixEngineer@yahoo.co.ukHey CitrixEngineer,
good to hear from you, yes there may be some confusion on understanding my use of the terms pio and ata because like you I do understand the diference as mentioned in my prior post, however I have defaulted to what the public commonly refers to pio as the older drives and ata meaning the newer drives (again this is what is commonly referred to as ata as in ata100 and ata133), I apologize for the misuse of the terms. So for arguments sake we may have been on different wavelenths.
I do like that piece on VMM, very well written although I like my articles a little bit more breezy, lol, especially since I'm not as technically astute as some of the authors are. You might want to think about reworking some of your articles so the lay person can better grasp your idea, of course if it was for industry insiders then thats a different story.
I remember working on an xt, at least I think it was an xt, no hard drive just dual floppies, and that was the hottest thing around at the time. At the time I thought that was a waste because I could do everything faster by hand (yellow pad and calculator) lol. Those were the days.
I've pretty much completed the pci sig site, almost thru with the pcguide site and will start on the networkingcomputing site shortly.
The pcguide site is pretty informative and does fill in some gaps that I was not aware of, it also includes something that you might find funny, because it completely agrees with my opinion/recommendation.
"...For optimal performance, buy an add-in controller card and use all four channels for the four devices..."
(exactly as I set up all my personal systems)
"...If you have only two channels, there are several options. I would put the fastest drive as the boot device on the primary master, and whatever drive is the least used as the slave on that channel..."
(my original advice to smuuge who started this thread, cross matching the devices. Hard drives on separate channels, cd-rom drives which are less used as slaves.)
"...Then put the other two devices on the secondary channel. Avoid putting drives that can't be run in Ultra DMA with the boot device..."
(as I've stated before - pio vs ata, the common usage of which I was refering to or as you might say PIO mode 4 and under should not be mixed with UDMA mode 2 and up, what happened with UDMA mode 0 and 1 you might ask, well I've never had them so I don't know the results.)
As to what PIIx chipset had to do with it? It really was about nothing as that was something that happened to be what the article used as an example. What they pointed out was that that the data transfers one way at a time and not concurrently, the rest of the article was not of importance to this discussion. In my opinion while chipsets can make a difference in speed in won't overcome the inherent limitations in the data transfers themselves.
As to Max Async IO Configuration Parameter and whether it backs you or me up, It was all about data transfers on a single channel and drives having to wait for the transfers to complete before requesting their turn to transfer data, in other words, one way traffic at a time, and this in turn supports my theory that hard drives configured as master and slave on the same channel will be less efficient then hard drives set up as masters on separate channels, with multiple drives, hard drives are used much more than "rom" drives, and I did mention that this is the way I used them and I also believe this is the way most people use them as well, I also mentioned that cd-rom and cdrw (not a "rom" btw, lol) without burn proof technology on the same channel will result in coasters, or will be much slower than optimal with burnproof technology, which is why I recommend that the cdrw/cd rom should be slaves to the hard drives. BTW the pcguide article does imply this although it doesn't explicitly state it when it recommends
"...I would put the fastest drive as the boot device on the primary master, and whatever drive is the least used as the slave on..."
otherwise why would pcguide recommend matching up the most used drive with the least used drive? And usually the least used is the slowest drive meaning cd/dvd rom. Could they be aware of only one device using the channel at one time? Seems that is what they are implying.
Its been fun (up to a point ( )
cya around.
good to hear from you, yes there may be some confusion on understanding my use of the terms pio and ata because like you I do understand the diference as mentioned in my prior post, however I have defaulted to what the public commonly refers to pio as the older drives and ata meaning the newer drives (again this is what is commonly referred to as ata as in ata100 and ata133), I apologize for the misuse of the terms. So for arguments sake we may have been on different wavelenths.
I do like that piece on VMM, very well written although I like my articles a little bit more breezy, lol, especially since I'm not as technically astute as some of the authors are. You might want to think about reworking some of your articles so the lay person can better grasp your idea, of course if it was for industry insiders then thats a different story.
I remember working on an xt, at least I think it was an xt, no hard drive just dual floppies, and that was the hottest thing around at the time. At the time I thought that was a waste because I could do everything faster by hand (yellow pad and calculator) lol. Those were the days.
I've pretty much completed the pci sig site, almost thru with the pcguide site and will start on the networkingcomputing site shortly.
The pcguide site is pretty informative and does fill in some gaps that I was not aware of, it also includes something that you might find funny, because it completely agrees with my opinion/recommendation.
"...For optimal performance, buy an add-in controller card and use all four channels for the four devices..."
(exactly as I set up all my personal systems)
"...If you have only two channels, there are several options. I would put the fastest drive as the boot device on the primary master, and whatever drive is the least used as the slave on that channel..."
(my original advice to smuuge who started this thread, cross matching the devices. Hard drives on separate channels, cd-rom drives which are less used as slaves.)
"...Then put the other two devices on the secondary channel. Avoid putting drives that can't be run in Ultra DMA with the boot device..."
(as I've stated before - pio vs ata, the common usage of which I was refering to or as you might say PIO mode 4 and under should not be mixed with UDMA mode 2 and up, what happened with UDMA mode 0 and 1 you might ask, well I've never had them so I don't know the results.)
As to what PIIx chipset had to do with it? It really was about nothing as that was something that happened to be what the article used as an example. What they pointed out was that that the data transfers one way at a time and not concurrently, the rest of the article was not of importance to this discussion. In my opinion while chipsets can make a difference in speed in won't overcome the inherent limitations in the data transfers themselves.
As to Max Async IO Configuration Parameter and whether it backs you or me up, It was all about data transfers on a single channel and drives having to wait for the transfers to complete before requesting their turn to transfer data, in other words, one way traffic at a time, and this in turn supports my theory that hard drives configured as master and slave on the same channel will be less efficient then hard drives set up as masters on separate channels, with multiple drives, hard drives are used much more than "rom" drives, and I did mention that this is the way I used them and I also believe this is the way most people use them as well, I also mentioned that cd-rom and cdrw (not a "rom" btw, lol) without burn proof technology on the same channel will result in coasters, or will be much slower than optimal with burnproof technology, which is why I recommend that the cdrw/cd rom should be slaves to the hard drives. BTW the pcguide article does imply this although it doesn't explicitly state it when it recommends
"...I would put the fastest drive as the boot device on the primary master, and whatever drive is the least used as the slave on..."
otherwise why would pcguide recommend matching up the most used drive with the least used drive? And usually the least used is the slowest drive meaning cd/dvd rom. Could they be aware of only one device using the channel at one time? Seems that is what they are implying.
Its been fun
(up to a point ( )cya around.
- Status