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 SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

AMD Sockets

Status
Not open for further replies.

edemiere

MIS
May 13, 2003
483
0
0
US

Hi guys. I am directing this question to cdogg, paparazi and ceh4702.

One of you had discussed sometime in the recent past about AMD changing sockets again from 939 to something new. I am not sure which one of you it was, but I am pretty certain it was one of you.

Anyway, can any of you point me to where you read about this? Because I'd like to read more about it.

Cheers!

 
Check me if I am wrong, but the socket 940 was replaced by the 939.

The original Athlon64 sockets were 754 and 940. Then AMD shifted to the 939 socket.

But I seem to recall one of the guys I mentioned talking about AMD coming to a new socket that would be their new "socket A" solution in the Athlon64 line.

Anyway, thanks for the link and replying!

Cheers!
 
Ok, my last reply was late, lol. I was trying to reply while dealing with my walk-ups. :)

Thanks paparazi, that's what I was looking for and now that I know what it is I am googling for, I will read up some more!

Cheers!
 
The New Socket A, thats a laugh!

Yeah the C2 (which is the name they seem to have settled on, although I have seen other names) has 940 pins but is in no way compatible with the original socket 940 (the one that appeared imediately after socket 754)

The main advantage seems just to be DDR2 (greater band width, at this point)
Martin

We like members to GIVE and not just TAKE.
Participate and help others.
 

You mean the M2? Funny thing is, you say M2 and I think Cyrix :) Remember the M1, M2 and M3 Cyrix processors? Those were GREAT CPU's for the money back in the day (if you weren't doing heavy gaming).

Well, I seemed to recall one of you calling it the new "socket A" or something for it being the new platform that everything will work in or something. I could be wrong though, no problem.

Thanks again!

ED
 
I think the HOPE WAS that this would finally be the socket AMD stick with for some time like socket A was.
But then, I think they said that about socket 939 hmmmmmm.

The underlying opinion was that AMD gained an aweful lot of friends by NOT doing the Intel trick! and changing CPU sockets all the time.
Problem is AMD have become worse in that respect than Intel, both as bad as each other.
Planned obsolessence, thats the terminology.
Making either want or have to have the newest technology.

Martin

We like members to GIVE and not just TAKE.
Participate and help others.
 
The concern was that we'd never see another "socket A"-like platform that would ever last that long again.

The reason, as I've been schooled on now [wink], is that the memory controller is now embedded or integrated on the CPU die itself. In the socket A days, it was on the northbridge of the chipset. So as a result, AMD will have to change the socket architecture (even if just slightly) to accomodate each new memory design that comes out.

With DDR3 on the way soon, I suspect there will be yet another chipset released in the next year or two.

kmcferrin had an excellent point though in another thread. Even with the frequent changes of the socket, it really doesn't matter. Every socket AMD releases sticks around for a couple years. When a replacement is announced (such as the M2 replacing 939), it's no big deal, because you will still be able to buy CPU's for the 939 chipset for another year or two.

If you ride the bandwagon early, you will have a chance to upgrade the CPU once or twice before the chipset is obsolete (if you're the type that likes to keep a system up to date every year or two). Most people, however, will be happy to just replace the motherboard/CPU every 4-5 years to get access to better technology, such as faster memory.

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Albert Einstein
[tab][navy]For general rules and guidelines to get better answers, click here:[/navy] faq219-2884
 
The underlying opinion was that AMD gained an aweful lot of friends by NOT doing the Intel trick! and changing CPU sockets all the time.
Martin,
That's the argument I used to make! But also in that other thread, kmcferrin pointed out that with the socket A architecture, you still had to upgrade the motherboard to go from an original Athlon to an Athlon XP, even though the socket architecture didn't change. The same was true as DDR speeds increased (going from 266 to 400 for example). Some motherboards just didn't have a BIOS flash that would allow this compatibility to extend more than 1 or 2 generations of CPU's.

The other thread is here if you want to take a peek:
thread602-1152863

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Albert Einstein
[tab][navy]For general rules and guidelines to get better answers, click here:[/navy] faq219-2884
 
I think it is about time for Intel to come out with a new Celeron as well. If they start going nuts with these dual core processors, then maybe you will see an 800mhz FSB Celeron. Maybe they will come out with some 1gig FSB dual core processors. The XBOX360 runs off of 3 Processors.

I think it will not be long and we will see the Quad Core Processor, with 2 cores one for each SLI video card or some kind of new Memory Hard Drive or something like that. I think the Price of memory might fall if they quit making DDR and go to DDR2 only.

If you do not like my post feel free to point out your opinion or my errors.
 
cdogg
kmcferrin pointed out that with the socket A architecture, you still had to upgrade the motherboard to go from an original Athlon to an Athlon XP

I agree to a certain extent but not the degree.

Take an early ECS K7S5A motherboard, released in the summer of 2001 and still available now if you search online.
This board supported all socket A processors from the very first Duron 600/650? Athlon Thunderbird 700/750? I forget exactly, all the way up to a 266fsb Athlon XP2.6+ over 21 jumps in CPU speed!
939 I guess was somewhere around 10 (not cunting all the variations in models)
But it's just the length of time socket A was around (over 6years?) again from memory, compared to 2/3 years? for 939

And whilst I don't doubt your explaination about memory controllers, call me old fashioned but 940pins are 940pins are 940pins, they are just contacts or better still, pin outs for the processor, surely the architecture could be altered and still keep the same pin pattern?

Martin

We like members to GIVE and not just TAKE.
Participate and help others.
 
woooops!

Anyway.
My point, if they had to go back to 940pins why couldn't it be 939pin CPU compatible as well as 940?
Well it's not going to happen, I know that but I'm guessing it could have if they had wanted it too.
Martin

We like members to GIVE and not just TAKE.
Participate and help others.
 
Martin,

Your example, the ECS K7S5A, is the only board I know of that has that long of a life span. It was released in the Summer of 2001, just like you said. But, that was after the Athlon Thunderbird "C" was released. So if you were an early adopter of the Athlon or Duron, this board wasn't even close to being available.

In other words, your analogy is like saying a year or two from now, they release a board that can work with both 939 and M2 processors, and support both DDR and DDR2 memory. Sure that would be nice, but would that matter to anyone that already has an Athlon64 or is about to buy one?

The same is the case for a majority of owners that had the original Athlon or Duron models. They had to upgrade their motherboard to get to the Athlon XP (Palomino).

The pin count is probably a moot point when the socket isn't always the reason you have to change out the motherboard...

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Albert Einstein
[tab][navy]For general rules and guidelines to get better answers, click here:[/navy] faq219-2884
 
Cdogg respect to you.
"The ECS K7S5A, is the only board I know of that has that long of a life span. It was released in the Summer of 2001"
I agree, but it was only the very first crop of socket "A" motherboards (those only supporting 100fsb) ie: Via KT133 for example, that didn't support the later 266fsb XP series.
We are talking about production 2000 to 2001 (about 9 months) right at the beginning, everything after that: mid 2001 and later (KT133A/AMD 760) and newer supported XP upto when the 333fsb's CPU's were released (XP2.5+ Barton)

Like I said, you are absolutely right, if you were so unfortunate as to buy a motherboard in those first 9 months of Socket A but that only accounts for an extremely small proportion of Socket 462 users in the following 5 years.

Martin


We like members to GIVE and not just TAKE.
Participate and help others.
 
OK, I was thinking that socket A had been around longer than June of 2000, but I forgot that AMD used the SLOT format much longer than Intel did, and didn't release Socket A until the Athlon about hit 1GHz.

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Albert Einstein
[tab][navy]For general rules and guidelines to get better answers, click here:[/navy] faq219-2884
 
Well, I agree with both of you, so there!

When I bought my Socket A board (Epox 8KHA+, I think in January of 2002), DDR333 and the 333MHz FSB wasn't even available, and DDR266 was brand new. In fact, IIRC, it was shortly after the KT266A chipset was available. But once AMD released the (Palomino?) 333 MHz FSB CPUs, they essentially obsoleted a large number of the Socket A boards that were around at that time, including my Epox 8KHA+ and the ECS K7S5A. And unless you were lucky enough to buy one of the few 333/400 MHz boards when you went to the 333 MHz-based Athlon XPs, you would have had to do it again when they released the 400 MHz.

But I think that Cdogg has probably hit it right on the head when he indicated that most people wouldn't upgrade on the same board. For most people the higher performing "next-generation" CPU can usually be had for just a little more money. From 1998 to the present (so 8 years) I only upgraded my CPU 4 times. I went from a 300 MHz K6-2 (FIC-PA2013) to a 450 MHz K6-III (still on the FIC-PA2013 - the only time that I kept the mainboard) to an Athlon XP 1600+ (Epox 8KHA+) to a Socket 754 Athlon 64 3000+ (Asus K8N). I've been toying with the idea of migrating to an S939/X2 system lately, but I will probably wait for the next generation CPUs (DDR2) because by the time I factor in a new CPU, mainboard, and video card, the cost difference will be minimal.
 
4 times in 8 years, eh?

Since we're on the topic, I'll just say that I've upgraded only 4 times as well, but that goes back to '96 (almost 10 years). In '96 I had a 166MHz Pentium socket 7, then I bought a used K6-2 350MHz in '98. In early 2000, I made the jump to an 800Mhz PIII. I managed to stay afloat for 4.5 years until recently when I built a 3.2GHz Athlon64 system on the 939 chipset (you can imagine what kind of difference that made)!

I imagine that as CPU clock speed becomes less significant over time, I'll find myself upgrading even less often than before.

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Albert Einstein
[tab][navy]For general rules and guidelines to get better answers, click here:[/navy] faq219-2884
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top