Intel vs. AMD (part deux)

Looks like I waited too long to reply to this old thread which has now been closed:
thread602-1262244

It has been almost 2 years now, and even with the new K10 architecture found in AMD's Phenom CPUs, there is still a lot of ground to make up with AMD leaving a lot to be desired. I've also been hearing rumors about a flaw in the early Phenoms that restricts the maximum clock to 2.3GHz. Anyone care to comment on that?

Well I hate to be the one to say it, but after being put on the spot earlier I can only feel justified by saying "See, I told you so!". There, I vented! Now if ya'll can put the past behind you, then so can I. Let's move on, shall we?


Comments/suggestions about AMD's future??

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.

 

[cdogg]

Ben, thanks for the info. You bring up an interesting point about the memory controller. Others can correct me if I'm wrong, but one reason why you wouldn't want to do so now if you're Intel is because we're still in the middle of a transition between DDR2 and DDR3. Having an on-die memory controller ties the CPU to the memory, making it more difficult to support a new type without changing both the CPU and mobo chipset (DDR2 wasn't supported until AM2). AMD has released four different chipsets (754, 939, 940, and AM2) in the same span that Intel released only two (478 and 775), though I'm not sure if the memory controller was the biggest reason or not.

Obviously, the advantages of having an integrated controller outweighs this particular disadvantage, but I suspect since Intel's top products are currently in the lead, they'll prefer to wait until DDR3 becomes more mainstream. It will definitely become a priority before the move to 8-core CPU's. This article hints that it may be closer than you think and even mentions the use of an integrated GPU:

http://arstechnica.com/news.ars/pos...ed-graphics-on-die-memory-controller-smt.html

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.

 

[cdogg]

Actually as a follow-up to that article, I just realized Intel is releasing the Nehalem architecture late this year into early next.

It's still unclear to me, however, if all Nehalem CPU's will have an integrated memory controller. From what I'm reading, only high-end models will and DDR3 is what will be supported.

 

[Sympology]

You have to remember AMD have some massive advancements over Intel in the Higher end regions. The Optereon design allows all sort of plug in mudules, such as physics processors, as well as using GPU's for complex maths calculations, a thing Intel is still working on.
With AMD also now integrating pretty good GPU onto the board, they are attacking from another front, one that has always been Intels domain.
They've had to fight hard to gain any sort of share in the server arena, and now it's going to be the same on lower end desktops and possibly laptops (the 780G is low power).

In my opinion, Intel has always had the speed, but AMD tend to hit back with innovation.

I like the battle, as the real winner is the consumer !



Only the truly stupid believe they know everything.
Stu.. 2004

 

[LawnBoy]

In my opinion, Intel has always had the speed, but AMD tend to hit back with innovation.

I would agree with that. Intel seems to let AMD prove a technique works, then cranks up it's massive manufacturing abilities.




"We must fall back upon the old axiom that when all other contingencies fail, whatever remains, however improbable, must be the truth." - Sherlock Holmes

 

[cdogg]

Sure, that's a common perception that is true in a lot of respects. However, I wouldn't be too quick to discount other areas that make computing what it is today. x86, AGP, PCI, PCI-Express, SSE, MMX, USB, and Bluetooth were all introduced by Intel. The list goes on and on. When it comes to the processor, a lot of what you see from AMD is an improvement to these designs though I'd rather not get into a p*ssing contest about innovation. It's like an assembly line that Intel and AMD share, always pushing for a better product.

I think it would be fair to say we wouldn't be sitting here today talking about AMD if it weren't for Intel. On the same note, Intel's products wouldn't be as remarkable today had it not been for AMD.

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.

 

[LawnBoy]

Neither product would be as good without the competition. Saying that one company wouldn't exist without the other is a bit of a stretch, though.

<flamebait>
I still think Motorola has the best designs; it took the x86 market years to adopt a linear addressing model.
</flamebait>


"We must fall back upon the old axiom that when all other contingencies fail, whatever remains, however improbable, must be the truth." - Sherlock Holmes

 

[kmcferrin]

For posterity's sake, I'd like to point out that 754, 939, 940, and AM2 are sockets, not chipsets. Also, during the same period Intel went through sockets 478, 775, and 771 (which was omitted).

On the AMD side, most of the changes were memory controller related. Socket 754 was single-channel DDR. Socket 939 was dual-channel DDR. Socket AM2 was dual-channel DDR2. Socket 940 was a server version of socket 939, it's primary purpose was to separate server CPUs from desktop CPUs (the same is true of socket 775 and 771 on the Intel side).

One other thing to point out is that not all socket 775 CPUs and mainboards are compatible. Socket 775 defines a 775 pin land grid array interface, but the electrical characteristics have changed over time as lower voltage CPUs have been released. Because of this it is unlikely that the current socket 775 CPUs would function in an early 775 board, and vice versa.



________________________________________
CompTIA A+, Network+, Server+, Security+
MCSE:Security 2003

 

[cdogg]

Chipsets are not identical across the sockets, am I right? The nForce3, for example, was found on both socket 754 and 939 but were different versions. The difference between sockets, slots, slockets, and chipsets is a moot point I'm afraid!


kmcferrin,
I've been sensing a little hesitation in you to give another prediction for AMD! Care to take a swing at it again, old buddy old pal? In case I wasn't clear before, I'm trying my best to move past the technicalities and have just a normal water-cooler conversation for a change!

Just some light-hearted fun of course, nothing personal!!


~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.

 

[Sympology]

Lawn boy...

No flamebait there I'm afraid, I had an Atari Falcon with an 040 (same as the Top Mac's at the time) and this blew Pentiums away. 99 track D2D recording anyone? The pc could manage a measly 4, and that was with some serious hardware....
Oh and it had proper mutlitasking, that could be easily switched by the user (once you knew how). Oh the joys of, assigning max processing power to the app of your choice, those were the days.....


Only the truly stupid believe they know everything.
Stu.. 2004

 

[kmcferrin]

Don't get me started on my Amiga days, Stu.

And yes, the chipset varied from socket to socket. Because of the integrated memory controller you only needed new chipsets when adding new features to the southbridge, so some chipsets were used on 754, 939, and AM2 (or two of the three).

Regarding making predictions, at this point I won't. I've been to the Intel roadmap breifings for the next two years worth of CPUs, but haven't seen what AMD has in store other than what is public knowledge. I will so though that had AMD actually been abel to pull off the native-quad core on their 3Q 07 timeline like originally planned, we'd be back to neck and neck. Falling behind there plus the TLB was a killer. I'm really keen to see how they recover.

Of course, it's hardly the first time tha AMD has been behind in performance. In fact, until the Athlon 64 came about, they were always behind in performance and managed to survive. So I doubt it's doom and gloom and the end of the company.

________________________________________
CompTIA A+, Network+, Server+, Security+
MCSE:Security 2003

 

[LawnBoy]

.


"We must fall back upon the old axiom that when all other contingencies fail, whatever remains, however improbable, must be the truth." - Sherlock Holmes

 

[cdogg]

Thanks for the response.

Yes, somewhere I read that AMD was able to obtain almost a quarter of the worldwide server market on the success of the Opteron. That was a huge gain for them in a short amount of time. I too, doubt that AMD is in any real trouble yet.

I saw this article the other day which discusses AMD's current situation in detail and a general outlook - perhaps the perfect ending to this thread:

http://in.news.yahoo.com/reuters_id...ces-tough-battle-against-intel-796e4e8_1.html

Thanks to everyone for your interest...


~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.

 

[LawnBoy]

Geez, cdogg, that analysis link is just garbage (I mean no slur on you, just the article). Obviously the author has a major agenda to push.

"If they can't get 65 nanometer to work, what says they can get 45 nanometer to work?"

Who said they can't get 65 nm to work? It is working, albeit with a few hitches in rollout which is pretty normal even for Intel.


I think it shortsighted to think that having the "biggest and baddest" chip is the only way to survive. Look at what Google is doing; building huge quantities of fairly low-spec servers to run in their clusters.

It may be a while before AMD has the top chip (if they ever do again) but that doesn't mean the game is over.



"We must fall back upon the old axiom that when all other contingencies fail, whatever remains, however improbable, must be the truth." - Sherlock Holmes

 

[cdogg]

Yes I agree it was poorly worded, trying to make a reference to Barcelona's stutter step. It should have been left out altogether. The article clearly suffers in the technical aspect, and taking the intended audience into consideration it makes sense.

Guess I didn't pay any mind to it the first time around. It was the financial report that caught my eye originally.

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.