CNET: AMD vs. Intel dualcore shootout

[Dr Mordrid]

http://reviews.cnet.com/4520-10442_7-6389077-9.html?tag=btn

I have one dualcore AMD64 X2 and one dualcore Opteron and IMO they're fantastic.

If anything the Opteron can be a better deal. Frequently you can get the lower end dualcore Opterons for the same price as an equally clocked Athlon64 X2. The advantage is a 1 mb cache/core vs. 512k, which can make a difference in repetative tasks like rendering.

Dr. Mordrid

[wkulecz]

I think the tests would have been much more fair and meaningful if an Intel chipset motherboard was used for the Intel CPU. Nvidia Nforce 4 SLI chipset may be the best chipset for AMD, but I'm pretty sure its not for Intel -- The article mentions the memory controller as a likely factor in the performace differences, AMD has theirs on the CPU, Intel puts theirs on the chipset, I doubt Nvidia has the info needed to make a better memory controller than Intel.

In any case, BANG/BUCK its no contest AMD wins whatever chipset is used.

--wally

[Dr Mordrid]

More than bang/buck, but to test both under equal conditions (or as equal as possible) the NVIDIA NFORCE4 was the only option.

Dr. Mordrid

[wkulecz]

Oh well, you can call it "equal conditions" but I see it as handicapping Intel, or a smart move by AMD to include such a performance critical item as the memory controller on chip so third parties can't screw it up, or another example of Intel's greed to help it in the chipset market.

I'd like to see AMD's best vs. Intel's best in video rendering performance and then review the prices to see if its "worth it".

I know AMD X2 rocks, just setup a 3800+ "cheap" system for a Linux video image processing project. Trivial to get two live Xwindows diplaying live 640x480 29.97 fps video from a pair of cheap BT878 "winTV" cards. Now I've got to learn the x86-64 extensions to use in porting our algorthims, should be fun.

--wally.

[Umfriend]

Copy that: equal conditions here should be the best possible mem.contr. for both chips given that you can;t test with the same mem.contr. I'd go as far as to say that aside from the mem.contr. there is little performance impat from one chipset over the other, so there really is no reason to test with the same south-bridge.

[Elie]

Intel handicapped Intel!!!!

The chipset whether it's Intel or Nvidia has a memory bandwidth limitation which hinders the CPU's performance, that plus the issue of having both CPU's share the same path or bus to the memory was always Intel's problem from the get go.
in other word's that 800fsb is getting OLD too fast and is causing Intel a big headache.

AMD played it smart and developed the hyper transport which is a huge independant path from each CPU to the memory, no bottlenecks what so ever.
The downside of this is stability.

This is why workstations today are prefered to be AMD and not Intel where servers are still Intel, but that is quickly changing.

[Umfriend]

Sure, but are you saying that the handicap would be less using the nFart 4 instead of Intels chipset?

[Elie]

No not at all, in fact the nFart 4 (lol) could further impact performance over and above the handicap.

My comment was regarding the overall archtecture of Intels chipset design, so even if they performed the tests based on an Intel motherboard with Intel chipsets, I say the results wouldn't have been that much better.

Cheers,
Elie

[wkulecz]

My comment was regarding the overall archtecture of Intels chipset design, so even if they performed the tests based on an Intel motherboard with Intel chipsets, I say the results wouldn't have been that much better.

Cheers,
Elie

I'd rather see actual test results of best vs. best rather than debate which might be better. Video rendering performance is the only thing I'm really willing to pay a premium price for.

In the absense of real data to guide my choice and evaluate the actual magnitude of the price premium, I'll generally go with the cheaper alternative.

--wally.