AMD Vs Intel on 64-bit perfomance

[Rich85]

With Intel now adopting 64bit technology into their chips (EM64T ? ?Extended Memory 64 Technology?), and AMD with their already engineered 64-bit extensions to the x86 instruction set. Which is going to be the best performer for 64bit technology?

Bearing in mind that MS built the operating system around AMD using the 64-bit extension on the x86 instruction set, which means that MS was not going to build a second 64bit OS for Intel, so Intel had to adopt AMD?s extended 64-bit instruction set to be able to run this OS.

I ask the question because I have been long awaiting the Microsoft?s XP64 OS, who have recently released their RC2 and soon to release their final build. I currently own a 64 bit proc which you can see by my sig as the 3500+ Winchester (which I am very happy with), I do not intend to purchase an Intel chip if that would be the better performer with 64-bit tech, but for information purposes I would be interested in people views on this?

🙂

 

[McArra]

Benchmarks

 

[Vee]

I've been looking to an Intel system for a while. Mainly because I'm old Intel user (since '78-'79), and I don't want to lose touch (I've only done AMD for almost 2 years now). I won't build any 32-bit only, so I've been waiting for 6x0 and 5x1.

Problems are, I look at prices and speak to friends in PC manufacturing about Prescott heat issues, and I get all depressed.

I can get a 3800+ AMD, for the kind of money a 3.2GHz would cost. Without all the possible heat issues. I've been burnt before by Intel throttling, so I consider sufficient cooling to guarantee sustained performance a must.
There's this other thing too. AMD is better working comfort than Intel, (except for some ht multitasking scenarios) for whatever reasons. I suppose it might be just superior performance on GUI/UI type of code? Then again, I haven't really tried the new Prescott yet.

...- Well, I'm just going to get yet another A64 instead. 🙁
It's the only sensible thing. 🙂

To round off, different CPU architectures are always going to have different performance characteristics. From what I've seen, 64-bit seem to be in AMD's favor sofar.

I've heard rumors about that EMT64 has some hidden performance issue at the 4GB border, which suggests the EMT64 bit addressing is built on top of the old P4 memory mapping. But I really don't know anything about it.

 

[Slaimus]

So what compilers ARE available that are 64-bit optimized besides the Intel one?

 

[Duvie]

Isn't it funny how all the real world test show AMD is strong with actually bigger leaps with 64bit yet sissoft as usual shows the opposite of real world??? pretty much why I dont even run that POS app anymore....It is clear that make room for Intels nterpretation of 64bit but not AMDs....

In ll fairness I know the povray is not optimized for Intels 64bit which I believe is slightly different. I am sure that will be fixed shortly....

There are some other benches floating with some other real world apps and it looks solidly like AMD holds the 64bit crown as well in desktops......


Who knows if MS will bow down and optimize more for Intel then AMD and may reverse this shortly....I am a fan of Intel products but I have to admit this type of influence pisses me off!!!

 

[Zebo]

I thought Intel 64 was a carbon copy of AMD? Least tech report says so.

"64-bit extensions ? Intel has dubbed its 64-bit extensions EM64T, for Extended Memory 64 Technology, but they are really just a functional clone of AMD's AMD64 extensions, first implemented in the Opteron processor a couple of years ago. "

Sisuck is not a valid benchmark. Niether are 3dmarks, PCmark or any other synthetic. Incidently I wonder why Intel does so well in those but is M.I.A. on real world? If you like Intel Lazy Sharky Extreme is your website as they love the synthetics.

 

[Duvie]

You may be right Zebo but what I meant was maybe the porgrams are not recognizing on ecpu makers 64bit app. Kinda like some old apps that would not recognize AMD's SSE but would Intels SSE.....I seen it happen and I am sure Intel has a hand in this about half the time....

 

[pm]

About the only problem that I have with the link to Xbit's review is that they don't mention which compiler - and compiler options - were used to compile any of the apps. A couple of the benchmarks used (probably all of them, but I have personal experience with a few of them) are extremely sensitive to the compiler used and the compiler flags used.

 

[jose]

They are not the same it seems..
http://www.theinquirer.net/?article=16879

Regards,
Jose

 

[Duvie]

?Software IOTLB ? Intel® EM64T does not support an IOMMU in hardware while AMD64 processors do. This means that physical addresses above 4GB (32 bits) cannot reliably be the source or destination of DMA operations. Therefore, the Red Hat Enterprise Linux 3 Update 2 kernel "bounces" all DMA operations to or from physical addresses above 4GB to buffers that the kernel pre-allocated below 4GB at boot time. This is likely to result in lower performance for IO-intensive workloads for Intel® EM64T as compared to AMD64 processors.?

I think we have enough proof to say the two chips are not 100% compatible, in fact it almost seems like a hack job. With what RedHat have suggested, in order to fully test Intel?s 64-Bit solution, we are going to need more than 4GB of memory. This seems completely against the point of running at 64-Bit.


OUCH!!!!!

 

[friedrice]

Instead of comparing say, an Intel 3.5 ghz to an Athlon 3500+, you should compare prices. In other words, compare a Athlon 3500+ to an Intel proc of the same price. I wonder what kind of results you would find then 😀

 

[Zebo]

Well a $200 3400 is faster than all Intel processors globally speaking.😀

Stock, nothing released in the P4 or A64 realm can touch this processor price/performance. You know it's faster than 3500 in every benchmark?