AMD rock Intel!

[ratas911]

Looks like no one has done this... So I only got this to say. AMD rule this time! FX trash EE like there is no tomorrow.

http://www.ukgamer.com/article.php4?id=152&page=13
===========================================================
There's very little we can say other than if every application that is recompiled to take advantage of AMD64 architecture produces similar performance gains, the Athlon 64 will be the processor to own.

The FX-51 excels in public key operations, especially when we use a recompiled binary for 64-bit mode operation. The difference between using the 32-bit binary and the 64-bit AMD64 binary is over 160%!!!.

In private key operations, when running in 64-bit mode we see a 261%!!!! increase in performance.


DELL sucks!
http://www.anandtech.com/systems/showdoc.html?i=1920
==========================================================
The Athlon64 FX is still the best-performing gaming machine that we have tested. We are in the process of testing a similarly equipped Athlon64 FX gaming rig from ElitePC, and we are anxious to compare performance to the XPS. It should be an interesting comparison.


Centrino sucks!!
http://www.anandtech.com/mobile/showdoc.html?i=1916
===========================================================
We are currently benchmarking the M:855 in more depth, and part 2 will cover what has become known as the AnandTech standard mobile review, as well as diving into the battery life of this powerhouse notebook. Voodoo claims that it fairs better than their Centrino notebook, which is indeed a very exciting claim to make. Considering it based on a K8T800 chipset and uses an Athlon 64 DTR, we aren¡¦t expecting incredibly high battery life. After all, these are desktop and desktop like components, which aren¡¦t designed to be that forgiving on power consumption. We will know more on this subject soon once we get the final benchmark numbers in.

 

[Whisper]

I don't believe these numbers. All it has extra in 64-bit mode is longer and more registers. Now, longer registers make RSA computations faster, but there already are 64-bit MMX registers. So clearly their 'legacy' code does not use MMX. And the extra registers are not going to double performance either unless you specifically do not optimize for low register numbers in legacy code. Without any doubt, this benchmark is seriously biased...

Nonetheless the Athlon 64 is an impressive chip. So I believe that biased benchmarks like this are not 'necessary' from a PR point of view. The chip can prove itself just fine under normal circumstances. Besides, 99.9% of applications is still 32-bit so benchmarks like this won't increase sales.

Oh, and DELL and Centrino do rock, on their own terrain!

 

[superpsa]

I never really trust sites like Anandtech where they seem biased to one chip brand...(now I wonder why I'm on TH )

AMD Is An Anagram Of MAD, Intel Is An Anagram Of INLET, Cyrix...Ah Who Cares?

 

[Guest]

>I don't believe these numbers. All it has extra in 64-bit
>mode is longer and more registers. Now, longer registers
>make RSA computations faster, but there already are 64-bit
>MMX registers

I don't know a whole lot about how MMX works, but actually I think the benchmark is perfectly credible and not "biased"; its cherry picked though, and its just not very representative for anything but RSA. I've never seen or heard MMX would boost RSA scores, while every new 64 bit architecture ever intro-ed used RSA scores to show off it performance (Itanium did, Power+ did, opteron/K8 does now). In theory, RSA performance can increase by 4x on a 64 bit cpu over a 32 bit design. But like I said, RSA is just RSA, and very few other apps require or use 64 bit integers. Don't think these speedups are any indication of how other AMD64 binaries will perform, but they show what a 64 bit cpu can do when working with 64 bit integer math (which is very rare).

As for your MMX thing; here is a quote I found:"The MMX data path includes two 64-bit ALUs, one barrel shifter, one integer memory/register access unit, and one multiplier capable of performing four 16x16->32-bit multiplications in a single instruction cycle."

Sounds like MMX can not handle 64 bit multiplications, but it can do several simultaneous 32 bit multiplies. More over, if RSA encrypting benefitted from MMX, rest assured software would have been out for nearly a decade to support it, if for no other reason than marketing the original Pentium MMX.

In short; trust the scores and the bench, but just don't put too much weight on the results.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[ratas911]

Opteron 248 Review

http://www.tech-report.com/reviews/2003q4/opteron-x48/index.x?pg=1


"Conclusions
The last time we looked at a pair of Opteron chips running in SMP, we were comparing the Opteron 240 to the Xeon 2.66GHz, and frankly, the Opteron 240 got its tail whipped. This time around, it's a very different story. Thanks in part to its built-in memory controller, the Opteron's performance seems to scale up exceptionally well with clock speed increases. Overall, a single Opteron 148 is every bit as fast as a Pentium 4 3.2GHz Extreme Edition chip, and running together, a pair of Opteron 248s makes for the fastest workstation system we've ever laid hands on.