Athlon Faster than P4 for FEA

[Clocker]

Cool!

http://www.aceshardware.com/#55000490

It is nice to see a benchmark using the software work with (LS-Dyna, a multi-physics analysis package by LSTC (Livermore Software Technology Comp.) www.lstc.com

I only wish they would have included a bench run on a HP J6000 (dual 64-bit PA-8600 RISC CPUs with 1.5MB onchip cache each) and 8GB RAM (or similar) for comparison...


Clocker

 

[Buck]

Cool!

http://www.aceshardware.com/#55000490

It is nice to see a benchmark using the software work with (LS-Dyna, a multi-physics analysis package by LSTC (Livermore Software Technology Comp.) www.lstc.com

I only wish they would have included a bench run on a HP J6000 (dual 64-bit PA-8600 RISC CPUs with 1.5MB onchip cache each) and 8GB RAM (or similar) for comparison...
Clocker

For a guy with an abnormally small cranium, and with an even smaller brain, you have a great interest in speed and power (we have yet to x-ray Tea’s cranium). Could this be a subconscious desire, which attempts to compensate for the well-documented, and x-rayed, lack of neurological matter and activity? Or is this an attempt to learn how the power of computing can be used to generate a six-pack of beer? (Either way, I’m behind you all the way – more power, more power!)

 

[Cliptin]

I figured he was analysing donuts.

 

[LiamC]

I posted my take on these benchs on my front page but here goes:

Now this (FEA benchmark) is both good and bad - good in the sense that this is yet another way to analyse performance - and in this particular setup, the Athlon MP and XP absolutely smacks the Intel offerings silly. Engineers being a conservative lot, it might help to get the message through that there are faster and cheaper alternatives to traditional hardware solutions out there - which encourages competition, which lowers prices. Das ist gut, ja?

The bad side is that there is not a lot of detail on the benchmark itself, which makes it very difficult to decide what is being measured. Here, (in the Aces message board)

http://www.aceshardware.com/forum?read=65045633

is an explanation of how the particular problem benchmarked could be solved, and how if done in a different manner, the Intel offerings might have the performance crown. This just illustrates further one of my mantra's - benchmarks don't tell the whole story - you must (absolutely must) know what is being measured before joining in a pissing contest (mine is bigger than yours nyah, nyah). BTW, the full thread of people's opinions on this "benchmark" starts here:

http://www.aceshardware.com/forum?read=65045474

Personally, what I think this benchmarks shows is that "traditional" engineering/scientific applications may be better served when run on the AMD platform than the Intel - unless the application has been specifically coded for the Pentium 4.