Sign in with
Sign up | Sign in
Your question

AMD has the edge....

Last response: in CPUs
Share
August 15, 2001 10:13:48 PM

http://www.idius.net/fpucomparison/

AMD REJOICE. FUGGER, MELTDOWN, AND OTHER PENTIUM IV LOVERS BOW DOWN!!

This PROVES that AMD is faster and more efficient, WAY more efficient clock for clock than the Pentium IV garbage. Quote from the site: "After looking at these results you can clearly see that Pentium 4 NEEDS a very large clock rate advantage and high memory bandwidth to even stay competitive with Athlon when the code is not SSE-2 optimized. Clock for clock the floating point unit of Pentium 4 only offers about half the performance of the original Pentium!"

Hahah you Pentium IV works WORSE than an original Pentium in FPU.....rofl that chips REALLY sucks moocho.

I now procede to laugh at how you Pentium IV people paid more for WAY less.....HAHAHAHA

More about : amd edge

August 15, 2001 10:17:19 PM

<A HREF="http://www.idius.net/fpucomparison/" target="_new">Click</A>

The words "Back to AMDZone" at the bottom worry me...



<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
Related resources
August 15, 2001 10:36:38 PM

No!!!!!!
No Van's Hardware links, no, no, no.



<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
August 15, 2001 10:39:18 PM

If it's solid data, I take it.
August 15, 2001 10:56:32 PM

1. Although the article was published today, he says that BapCo's page "continues to cite 2200 Mission College Blvd. as its official address" despite the fact that the source is a webpage almost <font color=green>two years</font color=green> old.

2.
Quote:
BAPCo, has seen its SysMark application level test suite become perhaps the most widely used CPU analysis tool in the industry.



Most people would disagree with that.

3. "By controlling the very lens through which all CPUs are viewed, Intel can show their processors in the best possible light while placing its rivals' products in the shadows."
Is a far cry from
"It is now evident that BAPCo's SysMark 2000 included Intel processor specific instructions to make the chip giant's CPUs look good, while handicapping the competition."

Having Intel optimized code is hardly "controlling the lens"


4.
Quote:
Despite its 200 MHz handicap, the 1.2 GHz Palomino easily defeated its faster clocked older brother. Although the Palomino also supports hardware prefetching, the performance difference is almost entirely due to SSE since the SysMark Photoshop test is relatively bandwidth insensitive as can be seen in this old chipset review


He claims that <font color=green>only</font color=green> because of SSE, but has no way to prove it.

5.
Quote:
There is little remaining doubt that Intel's SysMark 2000 played favorites with the chip giant's own processors.


There is?

6.
Quote:
We mentioned well over a year ago our belief that SysMark 2000 had Intel SSE optimizations to the exclusion of AMD's 3dNow! enhancements. However, testing this conjecture was difficult.


Duh. He states this as fact later in the article.
His proof that this is true? A link to THG that states
"For instance, deliberately choosing applications optimized for Intel's SSE instructions but not optimized for AMD's "3D Now!" instructions would skew the results in favor of Intel platforms."
No mention of BapCo or SysMark anywhere on the page.
There's also the fact that he wrote that article.



All in all, Van Smith is the kind of person that takes rumors and presents them as fact. That's why I dislike him so strongly.



<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
August 15, 2001 11:49:52 PM

<A HREF="http://www.spec.org/osg/cpu2000/results/cpu2000.html" target="_new">SPEC CPU 2000 Results</A>:

(Notes: Higher is better. The number in parenthesis is the average of the two previous numbers, computed by me.)

Integer tests:
Dell -- Precision WorkStation 330 (1.80 GHz Pentium 4) - 599 619 (609.0)
Dell -- Precision WorkStation 530 (1.70 GHz Pentium 4) - 575 593 (584.0)
Intel - Intel D850GB motherboard (1.60 GHz Pentium 4) - 552 563 (557.5)
Dell -- Precision WorkStation 530 (1.50 GHz Pentium 4) - 526 543 (534.5)
AMD - Gigabyte GA-7DX Motherboard (1.40GHz Athlon) - 495 554 (524.5)
AMD - Gigabyte GA-7DX Motherboard (1.33GHz Athlon) - 482 539 (510.5)
Dell -- Precision WorkStation 330 (1.40 GHz Pentium 4) - 501 517 (509.0)
AMD - ASUS A7V Motherboard (1.30GHz Athlon) -------- 438 491 (464.5)

FPU tests:
Dell -- Precision WorkStation 330 (1.80 GHz Pentium 4) - 615 631 (623.0)
Intel - Intel D850GB motherboard (1.70 GHz Pentium 4) - 598 608 (603.0)
Intel - Intel D850GB motherboard (1.60 GHz Pentium 4) - 578 587 (582.5)
Dell -- Precision WorkStation 530 (1.50 GHz Pentium 4) - 551 566 (558.5)
Dell -- Precision WorkStation 330 (1.40 GHz Pentium 4) - 527 541 (534.0)
AMD - Gigabyte GA-7DX Motherboard (1.40GHz Athlon) - 426 458 (442.0)
AMD - Gigabyte GA-7DX Motherboard (1.33GHz Athlon) - 414 445 (429.5)
AMD - ASUS A7V Motherboard (1.30GHz Athlon) -------- 348 374 (361.0)

Note that these SPEC tests are presented in source code form, giving both companies a good look at how the benchmark operates. Alterations to the source code is not permitted. The source code does not contain any processor-specific instructions. Compilers are allowed to optimize in any fashion they choose so long as they work with the same unaltered source code.

-Raystonn



= The views stated herein are my personal views, and not necessarily the views of my employer. =
August 16, 2001 12:00:15 AM

Raystonn, there is no way in hell the P4 beat an athlon in a FPU only test. There had to be some SSE2 support or something in there that gave the p4 the upper hand according to your data. Everyone knows the Athlon FPU is superior to anything intel has to offer right now! Even the P3 FPU is superior to the P4 FPU. So I take that data with a grain of salt!
August 16, 2001 12:05:58 AM

That is bs like he said. Everyone knows that without SSE2 support in 3D games, the PIV is dead in the water. This is a fact that the PIV FPU is NOT as strong as Athlon's.
August 16, 2001 12:25:06 AM

"...The source code does not contain any processor-specific instructions. Compilers are allowed to optimize in any fashion they choose so long as they work with the same unaltered source code. ..."

I don't think he precluded the use of the compiler generating SSE2 optimized code.
August 16, 2001 12:35:38 AM

"there is no way in hell the P4 beat an athlon"

With a properly optimizing compiler, it does.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =
August 16, 2001 12:39:04 AM

"Everyone knows that without SSE2 support in 3D games"

Who said anything about not using SSE2? I'm sure the executable produced by the compiler included SSE2 optimizations. This is the main difference between AMD's 3DNow attempts and Intel's SSE/SSE2 instruction set. SSE and SSE2 are automatically generated by the compiler without having to do anything special in your code. Adding SSE2 support to your application can be as easy as simply recompiling your application with the correct options. Using 3DNow however requires special libraries or assembly code. It's a huge hassle.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
August 16, 2001 2:17:45 AM

Gee and we're just going to run the SPEC FPU benchmark the whole time when we get our computers!

--
It's Princess Leia, the yodel of my life. Give me my sweater back or I'll play the guitar.
August 16, 2001 2:29:05 AM

the current P4 is a somewhat dampened design and I don't think the P4 was designed to run 100% perfect with legacy code, this is why the P4 has 144 ***new*** instructions!, however it does really well in multimedia programs and games that were coded for it, no one can deny that! you should really educate yourself first before posting a moronic biased benchmark and from someone who did not actually test a P4.

"<b>AMD/VIA!</b>...you are <i>still</i> the weakest link, good bye!"
August 16, 2001 8:13:24 AM

LOL!!!!!!!!!!
August 16, 2001 9:18:32 PM

It’s true that the Intel P4 processes data faster than the AMD (P4 has a higher clock speed), but the P4 is also less efficient. That’s why benchmarks that test Integer and FPU performance often go in favor of the P4. It all depends on the how the benchmark software receives the results from the CPU. I only pay attention to real world application and game tests (AMD wins most of 'em).

The Intel just wastes its clock cycles. (Partly because of it’s 8k of L1 cache)


- I got a board too: http://www.impactsites2000.com/cgi-bin/ikonboard/ikonbo...
Anonymous
a b à CPUs
August 16, 2001 11:16:34 PM

WOW
amd isnt optimized for sse2 or developers dont optimize greatly for 3dnow

wow
amazing how amd can pull its weight like it does!

--call it what you wish, with this machine I can make mercury flow in 3 directions at once--
August 20, 2001 10:44:36 PM

Quote:
SSE and SSE2 are automatically generated by the compiler without having to do anything special in your code.


Can you prove this? A friend of mine who's a programmer (C/C++, assembly, etc.) told me that the exact opposite is true. With SSE, SSE2 and 3DNow, you had to explicitly program it into your code.


Sorry to raise an old post from the dead :) 



<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
August 20, 2001 11:06:39 PM

Just grab the latest trial version of Intel's C/C++ compiler. It is available on Intel's website. This is a plugin that installs on top of Microsoft's Visual C++, the industry standard Windows compiler. You simply place a checkmark indicating you would like to use the Intel compiler instead of the VC++ compiler. You continue to use the same IDE and libraries, but with the optimizations available from a better compiler. There's also a Linux version available.

No other chipmakers have a compiler available for their CPUs. The use of 3DNow requires you to alter your source code; something very few programmers have been willing to do. A simple recompile requires no real effort and, when the options are selected, can make use of MMX, SSE, and SSE2 instructions in addition to other processor specific optimizations.

You can optimize for the whole range of CPUs: generic x86, Pentium, Pentium MMX, Pentium Pro, Pentium II, Pentium III, and Pentium 4. You can also have it generate code that will detect your CPU and run different versions of optimized sections of code, depending on what CPU you have. None of this requires any change to the actual source code for your project. It's all done automatically for you by the compiler.

There are libraries available to specifically make use of MMX, SSE, and SSE2, similar to the libraries available for 3DNow, but these are no longer necessary. If you have any doubt about the execution speed of applications generated by this compiler, check the Spec site I mentioned above. You will see that AMD chose to use Intel's compiler for their benchmarks. This says a great deal.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
Anonymous
a b à CPUs
August 21, 2001 7:57:07 AM

>You will see that AMD chose to use Intel's compiler for
> their benchmarks. This says a great deal.

It does.. doesnt it.. intel is king of compilers; there is no doubt about that. If only AMD could release a compiler that was half as well optimized for their own CPU's,...

Also, I have to credit Intel for not building compilers that artificially handicap AMD cpu preformance. (of course, no one can blame them for not optimizing for Athlons or 3D )now..).

Its really the compiler team that is helping intel keep pace. I wonder if they'll succeed doing so with the Itanium...

---- Owner of the only Dell computer with an AMD chip
August 21, 2001 2:58:31 PM

Cool, thanks Raystonn.



<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
August 21, 2001 3:20:32 PM

Actually, i wonder if Intel used to optimize compiler code for AMD? Surely Intel likes the fact that AMD exists and is doing pretty well. AMD may have a little more market share than Intel would like, but overall I think Intel's pleased with them.

Althons and Pentiums are just melted rock. Who’s rock is better? Who cares, let’s play some games
August 21, 2001 3:54:49 PM

Codewarrior has a compiler that compiles and optimises normal code into 3D-Now! instructions.


<font color=red><i>Tomorrow I will live, the fool does say
today itself's too late; the wise lived yesterday
August 21, 2001 4:10:42 PM

At 22%, I'd say "a little" is a bit of an understatement, my friend :) 



<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
August 21, 2001 6:51:24 PM

I doubt they'd want AMD to have under 10 (that reaks of monopoly. Probably 15 would be perfect. so 7% difference? I guess it's ur perspective. Percentage wise to intel it's not that much. Dollarwise to both companies, it is quite a hefty amount. So ya, maybe an understatement :) 

Althons and Pentiums are just melted rock. Who’s rock is better? Who cares, let’s play some games
August 21, 2001 10:39:54 PM

Codewarrior is apparently so good that AMD dumped it and used the Intel compiler in all the benchmarks. Go figure. Codewarrior has never been a very good compiler as far as optimizations go.

Microsoft's Visual C++ is the industry standard for Windows development. The Intel C/C++ Compiler plugs into VC++ and works without changing your source code or projects. This is a solution that requires no additional work and provides the best performance, even for AMD CPUs.

-Raystonn


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