[link lost in import]
Thursday, August 19, 1999 - 14:47 CDT
AMD Athlon Test Results Questioned
Questions have arisen regarding Advanced Micro Devices Athlon test
results since a Toronto-based company revealed that AMD modified its
test suite to give the new processor an unfair advantage over the
Pentium III. According to this statement, AMD engineers optimized the
test suite in violation of the licensing agreement, something no other
company, incuding Intel, has been given the chance to do. This means
AMD compared the optimized Athlon results to the generic test suite
used with the Pentium III.
AMD's test results, referred to in this release announcing the
shipping of the Athlon processor, demonstrated that in one of two
tests, the 600 MHz Athlon outperformed the 600 MHz Pentium III by over
30 percent. An AMD representative told TechWeb the customized test
suite was fully disclosed on page 39 of a 42-page software performance
guide accompanying the test results. However, in the guide, AMD does
not mention optimizing its code for other test suites. Other firms
have said the Athlon generally outperforms the Pentium III in a
variety of applications.
Reported by: David Laprad
--
* rrevved at mindspring dot com
Couple things:
1) Its only one benchmark they've admitted to optimizing so far. If
thats it, and they go "whoops! Sorry", It'll be ok.
2) Read the original Techweb article at:
http://www.techweb.com/wire/story/TWB19990819S0008
The Avault article is just a really obvious paraphrase. I suppose
that's ok, but it makes people wonder how many places actually put
*out* news instead of just regurgitating news from other sites....
Ah well. The best (or worst, depending on your point of view) type of
meta-content that I saw was when one of those Freespace2 sites
interviewed Evil Avatar for his thoughts. A fan site interviewing a
news site guy.... yeah.... that's real content! :-)
snip
snip
Really? What else does that phrase work for in big business?
Zog
According to 3DNow.net:
"Futuremark allowed AMD to modify a version of 3DMark '99. Even still the
Athlon outperforms the PentiumIII in many other NON-optimized benchmarks."
As if Intel never did that..duh! Time for the old Has-Been to start being
competitive again. Wake Up Intel!
If this is the case, AMD is looking for trouble in courts.
Falcification is a criminal act. And if they falcified benchmark results and
people baught the product because of these results:
- They should request a full refund, else the courts, will be more than
happy to handle the case. And I don't think that the company wants to get
this type of media attention. I'm sure that Intel will be also happy to
serve it's competitor's curtomers.
I recently aquired a AMD K6-III 450 MHz and I will be *NOT VERY HAPPY* if I
will find out that it doesn't perform as it's supposed to be.
:-|
Then why Futuremark is the first to issue a press release accusing AMD
falsify the benchmark?
Foxbat
>> According to 3DNow.net:
>> "Futuremark allowed AMD to modify a version of 3DMark '99. Even still the
>> Athlon outperforms the PentiumIII in many other NON-optimized benchmarks."
>Then why Futuremark is the first to issue a press release accusing AMD
>falsify the benchmark?
I don't know, but the fact remains that the K7 still outperforms the PIII
even without these alleged optimizations. Everything else is simply fuel
for skeptic so IMO it's no big deal. Some people will obviously "make" a
big deal about it for their own gains. Hell, the graphic card industry has
been doing this sort of thing for years and nobody really cared. Marketing
mistake or not, the K7 is still a kick ass CPU, faster than the PIII and I
am routing for it and AMD all the way.
Yes, so what? It is about 15% faster than a P3 running at the same mhz.
About time, AMD has had a very crappy floating point unit for years on end.
So its nice to see them actually get it right finally. So congrats to AMD.
Competition is good. But I wonder want competition AMD will compete with
once the IA32/IA64 architectures from Intel enter the consumer market since
its NOT x86 in any matter whatsoever. :) Perhaps they can compete with
whomever it was that bought up Cyrix.
check out the review of the athlon on CNET,
I've seen a couple of tables for frame rates for games on athlon/intel
platforms and...the Athlon wins - and who buys an athlon for non *** use
anyways ? anyone who cites specific benchmarks is taking a tiny part of the
bigger picture and I'm sure any proccessor can be made to look good or bad
on one benchmark
"so what?"? People don't understand the significance of the K7 I guess.
But then again, you are obviously a skeptic so Im not going to waste my time
trying to explain the K7's importance in the market and what it means to
Intel and to competition.
Comparing MHz to MHz doesn't mean squat when you are comparing a completely
different architecture. You can put wider tires on your Chevette than your
neighbors Lotus and then brag about it but that doesn't mean that it will
handle better than the Lotus. You have to use real world apps to judge it.
You can't compare MHz between apples and oranges.
Yeah well now it's time for the Has-Been to play catch up. Anyway, what the
K6 lacked in floating point it made up for in other areas.
Ditto.
With the K7 in the market, IA64 will be overpriced and not necessary. There
will not be a market for it IMO and the K7 will be one of the reasons for
this. Anybody that purchased a K7 system is not even going to notice or
care about IA64 if and when it emerges. As you posted above "so what?".
AMD optimized the driver for it's 3Dnow! instructions.
Futuremark was fully aware of it and AMD stated that they did it weeks ago.
Intel did the same for the P3 and SSE.
Being that CPU's are so diverse now it is an everyday occurance.
As usual grade school intelligence shines on the Internet.
Read this..
http://www.fullon3d.com/
http://www.3dmark.com/press/rel3.html
3D Mark 99 and Athlon DLL | armin | 15:24 EDT
There has been a row over the use of a DLL
that had been optimized
by AMD and added to the 3D Mark 99 MAX
benchmark software
shipped with their review systems.
Futuremark released the following
statement on the issue. I'd assume
they received heat from Intel and chose to
move themselves into a
neutral position.
This was followed by this article on EB
News, which was a poor
interpretation of the press release.
Especially there is no claim by
Futuremark that their trademark is
violated by AMD, but rather that
such a violation would occur by using 3D
Mark scores with the Athlon
DLL in a way that was not following their
detailed instructions in the
press release.
More specifically still there are no
claims made that the benchmark was
rigged by using the DLL in question - my
information is that it simply
makes use of the new 3DNow! instructions,
which, among other
things, will crash the K6-2 and K6-III if
it is used.
The integrity of the benchmark is not
threatened, the performance is
just optimized for the newer CPU that was
not available when 3D Mark
99 MAX was released. Any questions whether
Intel would have pushed
for a SSE DLL if their Pentium III had
come out after 3D Mark 99 MAX
was launched?
In closing, let me quote one of our most
reliable 3D graphics industry
sources:
"... As the register has
demonstrated, not all journalists have the ability
to write anything but inaccurate,
poor quality, sensationalist
journalism. I was disgusted by their
article and the article that it was
spawned from. [...] This is gutter
press."
AMD has the K8 in beta and the K9 in the planning stages.
The did not build the best and fastest CPU in the World by sitting on thier
butts.
And the 200-400MHz Alpha bus that the K7+ are using will not be beat anytime
soon.
They didn't accuse AMD about the benchmarks, that's just the way the
press twisted it (if you read the press release and compare to the
press versions you'll notice that press have given the story a _huge_
spin :). In the press release they make a point about how the scores
should be _reported_ so that no mixups would happen between the
downloadable and Athlon optimized version. 3DMark 2000 is supposeed to
have optimizations for everything that's on the markets this fall.
Guess, they figured that they can only officially support their own
products... ;)
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
On Tue, 24 Aug 1999 18:08:22 GMT, Dave Swanson
I've seen quite a few independently run benchmarks using everything
from games to apps and winstone etc. The Athlon *is* faster in every
way.
Amplified
