noko
Distinguished
- Jan 22, 2001
- 2,414
- 1
- 19,785
The article was By:
Frank Völkel
Bert Töpelt
Not Tom, just as a reminder. Interesting article and the point about noise and price seems to reflect the $50 price tag for the HS. To get an Athlon or faster Athlon XP cool and noise free as a P4, the price does indeed go up. My computer I think sounds like a vacuum cleaner at times, sucking up all the dirt in the room meaning I have to change out the bag (inside case) periodically. Anyways I thought the review was a good review. Would like to know what compatibility problems where shown by the XP processors and not by the P4's.
Frank Völkel
Bert Töpelt
Not Tom, just as a reminder. Interesting article and the point about noise and price seems to reflect the $50 price tag for the HS. To get an Athlon or faster Athlon XP cool and noise free as a P4, the price does indeed go up. My computer I think sounds like a vacuum cleaner at times, sucking up all the dirt in the room meaning I have to change out the bag (inside case) periodically. Anyways I thought the review was a good review. Would like to know what compatibility problems where shown by the XP processors and not by the P4's.
Kelledin
Distinguished
- Mar 1, 2001
- 2,183
- 0
- 19,780
It's as simple as checking a bit in the CPU feature register. That's the <b>Intel-specified</b> method for properly detecting SSE and SSE2, and it's far simpler than grepping the CPU ID string for "GenuineIntel". I may not be a benchmark programmer, but I'm at least dumb enough not to screw around with regards to established standards like that. :tongue:
BTW, my condolences for your idol being in second place. :lol:
Kelledin
"/join #hackerz. See the Web. DoS interesting people."
BTW, my condolences for your idol being in second place. :lol:
Kelledin
"/join #hackerz. See the Web. DoS interesting people."
ROFL
Meltdown, I have proven you wrong on so many occasions it's not funny. I hope you're not trying to start another argument.
Would you care to provide a detailed reason why it can't look for SSE itself instead of 'GenuineIntel'? I'm not a programmer, which is why I asked someone to correct me if I was wrong.
The P4 beating the XP by so much in WinACE is sort of strange, but not in the way some might think. The P4's greater (effective) bus speed and RAM bandwidth should easily carry it forward in calculations like these that are simpler integer operations.
Now the question is, why didn't the MP3 conversion test show the same results? I would think they are very similar, but the results were very different. Anybody care to comment on that?
Further comments:
I added up the totals of the tests, and the score is P4: 9, XP: 10. Important to note is that Frank and Bert show the P4 winning the 2nd SiSoft Multimedia test, when it did not. Here is an excerpt from my email to Frank:
Interesting, to say the lest. Considering it is a multimedia test, you'd think it would swing the other way.
Anyhow, feel free to comment or correct.
<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
Meltdown, I have proven you wrong on so many occasions it's not funny. I hope you're not trying to start another argument.
Would you care to provide a detailed reason why it can't look for SSE itself instead of 'GenuineIntel'? I'm not a programmer, which is why I asked someone to correct me if I was wrong.
The P4 beating the XP by so much in WinACE is sort of strange, but not in the way some might think. The P4's greater (effective) bus speed and RAM bandwidth should easily carry it forward in calculations like these that are simpler integer operations.
Now the question is, why didn't the MP3 conversion test show the same results? I would think they are very similar, but the results were very different. Anybody care to comment on that?
Further comments:
I added up the totals of the tests, and the score is P4: 9, XP: 10. Important to note is that Frank and Bert show the P4 winning the 2nd SiSoft Multimedia test, when it did not. Here is an excerpt from my email to Frank:
Interesting, to say the lest. Considering it is a multimedia test, you'd think it would swing the other way.
Anyhow, feel free to comment or correct.
<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
mr_gobbledegook
Distinguished
- Sep 3, 2001
- 468
- 0
- 18,780
Well Tom finally got out an offical report on the Athlon XP and P4. The report is well written and I don't mind waiting if the results are accurate and fair.
As with the Windows Media encoder I pointed this out in one of my previous posts when Anandtech reviewed the Athlon XP and is old news. I support Tom in that he did not enable SSE for Win Encoder since there is no offical patch release for it. Only enthusiasts will see benifits (who change the CPUID string in the dll) not the average user therefore is an accurate to use the non SSE optimised benchmarks since that is what will be portrayed in real life for the vast majority of users. Out of interest the Athlon XP is on par with the P4 for Windows Media Encoding according to Anandtech who enabled SSE for the Athlon XP.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
As with the Windows Media encoder I pointed this out in one of my previous posts when Anandtech reviewed the Athlon XP and is old news. I support Tom in that he did not enable SSE for Win Encoder since there is no offical patch release for it. Only enthusiasts will see benifits (who change the CPUID string in the dll) not the average user therefore is an accurate to use the non SSE optimised benchmarks since that is what will be portrayed in real life for the vast majority of users. Out of interest the Athlon XP is on par with the P4 for Windows Media Encoding according to Anandtech who enabled SSE for the Athlon XP.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
If there was a moral question to a benchmark, one which requires an explination, they should NOT have used the benchmark, if si soft cannot use the features of both processors, OR MORE APTLY, if both processors can use those features but si soft only recognizes one of them, then the test should either be patched, or not ran, period.
~Matisaro~
"The Cash Left In My Pocket,The BEST Benchmark"
~Tbird1.3@1.5~
~Matisaro~
"The Cash Left In My Pocket,The BEST Benchmark"
~Tbird1.3@1.5~
Tom? What happened?
I used to rely on you to be objective and unbiased even in the face of Intel. I could always tell you liked AMD but in your test/comparison articles you always managed to stay on the level.
I found a few inconsistencies and over-tones in your article in your article that kinda bothered me. Take for instance….
“We've got our performance winner in this extensive CPU test - the AMD Athlon XP 1800+ tops the Intel Pentium 4/2000 in most of the applications benchmarks we selected.”
I counted. 9-10 Intel
The other thing I noticed, was the general tone of the article, in many of the comments where a P4 would come out on top of a test there was an explanation or excuse of why the Athlon couldn’t pull it off or a point would be made to frown on the Intel and praise the AMD. Such as:
“The "Lightscape" benchmark produced very clear results - the Pentium 4 CPUs, with their faster clock speeds, were clearly in the lead, while the fastest AMD processor, the Athlon XP 1800+, took fifth in the rankings. The big loser is the Intel Pentium 4/1400.”
There’s a loser in every test, and many times it was an AMD, besides the P4/1400 was only 2% slower than the AMD?? Not a large margin.
Or
“The Lame MP3 Encoder under Windows XP is used to convert a 178 MB sound file from a WAV format to a MPEG-1 Layer 3 format. In this discipline, the Intel Pentium 4/2000 makes a clean sweep of the competition. Nonetheless, the new Athlon XP CPUs don't cut a bad figure, either.”
I’m not a raving Intel fan or anything (I own both) but I think this kind of writing belongs in an editorial rather than a comparison test.
<P ID="edit"><FONT SIZE=-1><EM>Edited by macutty on 10/31/01 04:39 PM.</EM></FONT></P>
I used to rely on you to be objective and unbiased even in the face of Intel. I could always tell you liked AMD but in your test/comparison articles you always managed to stay on the level.
I found a few inconsistencies and over-tones in your article in your article that kinda bothered me. Take for instance….
“We've got our performance winner in this extensive CPU test - the AMD Athlon XP 1800+ tops the Intel Pentium 4/2000 in most of the applications benchmarks we selected.”
I counted. 9-10 Intel
The other thing I noticed, was the general tone of the article, in many of the comments where a P4 would come out on top of a test there was an explanation or excuse of why the Athlon couldn’t pull it off or a point would be made to frown on the Intel and praise the AMD. Such as:
“The "Lightscape" benchmark produced very clear results - the Pentium 4 CPUs, with their faster clock speeds, were clearly in the lead, while the fastest AMD processor, the Athlon XP 1800+, took fifth in the rankings. The big loser is the Intel Pentium 4/1400.”
There’s a loser in every test, and many times it was an AMD, besides the P4/1400 was only 2% slower than the AMD?? Not a large margin.
Or
“The Lame MP3 Encoder under Windows XP is used to convert a 178 MB sound file from a WAV format to a MPEG-1 Layer 3 format. In this discipline, the Intel Pentium 4/2000 makes a clean sweep of the competition. Nonetheless, the new Athlon XP CPUs don't cut a bad figure, either.”
I’m not a raving Intel fan or anything (I own both) but I think this kind of writing belongs in an editorial rather than a comparison test.
<P ID="edit"><FONT SIZE=-1><EM>Edited by macutty on 10/31/01 04:39 PM.</EM></FONT></P>
Wrong. 10-9 in favor of the XP. Read my above post.
Interesting how the Intel fan is making the exact same points in the opposite respect. (In other words, using the same statements to mean a different thing.)
<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
it goes both ways the P4 is down 248 in FP but beats the AMD by 346 MIPS? so sit down, and as Tom posted the article the Intel beat AMD 10-9.
you should listen to your mother: "crying doesn't help anything"
you should listen to your mother: "crying doesn't help anything"
lhgpoobaa
Illustrious
- Dec 31, 2007
- 14,460
- 1
- 40,780
dunno bout you but im getting very... bored.. about the p4 vs athlon benchmarks.
another comment i disagree with is toms "intel has galloped to 2000mhz, while amd is creaping to achieve 1400 then 1533"
a. benchmarks mostly show the 1.533 giving the p4 2.0 a good spanking. why go further right now?
b. people are getting good overclocks from the 1.53, up to 1.8 with quality cooling.
ugh.
boring.
Is that a Northwoody in your pocket or are you just eXPited to see me?
another comment i disagree with is toms "intel has galloped to 2000mhz, while amd is creaping to achieve 1400 then 1533"
a. benchmarks mostly show the 1.533 giving the p4 2.0 a good spanking. why go further right now?
b. people are getting good overclocks from the 1.53, up to 1.8 with quality cooling.
ugh.
boring.
Is that a Northwoody in your pocket or are you just eXPited to see me?
I think your taking it too personal. I didn't have any problems with the review. Things to note:
a) It wasn't written by Tom
b) Cheap shots were made at both the p4 and XP
c) If windows media encoder won't recoginize the SSE support in the XP, why should the benchmark be patched to allow it? After all a benchmark is supposed to give you an idea of real world performance isn't it?
All in all the report showed us what most of us already new, the xp 1800+ and the p4 2000 are about on par, one being better than the other only depending on ones particular useage. What it comes down to is the price/performance ratio. At that the xp1800+ is the clear winner of the two. Also of interest is that as much maligned as AMD is for going to the performance rating at least it is obvious that they are at least using realistic and even conservative numbers for this.
Video editing?? Ha, I don't even own a camera!
a) It wasn't written by Tom
b) Cheap shots were made at both the p4 and XP
c) If windows media encoder won't recoginize the SSE support in the XP, why should the benchmark be patched to allow it? After all a benchmark is supposed to give you an idea of real world performance isn't it?
All in all the report showed us what most of us already new, the xp 1800+ and the p4 2000 are about on par, one being better than the other only depending on ones particular useage. What it comes down to is the price/performance ratio. At that the xp1800+ is the clear winner of the two. Also of interest is that as much maligned as AMD is for going to the performance rating at least it is obvious that they are at least using realistic and even conservative numbers for this.
Video editing?? Ha, I don't even own a camera!
The issue with the SSE portion of the benchmark is that, because the program used as an *example* of real world use, is flawed and, by using that program to test with, the real world performance gets skewed.
If 99% of programs recognize SSE support properly, but the program you choose to put in your benchmarks does NOT, then the benchmark provided is incorrect in the real world. And THAT is the issue to me.
mark-
When all else fails, throw your computer out the window!!!
If 99% of programs recognize SSE support properly, but the program you choose to put in your benchmarks does NOT, then the benchmark provided is incorrect in the real world. And THAT is the issue to me.
mark-
When all else fails, throw your computer out the window!!!
you know what was a bit unfair about this test? the athlon xp cpu's are a fairly new processor while the P4 has been out for some time now. i'm not saying that the athlon xp didn't deserve the crown-by all means it deserves it, i have always respected amd because an athlon 1.3 ghz WILL perform at 1.3 ghz while a P4 1.3 ghz will flucuate. anyways, looks like intel needs to work on that northwood a bit quicker if they want to compete against the athon xp.
I don't claim to know anything about everything, I just tell people what I know.
-PSB
I don't claim to know anything about everything, I just tell people what I know.
-PSB
Fatburger and anyone else that thinks SSE should remain off in SYSMARK 2001. Read the data I posted at HardOCP's forum. It ends all discussion on the issue. If we're trying to simulate the REAL WORLD, this benchmark is useless without it turned ON. RESPONSE BELOW:
As a whole the article was great. I only question one thing. The absence of the correct ID for windows media encoder (to identify the AthlonXP). This ID patch in no way changes the software itself. It merely tells the software WHO the chip is thats running said software. By leaving it out you are giving the world the impression SSE will not work at all out there. Which is simply NOT true. Only Windows Media Encoder has this problem, other software doesn't. So which gives a clearer picture of AthlonXP: One program that doesn't act as it should, or all the other software that DOES act as it should? What kind of logic is this? Below was taken from AMDZONE's response FROM BAPCO about the issue:
"In the case of Windows Media Encoder, support for hardware features is included in the application itself and therefore is represented in the benchmark accordingly. The benchmark itself does not run any code paths based on detection of a specific platform. That decision is left to the application software and the operating system, "not" the benchmark. This is consistent with the way actual applications behave. This is also the default behavior of other industry standard application-based benchmarks."
http://www.amdzone.com/articleview.cfm?ArticleID=836
Since Bapco is not SOFTWARE, and rather a SCRIPT that runs the software you clearly should be able to implement a newer version even (like Intel optimized compilers for SSE2 in SpecViewPerf). Let alone a simple little patch to SEE what is already available IN HARDWARE/Software already. It's just not turned on. Previously INTEL was allowed to modify the compiler that IS used in SPEC viewperf to enable SSE2 (which actually REQUIRED a complete RECOMPILE of parts of code to use SSE2 that was NEVER in the code to begin with). Why is it OK for Intel and not AMD? Amd is not even modifying anything in how the CODE works. Merely telling the code who's knocking on the door. This is a far cry from what Intel pulled with spec (I saw no complaints from you over recompiled Spec benchmarks...and have heard NOBODY else complain either). Do you honestly think people will not patch the software if they are planning on ripping etc with WME7? The story is out now, and most will go get a patch immediately.
This will even affect Webmark. Is the same .dll running Media player in webmark? You bet. You love this benchmark but it DOES NOT reflect real world performance at home (people your site is aimed at...you pretty much benchmark home software/games). How many people out there have a web connection that is governed by their PC? Raise your hand. I suspect NOBODY is raising their hand. You are ALL governed by your ISP's servers, and usually [worse yet], the server your trying to get data from. This is why the business to business (the part where Intel blows Amd away, AMD wins the other two parts) benchmark is not indicative of WEB performance at all. "faster access" to the internet is not what you get from the P4. Even ZDnet proved this many times in PCmag. IE runs just as fast no matter what platform you're on. Webmark is just another Intel/Bapco scam at work.
Anandtech proved here http://www.anandtech.com/cpu/showdoc.html?i=1441&p=13
that a full 65% of one of the 3 tests in this benchmark (WEBMARK 2001) come from guess what: Windows Media Encoder 7. 37% of another (B2B part) of the 3 parts of this benchmark is using 3D visualizations (cult3D I guess, one straight from Intels list of SSE2 apps...one of the VERY FEW apps in existence). Forward looking benchmark is what Kyle calls it. Is "forward looking" very real world today? No, and how the world will look to us next year can't be clear to anyone today. Why run the benchmark? Clearly it has always been weighted towards Intel with 65% coming from ONE app WME7. While the parts Amd wins in are only weighted in the 12-26% range. Either turn on what is ALREADY there or dump this totally biased (until AthlonXP was released anyway) benchmark. Nobody I know rips or encodes anything with WME7 anyway (160bit max?...320bit CBR or at worst 320VBR is what everybody rips at, most pro's use Lame...it's free too). One of the primary benefits of the new chip is SSE. To leave it disabled when no code change is required to merely turn it on is crazy.
Doesn't everyone update Win2K every time a service pack comes out? Do you go and get game patches when you have problems? Did you crawl through all six service packs for NT4? Do you use new versions of Via's 4in1 drivers each time they are available for benchmarking that next article? Don't you flash your motherboard bios's before benchmarking a new review? It's not in the hardware, YOU'RE CHANING THE CODE MAN! Are you still using Detonator 3 drivers, or do you like the new performance you got from downloading the new detonator XP drivers? We live in a world of completely changing code all the time, yet you balk at a puny 12 text characters to ID a chip properly as 'AuthenticAMD'. Even when Microsoft worked on it and will implement it shortly! Give me a break. WE LIVE IN A WORLD OF PATCHES (UNFORTUNATELY). To leave it UNPATCHED only makes us all wonder about the validity of your site (another one bought by Intel?). If Kyle holds out until Sysmark/webmark 2002 (when Intel figures out how to beat AMD again via Bapco's fake benchmarks) I will be extremely disappointed in this site. The least you could do is post the results. Patching the OS or Application is not bapco's area, nor is it ILLEGAL. Patching their SCRIPT IS.
Warning people to NOT believe results of tests using the patch makes you sound like an Intel ad. Especially when the patch only ID's the chip, and does not in any way alter code. Twenty bucks says Intel (er, I mean bapco) plans to cheat again with Sysmark 2002. They are beavering away right now to find out where the P4 can beat an AthlonXP no doubt..ROFL. Release Northwood now, and charge $300 and I'll be happy. Athlons are so cheap I don't make money on them. P4's are so expensive not many want to buy them. AthlonXP will help (higher margins), but I need Intel's help too. Oh well, enough whining.
As a whole the article was great. I only question one thing. The absence of the correct ID for windows media encoder (to identify the AthlonXP). This ID patch in no way changes the software itself. It merely tells the software WHO the chip is thats running said software. By leaving it out you are giving the world the impression SSE will not work at all out there. Which is simply NOT true. Only Windows Media Encoder has this problem, other software doesn't. So which gives a clearer picture of AthlonXP: One program that doesn't act as it should, or all the other software that DOES act as it should? What kind of logic is this? Below was taken from AMDZONE's response FROM BAPCO about the issue:
"In the case of Windows Media Encoder, support for hardware features is included in the application itself and therefore is represented in the benchmark accordingly. The benchmark itself does not run any code paths based on detection of a specific platform. That decision is left to the application software and the operating system, "not" the benchmark. This is consistent with the way actual applications behave. This is also the default behavior of other industry standard application-based benchmarks."
http://www.amdzone.com/articleview.cfm?ArticleID=836
Since Bapco is not SOFTWARE, and rather a SCRIPT that runs the software you clearly should be able to implement a newer version even (like Intel optimized compilers for SSE2 in SpecViewPerf). Let alone a simple little patch to SEE what is already available IN HARDWARE/Software already. It's just not turned on. Previously INTEL was allowed to modify the compiler that IS used in SPEC viewperf to enable SSE2 (which actually REQUIRED a complete RECOMPILE of parts of code to use SSE2 that was NEVER in the code to begin with). Why is it OK for Intel and not AMD? Amd is not even modifying anything in how the CODE works. Merely telling the code who's knocking on the door. This is a far cry from what Intel pulled with spec (I saw no complaints from you over recompiled Spec benchmarks...and have heard NOBODY else complain either). Do you honestly think people will not patch the software if they are planning on ripping etc with WME7? The story is out now, and most will go get a patch immediately.
This will even affect Webmark. Is the same .dll running Media player in webmark? You bet. You love this benchmark but it DOES NOT reflect real world performance at home (people your site is aimed at...you pretty much benchmark home software/games). How many people out there have a web connection that is governed by their PC? Raise your hand. I suspect NOBODY is raising their hand. You are ALL governed by your ISP's servers, and usually [worse yet], the server your trying to get data from. This is why the business to business (the part where Intel blows Amd away, AMD wins the other two parts) benchmark is not indicative of WEB performance at all. "faster access" to the internet is not what you get from the P4. Even ZDnet proved this many times in PCmag. IE runs just as fast no matter what platform you're on. Webmark is just another Intel/Bapco scam at work.
Anandtech proved here http://www.anandtech.com/cpu/showdoc.html?i=1441&p=13
that a full 65% of one of the 3 tests in this benchmark (WEBMARK 2001) come from guess what: Windows Media Encoder 7. 37% of another (B2B part) of the 3 parts of this benchmark is using 3D visualizations (cult3D I guess, one straight from Intels list of SSE2 apps...one of the VERY FEW apps in existence). Forward looking benchmark is what Kyle calls it. Is "forward looking" very real world today? No, and how the world will look to us next year can't be clear to anyone today. Why run the benchmark? Clearly it has always been weighted towards Intel with 65% coming from ONE app WME7. While the parts Amd wins in are only weighted in the 12-26% range. Either turn on what is ALREADY there or dump this totally biased (until AthlonXP was released anyway) benchmark. Nobody I know rips or encodes anything with WME7 anyway (160bit max?...320bit CBR or at worst 320VBR is what everybody rips at, most pro's use Lame...it's free too). One of the primary benefits of the new chip is SSE. To leave it disabled when no code change is required to merely turn it on is crazy.
Doesn't everyone update Win2K every time a service pack comes out? Do you go and get game patches when you have problems? Did you crawl through all six service packs for NT4? Do you use new versions of Via's 4in1 drivers each time they are available for benchmarking that next article? Don't you flash your motherboard bios's before benchmarking a new review? It's not in the hardware, YOU'RE CHANING THE CODE MAN! Are you still using Detonator 3 drivers, or do you like the new performance you got from downloading the new detonator XP drivers? We live in a world of completely changing code all the time, yet you balk at a puny 12 text characters to ID a chip properly as 'AuthenticAMD'. Even when Microsoft worked on it and will implement it shortly! Give me a break. WE LIVE IN A WORLD OF PATCHES (UNFORTUNATELY). To leave it UNPATCHED only makes us all wonder about the validity of your site (another one bought by Intel?). If Kyle holds out until Sysmark/webmark 2002 (when Intel figures out how to beat AMD again via Bapco's fake benchmarks) I will be extremely disappointed in this site. The least you could do is post the results. Patching the OS or Application is not bapco's area, nor is it ILLEGAL. Patching their SCRIPT IS.
Warning people to NOT believe results of tests using the patch makes you sound like an Intel ad. Especially when the patch only ID's the chip, and does not in any way alter code. Twenty bucks says Intel (er, I mean bapco) plans to cheat again with Sysmark 2002. They are beavering away right now to find out where the P4 can beat an AthlonXP no doubt..ROFL. Release Northwood now, and charge $300 and I'll be happy. Athlons are so cheap I don't make money on them. P4's are so expensive not many want to buy them. AthlonXP will help (higher margins), but I need Intel's help too. Oh well, enough whining.
MichaelDaugherty
Distinguished
- Nov 1, 2001
- 1
- 0
- 18,510
I enjoyed both the article and the comments. I think tom's hardware guide to the best place on the web for computer information; I would be lost without it. I thought some of the comments were a little insensitive to the people who must have worked hard on this article.
I wasn't aware a patch was yet available for the real application, only the benchmark... The last I read MS said it would only be patched in the next official release of the software... That may be outdated though....
-* <font color=red> Under Offer </font color=red> *-
email for application details
mr_gobbledegook
Distinguished
- Sep 3, 2001
- 468
- 0
- 18,780
Morally yes it is wrong. BUT whether you like it or not it is a 'TRUE' representation of the performance of the Athlon XP at this moment in time, life is not fair.
Let me clarify this... As Anandtech stated this problem has nothing to do with BAPCO (who make Sysmark) it is more to do with Microsoft.
(Source:<A HREF="http://www.anandtech.com/cpu/showdoc.html?i=1543&p=5"" target="_new">http://www.anandtech.com/cpu/showdoc.html?i=1543&p=5"</A>
Specfically Microsoft's Windows Media Encoder determines whether to use SSE by looking at the CPUID. BAPCO happens to use Windows Media Encoder for thier benchmark.
It is up to AMD to work with MICROSOFT to release an offical patch to enable SSE support for Windows Media Encoder. When the patch is released only THEN will BAPCO use this 'patched' version of Windows Media Encoder in thier benchmarking software. This is the correct way to go about the issue.
By AMD releasing thier own unoffical patch will not solve thier problems, and BAPCO / Tom are correct in using thier current benching s/w until an offical patch is released.
Tom should not have excluded the benchmarks just because they are 'morally wrong' he is simply reflecting on the performance of the processor at this moment in time and he has always kept to this policy.
In addtion this benchmark is quite important since it shows how a processor copes with the ever increasing need to produce multimedia content for many people in the industry, therefore should be included in the report.
So when the next speed grade of the P4 or AthlonXP is released and benchmarked hopefully Microsoft may have released a patch and BAPCO in turn updated thier s/w.
But until then it is up to AMD to work with MICROSOFT to release a patch who in turn should notify BAPCO.
I know this seems unfair and doesn’t help AMD but it is the only way this can be resolved, AMD have a habit of doing things in an adhoc way and is about time they start doing things through offical channels, this is something they are starting to realise as they move into the big enterprise sector. I just hope they don’t pull these kind of practices when they release Hammer – here is a little scenerio I thought up...
AMD: "Oh here is small patch that will correct a bug in XX benchmark tool since it doesn’t properly utilise our 64 bit features ! Don’t worry a patch for the actual application hasn’t been released by the software company but I’m sure they will release one soon !"
REVIEWER:"Errr..Yeah OK whatever you say."
To put it mildly this simply looks BAD and ruins AMD’s reputation. Reviewers and major companies like SUN, IBM will look the other way.
I’m sure Intel have done similar things but 2 wrongs don’t make a right.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
Let me clarify this... As Anandtech stated this problem has nothing to do with BAPCO (who make Sysmark) it is more to do with Microsoft.
(Source:<A HREF="http://www.anandtech.com/cpu/showdoc.html?i=1543&p=5"" target="_new">http://www.anandtech.com/cpu/showdoc.html?i=1543&p=5"</A>
Specfically Microsoft's Windows Media Encoder determines whether to use SSE by looking at the CPUID. BAPCO happens to use Windows Media Encoder for thier benchmark.
It is up to AMD to work with MICROSOFT to release an offical patch to enable SSE support for Windows Media Encoder. When the patch is released only THEN will BAPCO use this 'patched' version of Windows Media Encoder in thier benchmarking software. This is the correct way to go about the issue.
By AMD releasing thier own unoffical patch will not solve thier problems, and BAPCO / Tom are correct in using thier current benching s/w until an offical patch is released.
Tom should not have excluded the benchmarks just because they are 'morally wrong' he is simply reflecting on the performance of the processor at this moment in time and he has always kept to this policy.
In addtion this benchmark is quite important since it shows how a processor copes with the ever increasing need to produce multimedia content for many people in the industry, therefore should be included in the report.
So when the next speed grade of the P4 or AthlonXP is released and benchmarked hopefully Microsoft may have released a patch and BAPCO in turn updated thier s/w.
But until then it is up to AMD to work with MICROSOFT to release a patch who in turn should notify BAPCO.
I know this seems unfair and doesn’t help AMD but it is the only way this can be resolved, AMD have a habit of doing things in an adhoc way and is about time they start doing things through offical channels, this is something they are starting to realise as they move into the big enterprise sector. I just hope they don’t pull these kind of practices when they release Hammer – here is a little scenerio I thought up...
AMD: "Oh here is small patch that will correct a bug in XX benchmark tool since it doesn’t properly utilise our 64 bit features ! Don’t worry a patch for the actual application hasn’t been released by the software company but I’m sure they will release one soon !"
REVIEWER:"Errr..Yeah OK whatever you say."
To put it mildly this simply looks BAD and ruins AMD’s reputation. Reviewers and major companies like SUN, IBM will look the other way.
I’m sure Intel have done similar things but 2 wrongs don’t make a right.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
Fatburger,
If you count all the benchmarks except the SiSoft ones. The result is 8-8. A tie between the P4 and the AXP.
If you count the SiSoft test as Frank did, P4 wins 2 and the AXP wins 1. Which would give an end result of 10-9 to the P4.
Ok, so you say the AXP won part of a SiSoft benchmark. Lets split the SiSoft up into 5 benchmarks instead of as Frank did 3.
Now the P4 wins 3 and the AXP wins 2. The final result would be 11-10 to the P4.
No matter what way you count. It is intresting that the person writing the article claims that one CPU 'won' over the other. More correct would be that the AXP 1800+ is basically tied with the P4 2000.
/Knossos
If you count all the benchmarks except the SiSoft ones. The result is 8-8. A tie between the P4 and the AXP.
If you count the SiSoft test as Frank did, P4 wins 2 and the AXP wins 1. Which would give an end result of 10-9 to the P4.
Ok, so you say the AXP won part of a SiSoft benchmark. Lets split the SiSoft up into 5 benchmarks instead of as Frank did 3.
Now the P4 wins 3 and the AXP wins 2. The final result would be 11-10 to the P4.
No matter what way you count. It is intresting that the person writing the article claims that one CPU 'won' over the other. More correct would be that the AXP 1800+ is basically tied with the P4 2000.
/Knossos
SerArthurDayne
Distinguished
- Jan 21, 2001
- 344
- 0
- 18,780
Ya know....
ideally the review should have posted scores from a patched and unpatched version of the benchmark, while keeping the unpatched version as the official score. This would allow people to see what kind of performance they can expect now without the patch as well as a REAL demonstration of the capabilities of the AMD processor. Seems simple enough to me, and I believe others have done just that...
"Laziness is a talent to be cultivated like any other" - Walter Slovotsky
ideally the review should have posted scores from a patched and unpatched version of the benchmark, while keeping the unpatched version as the official score. This would allow people to see what kind of performance they can expect now without the patch as well as a REAL demonstration of the capabilities of the AMD processor. Seems simple enough to me, and I believe others have done just that...
"Laziness is a talent to be cultivated like any other" - Walter Slovotsky
I just though I’d mention another area where identification of a processor’s ability to do something is very skewed. Although its gotten better in the past year, many graphic card drivers enable bits for functions they nowhere near support by fault of their software or hardware. I used to use vendor strings to identify what cards could do what because half the cards would crash if you tried to use a feature they claimed to have. I would definitely blame the graphic card industry for causing mistrust in any processors ability to accomplish some task based on a flag.
Apparently some of you aren't getting my point...
<font color=red>THE PENTIUM 4 DID NOT WIN ONE BENCHMARK WHERE IT WAS ON THE TOP OF THE GRAPH</font color=red>
Better?
<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
<font color=red>THE PENTIUM 4 DID NOT WIN ONE BENCHMARK WHERE IT WAS ON THE TOP OF THE GRAPH</font color=red>
Better?
<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
mr_gobbledegook
Distinguished
- Sep 3, 2001
- 468
- 0
- 18,780
I agree with you.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
charliec2uk
Distinguished
- Jul 26, 2001
- 249
- 0
- 18,680
Yeah, I agree, Intel has a processor in hand if you like.
Woke up in Borneo one day, stuck in the pouch of a big marsupial, but the boat was made of marsipan!
Woke up in Borneo one day, stuck in the pouch of a big marsupial, but the boat was made of marsipan!
TRENDING THREADS
-
Question No POST on new AM5 build - - - and the CPU & DRAM lights are on ?
- Started by Uknownflowet
- Replies: 13
-
-
-
-
-
Facebook
Twitter
Instagram