AMD, VIA and Nvidia Quit BAPCO over Sysmark 2012

[STaRGaZeR]

Slower but smoother

[NEOAethyr]

I don't know anyone that uses pcmark.
I've used it in the past a few times, long long ago, but it never seemed to really have any relevant use, that's what sandra and everest/aida is for...

I did not know they scewed results baised of the device id's, I do know that via has a pci reg back door on many of there parts which allow you to change it.

I wonder if 3dmark does the same...

I might run a 03 pass once in a great while, but I don't bother with the newer ver's.
What erk's me is that even when you get 120fps in some of these tests, take 03 test3? for ex. (ogres and chick with a sword), the thing runs in slow mo.
Like they purposely made it run half the speed.

Look at 05 as another ex., the graphics have never been anything to write home about.
Yet the performance is so shiz, it's like wth...
Bad programming comes to mind, but these days I lean more towards that it was done on purpose.

As for superpi scewing results.
I have my doubts about that one.
Though I don't have a decent via board to try that on anymore, well that's only for the chipset, ... the cpu, hmm...
A simple re-programming of the cpu-id (it's a msr, real easy to change) would doit foir some programs but others "might" know better, I "think", I'm not actually sure...

I can test this cpu-id theory if you guys want.
I already have code that I made for the nf2 to redo the cpu-id (had a buggy bios where it would sometimes fail to program it correctly).
I would just need a cpu-id string for a intel cpu, a core i7 of some sort, but it would definitely need to be 100% exact before I'de try it.
This could turn out to be an interesting mod if it actually effects programs we use (I don't really care about pcmark to be honest).

One interesting mod back in the day, the amd socket a mobile xp's were infact mp cpu's.
If you fixed up the cpu-id, it would enable the mp instructions.
Not that it would do you any good with a single socket but it was still pretty neat, had a l33t factor toit.


Anyways if you guys want to play, put up some id strings of the intel cpu's that are worth while, the top of the line sandy or i7.
I rather not call my cpu an i3 lol , if you know what I mean.
If we're gonna doit, might as well doit right .

Edit:
What's thee best corei7 or sandy out right now?
If I knew that I could look for screens for the cpu-id and it would be all done.

[-Boris-]

Slower but smoother

Yes, nothing weird with that. Just as GTX 480 is slower than a HD 5990, it is still smoother in many scenarios. P4 was considered smoother in windows since it had HT wich helped a bit. While Athlon XP and singlecore Athlon 64 could be stalled by bad code. Mostly unnoticeable but could be seen sometimes. The same way Phenom and Phenom II was said to give a smoother feeling than C2Q in games since Core 2 had an really old bus which could be bottlenecked from time to time, even if the C2Q had a higher average framerate. I never heard anyone claim that an i5 or i7 would be less smooth than Phenom II.

[NEOAethyr]

Oh I forgot to post this.
I think p4's were alright as long as they were clocked at 2.4ghz or higher.
The same could be said for athlon's but in a diff way.

P4's just sucked though at lower speeds, I've had 900mhz p3's that beat out p4's 1.8ghz's with ease, when it came to playing media.
A p4 2.4ghz will play divx, heck it'll even do a 720p h.264.
But slower then that, won't even play a darned divx file without a few ms of lag.

Athlons on the other hand, 1.4ghz is fine, it'll play divx.
To play some games though, 2.4ghz is a sweet spot, you didn't need it, but it helped.

[kl0012]

http://www.vanshardware.com/reviews/...hlonXP2600.htm

Wonder why the guys at the OEMs claimed Intel to be faster during the Athlon - Athlon 64 eras.

Well, I find it actually funny - about 10 years ago when AMD had faster processors and BARCo was biased toward Intel, AMD decided to join BARCo in order to correct BARCo's behavior. Since then no one have been complaining about Sysmark2007 (as I remember changing CPUID affected PCMark05 and not Sysmark). Now I wonder what was the AMD's reason to leave BARCo (if it indeed true). I guess every chip firm will only love benchmark where its product has superiority.
BTW, here is an interesting article about changing CPUID.
http://ixbtlabs.com/articles3/cpu/vi...d-fake-p1.html

[crash5s]

You are not an AMD fanboy then, :P Yes AMD fanboys do claim that Phenom II is faster than i7 in the applications *they* run and Intel has just paid reviewers and application creators for unfair advantage in the scores.....

I've never really see anybody make a straight up apples to apples comparison. I've seen some people claim that in some cases the larger core count in some AMD CPUs helps in some applications. But I haven't seen anybody make the argument that AMD is faster per core or even per clock.

As for the "smoother" issue, there was some truth to that but it's largely horse crap as well. I got higher FPS consistently on a Q9650 than on a 940 BE, but lower FPS were also, well, lower. The 940 never got as many FPS but it also never plummeted, there wasn't as much fluctuation. However I always pinned the blame on the fact that nvidias 775 chipsets were problematic and finicky to say the least. I haven't had that problem with 1366.

[NEOAethyr]

Well, I find it actually funny - about 10 years ago when AMD had faster processors and BARCo was biased toward Intel, AMD decided to join BARCo in order to correct BARCo's behavior. Since then no one have been complaining about Sysmark2007 (as I remember changing CPUID affected PCMark05 and not Sysmark). Now I wonder what was the AMD's reason to leave BARCo (if it indeed true). I guess every chip firm will only love benchmark where its product has superiority.
BTW, here is an interesting article about changing CPUID.
http://ixbtlabs.com/articles3/cpu/vi...d-fake-p1.html

Very interesting link you posted .
There could very well be an advantage in spoofing your cpu-id as an intel.
Now I just gotta look around for cpu-z shots of the top of the line intel's.
I'll look this up tonight for sure so I can get testing on it.

It would be very interesting if it was 1 sec faster or so on superpi and a bit faster on mem bandwith in aida64...
I now understand now that it is "potentially" possible.

Edit:
I think the latest i7 is the 990x extreme.
The exact strings, I believe are:
"Intel(R) Core(TM) i7 CPU X 990 @ 3.47Ghz"
"Intel(R) Core(TM) i7 CPU X 990 @ 3.47Ghz (ES)"

I could be off by a few spaces here and there though.
I'm gonna start work on a program and try it out to see if I can match up the name to the cpu-z shots on the net, to make sure I doit correctly, I don't want to miss a blank space or anything like that.

One more edit:



Obviously I screwed up somehow .
Current code:

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; CPUID
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; Notes
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;"Mobile AMD Athlon(tm) MP processor", 0x00

;???
;"Intel(R) Core(TM) i7 CPU  X 990 @ 3.47Ghz", 0x00
;"Intel(R) Core(TM) i7 CPU  X 990 @ 3.47Ghz (ES)", 0x00

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; Entry Ptr
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

org 0x100

jmp Init

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; Defines
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; Functions
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;"Intel(R) Core(TM) i7 CPU  X 990 @ 3.47Ghz (ES)", 0x0000

Fix_CPU_Name_String:
mov ecx, 0xC0010030	;MSR Address
call Clear_CPU_Name_String
mov eax, "Inte"		;MSR Data
mov edx, "l(R)"		;MSR Data
mov ecx, 0xC0010030	;MSR Address
wrmsr
mov eax, " Cor"		;MSR Data
mov edx, "e(TM"		;MSR Data
mov ecx, 0xC0010031	;MSR Address
wrmsr
mov eax, ") i7"		;MSR Data
mov edx, " CPU"		;MSR Data
mov ecx, 0xC0010032	;MSR Address
wrmsr
mov eax, "  X "		;MSR Data
mov edx, "990 "		;MSR Data
mov ecx, 0xC0010033	;MSR Address
wrmsr
mov eax, "@ 3."		;MSR Data
mov edx, "47Gh"		;MSR Data
mov ecx, 0xC0010034	;MSR Address
wrmsr
mov eax, "z (E"		;MSR Data
mov edx, 0x00000000	;MSR Data
mov dx, "S)"		;MSR Data
mov ecx, 0xC0010035	;MSR Address
wrmsr
clc
ret

Clear_CPU_Name_String:
mov eax, 0x00000000	;MSR Data
mov edx, 0x00000000	;MSR Data
wrmsr
inc cl
cmp cl, 0x36
je Ret_Function
jmp Clear_CPU_Name_String

Ret_Function:
ret

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; Init
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

Init:

call Fix_CPU_Name_String

jmp Exit

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; Exit
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

Exit:

ret

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

There must be a few more msr's that need modding...

[Lanek]

@NEOAethyr: im not sure why you go about PCmark and Futuremark benchmarks. when the news is about Sysmark, or i have miss something and you had wanted make a relation between scores on them and Sysmark ?

[flippin_waffles]

Well, I find it actually funny - about 10 years ago when AMD had faster processors and BARCo was biased toward Intel, AMD decided to join BARCo in order to correct BARCo's behavior. Since then no one have been complaining about Sysmark2007 (as I remember changing CPUID affected PCMark05 and not Sysmark). Now I wonder what was the AMD's reason to leave BARCo (if it indeed true). I guess every chip firm will only love benchmark where its product has superiority.
BTW, here is an interesting article about changing CPUID.
http://ixbtlabs.com/articles3/cpu/vi...d-fake-p1.html

How can you even try to spin a defence, it's futile. It is well known that intel has BAPCo in their pocket, and Bing or Google will likely produce a Million hits of evidence. Obviously you aren't going to find an intel or BAPCo confession, but what you will find is proof far beyond a reasonable doubt.

From AMD's blog:

....AMD has a long history of supporting open standards; if you have any doubt just look at our support for OpenCL. And this support extends to active involvement with open industry consortia that likewise promote open standards. The beauty of open standards is that they are just that – open. Open to analysis, open to improvement and open to criticism.

AMD has for some time been a member of BAPCo, an industry organization that promotes, among other things, a benchmark known as SYSmark. In the past year or so AMD, with openness and transparency, has tried to explain why we believe this benchmark is misleading with respect to today’s commonplace applications − about a year ago I published a blog designed to explore this. If you work for a company that believes in transparency and integrity – and I do – then you have to take a stand and speak up when something is wrong.

BAPCo’s response to this blog was a threat to expel AMD from the consortium....

...AMD decided to do what we believed was the right thing for the industry and our customers, so we continued to work within BAPCo to try to get the next-generation benchmark, SYSmark12 (“SM2012”), right. Our hope was to effect change so that it would be open, transparent and processor-neutral. We got workloads included that represent the things you and I actually do in a day (instead of 35,000 line spreadsheets!).

But the question remained: what weighting would BAPCo ultimately give to the real-world workloads − since it is this weighting that defines the actual benchmark scores.

Unfortunately, our good intentions were met with an outcome that we believe does a disservice to the industry and our customers. We weren’t able to effect positive change within BAPCo, and the resulting benchmark continues to distort workload performance and offers even less transparency to end users. Once again, BAPCo chose to ignore the opportunity to promote openness and transparency....

...So how can AMD stay in BAPCo? Simply put, we can’t. We have resigned from BAPCo and asked that our name and logo be removed from marketing materials promoting SM2012.

Now I hear some of you asking, “Isn’t this really just about the long-running antagonism between AMD and your competitor?”

No, it’s not....

http://blogs.amd.com/nigel-dessau/20...1/#comment-603

[NEOAethyr]

@NEOAethyr: im not sure why you go about PCmark and Futuremark benchmarks. when the news is about Sysmark, or i have miss something and you had wanted make a relation between scores on them and Sysmark ?

Oh my bad.
Thought pcmark was sysmark, names were similar.
Lol ^^, now I don't know which is which lol, not that it really matters...

[flippin_waffles]

Imagine if automakers got together and started measuring the gas mileage of new cars with a cool test of their own making—one in which the cars were rolling downhill with their engines idling. Suddenly you'd have some pretty amazing claims: Why, that three-ton SUV gets 300 miles per gallon! This subcompact gets 500! In tiny print at the bottom of the window sticker you'd find a disclaimer saying that, well, um, you know, your mileage may vary

"There's only three endings to this story," says Patrick Moorhead, a marketing vice president at AMD. "Either the industry regulates itself, or the FTC steps in and regulates us, or we get hit with a class-action lawsuit. I suggest the industry go with the first option."

AMD is recommending computer makers adopt a new way of measuring battery life, using two states: "active time" and "resting time," similar to the way cell-phone makers describe the "talk time" and "standby time" of a phone. A Dell executive says that approach makes sense, and that the company is considering providing customers with information beyond the MM07 scores. "Customers expect the advertised battery life to reflect the way they really use the product," says Ketan Pandya, head of AMD-based products at Dell.

AMD also points out that the president of BAPCo happens to be the head of performance benchmarking at Intel.

http://www.newsweek.com/2009/06/18/h...-and-type.html


From 2009:

http://video.nytimes.com/video/2009/...%20rant&st=cse

[Manicdan]

^caught with pants down comes to mind

[Hornet331]

Imagine if automakers got together and started measuring the gas mileage of new cars with a cool test of their own making—one in which the cars were rolling downhill with their engines idling. Suddenly you'd have some pretty amazing claims: Why, that three-ton SUV gets 300 miles per gallon! This subcompact gets 500! In tiny print at the bottom of the window sticker you'd find a disclaimer saying that, well, um, you know, your mileage may vary

Oh god I loled at that... I think whoever wrote that whould do a reality check... its exactly like that already.

Regardless where you live in this world and what the code is called (NEDC, 10/15-Mode, FTP75), they are all faaaar of reality. The diffence is only that tihs regulations are done by goverments... so calling for the FTC wont change one step in that regard...

Also its quite funny to look at that statement and what amd is now suggesting to get there "AMD AllDay™ Power" (which is 8h+ at idle, which also doesn't refelct rl at all).

[Shadov]

Oh god I loled at that... I think whoever wrote that whould do a reality check... its exactly like that already.

Regardless where you live in this world and what the code is called (NEDC, 10/15-Mode, FTP75), they are all faaaar of reality. The diffence is only that tihs regulations are done by goverments... so calling for the FTC wont change one step in that regard...

Also its quite funny to look at that statement and what amd is now suggesting to get there "AMD AllDay™ Power" (which is 8h+ at idle, which also doesn't refelct rl at all).

Are you suggesting they should offer "real numbers" while the competitor is promoting their idle work time? Talk about a marketing suicide.

It goes for the whole industry or no one.

[STaRGaZeR]

Also its quite funny to look at that statement and what amd is now suggesting to get there "AMD AllDay™ Power" (which is 8h+ at idle, which also doesn't refelct rl at all).

Dude, days in AMDland have 8 hours!

[-Boris-]

Dude, days in AMDland have 8 hours!

Yeah. That's because they are talking work days. How long is the night, evening and morning in your world btw?

[Shadov]

Slower but smoother

Dude, days in AMDland have 8 hours!

An Intel Fanboy like you that adds nothing valuable to the discussion should be banned from all threads that concern X86 and fair practises.

Just FYI: Most people do work 8 hours a day.

[-Boris-]

As for the "smoother" issue, there was some truth to that but it's largely horse crap as well. I got higher FPS consistently on a Q9650 than on a 940 BE, but lower FPS were also, well, lower. The 940 never got as many FPS but it also never plummeted, there wasn't as much fluctuation. However I always pinned the blame on the fact that nvidias 775 chipsets were problematic and finicky to say the least. I haven't had that problem with 1366.

That's typical narrow FSB behavior, you won't see that on an i7. I saw it on Q6600, sometimes some slight stutters, my Phenom 9850 didn't have that problem, the difference was noticable in GTA 4 among other games that had lots of data loaded all the time. On the other hand the Phenom was slower overall so if price and upgradeability wasn't an issue I'd go for the C2Q and try to overclock the FSB far enough for the problem to go away. Still happy with my choice though, Phenom II on the same motherboard is worth it.

[crash5s]

That's typical narrow FSB behavior, you won't see that on an i7. I saw it on Q6600, sometimes some slight stutters, my Phenom 9850 didn't have that problem, the difference was noticable in GTA 4 among other games that had lots of data loaded all the time. On the other hand the Phenom was slower overall so if price and upgradeability wasn't an issue I'd go for the C2Q and try to overclock the FSB far enough for the problem to go away. Still happy with my choice though, Phenom II on the same motherboard is worth it.

Hence why I said I didn't see it on 1366. And I didn't see the problem fixed by overclocking 775 CPUs at all. But my 680i, 780i, and 750i systems that I used the most all had issues left and right. So much so that I simply swapped GPUs and mobo's and went to Phenom II 780a configurations, and it went away. I haven't seen it on 1366 though.

[Hornet331]

Are you suggesting they should offer "real numbers" while the competitor is promoting their idle work time? Talk about a marketing suicide.

It goes for the whole industry or no one.

Its not about the all day thing only beeing 8h (or more), but the thing that it is measured in idle state.... they claim Mobile Mark is a useless benchmark tool not refecting realworld and then they suggested this as a metric?
Come on srsly you can't defend that...

[Manicdan]

marketing does what it wants, its still a higher number than the competition offers when things are equal.

if cars are compared at 45mph for mpg, then its still fair, but shouldnt be a measure of highway millage.

[Manicdan]

double post

[STaRGaZeR]

Just FYI: Most people do work 8 hours a day.

Just FYI: A Real Life Day™ has 24 hours.

Well, maybe a bit more, as seen here: http://upload.wikimedia.org/wikipedi...om_SI_day_.svg

[tifosi]

Just FYI: A Real Life Day™ has 24 hours.

Well, maybe a bit more, as seen here: http://upload.wikimedia.org/wikipedi...om_SI_day_.svg

Intel doesn't really advertise themselves as SSE5 killers, now do they? They also do not advertise the fact that they threatened OEM's to choke AMD. They also don't advertise that they cripple AMD processors, by rigging their compilers. Need more examples as to how marketing works?

[Hornet331]

Oh please not the crap "compiler conspiracy" again...

I just leave this here:
Jf-Compiler end discussion.jpg