VOGONS

Common searches


First post, by sliderider

User metadata
Rank l33t++
Rank
l33t++

http://news.cnet.com/8301-13924_3-20073556-64 … ing-intel-bias/

Advanced Micro Devices has quit a PC industry consortium, implying the integrity of a widely used benchmark is biased toward Intel chips.

In a blog Wednesday, an AMD executive provided a long explanation about why AMD has quit the BAPCo industry consortium, which develops and distributes the SYSmark benchmark.

"Customers need clear and reliable measurements to understand the expected performance and value of their systems," Nigel Dessau, senior vice president and chief marketing officer at AMD, said in a statement. "AMD does not believe SM2012 (SYSMark 2012) achieves this objective. Hence AMD cannot endorse or support SM2012 or remain part of the BAPCo consortium."

Graphics chip supplier Nvidia has also quit the group. The company confirmed to CNET that it has quit but declined to comment further.

AMD's argument revolves around the lack of emphasis on what is called general-purpose computing on graphics processing units, or GPGPU. This is an evolving computing paradigm that attaches increasing importance to the GPU for accelerating common tasks such as encoding/decoding of video and audio and Web browsing.

Dessau continued. "SM2012 doesn't represent the evolution of computer processing...SM2012 focuses only on the serial processing performance of the CPU, and virtually ignores the parallel processing performance of the GPU." Serial processing refers to the computation done on the main central processing unit, or CPU.

AMD's newest "Llano" chips allocate roughly equal real estate to the CPU and GPU, a break from the past when AMD processors were mostly composed of circuits devoted to the CPU. Intel still allocates more real estate to the CPU than the GPU component.

BAPCo responded to AMD's accusations. "Each member in BAPCo gets one vote on any proposals made by member companies. AMD voted in support of over 80% of the SYSmark 2012 development milestones, and were supported by BAPCo in 100% of the SYSmark 2012 proposals they put forward to the consortium...BAPCo never threatened AMD with expulsion from the consortium."

And AMD is certainly not the only company that provides input to the group. Members with equal standing include Microsoft, Dell, Hewlett-Packard, Lenovo, Samsung, Sony, and Toshiba. And, of course, Intel.

"It is a democratic process," Intel spokesman Dan Snyder said. "Intel believes that no single benchmark alone is a definitive measure of a PC's performance. Intel encourages all technical reviewers to use multiple benchmarks and applications--synthetic and real world--in determining the performance of a PC," he said in an e-mail.

Moreover, Intel is moving increasingly in the same direction of AMD and Nvidia. With its newest Sandy Bridge generation, Intel has, for the first time in a mainstream chip, integrated graphics processing silicon onto the main processor. And Intel will continue to beef up graphics processing on its upcoming Ivy Bridge chips, due late this year.

Reply 1 of 45, by sliderider

User metadata
Rank l33t++
Rank
l33t++

So does SYSMark have any legitimacy left after the second largest CPU maker and both of the two biggest GPU makers have withdrawn from the program? I can't see how it does if Intel the only major player left and will be working feverishly to make it's own processors and graphics chips look better.

Reply 3 of 45, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

AMD has a massive focus on their fusion on-die GPUs and multithreading. Anandtech always shows SysMark to highlight the single thread performance lead of Intel.

But wouldn't leaving the group just give even more influence to Intel?

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 6 of 45, by sliderider

User metadata
Rank l33t++
Rank
l33t++
Dant wrote:

So why then would Nvidia quit when they're not even in the same field?

GPU computing. nVidia does it really well, Intel doesn't. nVidia and AMD both feel GPU computing is going to be very important and Intel doesn't so naturally Intel get their way.

Last edited by sliderider on 2011-06-23, 22:11. Edited 1 time in total.

Reply 7 of 45, by sliderider

User metadata
Rank l33t++
Rank
l33t++
Mau1wurf1977 wrote:

AMD has a massive focus on their fusion on-die GPUs and multithreading. Anandtech always shows SysMark to highlight the single thread performance lead of Intel.

But wouldn't leaving the group just give even more influence to Intel?

If Intel is the only one left in the group it makes SYSMark less legitimate because everyone will know Intel is manipulating the results.

Reply 8 of 45, by Barry_Purplelips

User metadata
Rank Member
Rank
Member

I think the quitters have secretly teamed up to delegitimize bapco as an attempt to stop Intel from kicking there asses.

See how easy it is to come up with a conspiracy theory these days? really, we weren't there, we can't possibly know.

Reply 9 of 45, by DonutKing

User metadata
Rank Oldbie
Rank
Oldbie

I remember reading about this 10 years ago.... http://www.anandtech.com/show/835/5

IIRC there was also bias claims against Sisoft Sandra as it greatly favoured SSE2 which AMD didn't support at the time

If you are squeamish, don't prod the beach rubble.

Reply 10 of 45, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

You might also say that AMD is now only looking at GPU benchmarks. When they talked about Llano and Trinity, they do talk about performance gain, but when you read the fine print it's mostly about the GPU.

This whole AMD Compute / Fusion doesn't excite me one bit.

Last week a friend asked me about advice for a PC, I looked at the prices and recommended him a i5 2500, 8 GB DDR3 and a H61 board. The whole PC with 1TB drive, DVD burner, case and CPU cost only a little over AUD 500.

The CPU was ~ 200 bucks and that's Intel’s second fastest CPU money can buy. I don't see the point in saving money on hardware anymore. It's cheap as chips.

Back in the day when CPUs set you back 600+ I can understand why you would go with AMD and save some money. But now?

I love AMD, but Intel will be at 28nm by the time Bulldozer is out and all the bugs have been sorted out. And Trinity will have another socket again as well...

Reply 11 of 45, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

Gah, I made a long post yesterday but it got lost 🙁

Vogons wrote:

I think the quitters have secretly teamed up to delegitimize bapco as an attempt to stop Intel from kicking there asses.

See how easy it is to come up with a conspiracy theory these days? really, we weren't there, we can't possibly know.

Personally I don't like Intel's socket hopping thing ever since they left the LGA775 platform.
I like AMD for their price/performance and the flexible hardware that comes with it.

My lost post was 10 times better but eh...early mornings 😁

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 12 of 45, by DonutKing

User metadata
Rank Oldbie
Rank
Oldbie

Intel's been hopping sockets for ages. Socket 4, 5, 7... Slot 1, socket 370, Tualatin 370... 423, 478, LGA775... sure you could say that 7 was backwards compatible with 5, and there were 370 slockets and Tualatin adapters, but still...

If you are squeamish, don't prod the beach rubble.

Reply 13 of 45, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

My point is that bulldozer will be a very short lived product with trinity really what they want to push. I mean they pushed trinity even at that recent trade show. And trinity won't be out for like another year!

And why wait for it, when you can buy a 32nm i5 2500 for ~ AUD 200. It will be AMDs first 32nm CPU, it's already late because of issues and just like with Phenom it will take them several revisions to get going. Buy that time Intel will have moved to 28nm.

Reply 14 of 45, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
DonutKing wrote:

Intel's been hopping sockets for ages. Socket 4, 5, 7... Slot 1, socket 370, Tualatin 370... 423, 478, LGA775... sure you could say that 7 was backwards compatible with 5, and there were 370 slockets and Tualatin adapters, but still...

What I meant with socket hopping is when they introduce a new chip, the chip is released in a new socket which is incompatible with the older socket and then abandoning the new socket for an even newer socket when another "new" chip is introduced.

It's about the time a socket is "alive". Intel seems to like jumping sockets within 2 years or so, meaning theres not many chips to choose from.
Socket 7, A, 775 and AM"X" have existed for a longer time and have many different chips available for them.

Socket hopping means a computer that's about 2 1/2 years old and breaks down will be more costly to repair as it will be more difficult to find replacement parts.

The last part is subjective though, but it matters to me.
The only socket hopping AMD ever did was when they first introduced their A64, but Intel just loves to make changes and leave people with an older Intel platform without an upgrade path (Tualatin, s423 and 1156 come to mind).

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 15 of 45, by Dominus

User metadata
Rank DOSBox Moderator
Rank
DOSBox Moderator

Quoting whole articles should be avoided. Wherever that was posted probably lives on ad income. You should only quote a small part and link to the original article. Full quote only makes this site look bad and vulnerable to copyright infringement claims...
People should be glad I'm not a mod 😉

Windows 3.1x guide for DOSBox
60 seconds guide to DOSBox
DOSBox SVN snapshot for macOS (10.4-11.x ppc/intel 32/64bit) notarized for gatekeeper

Reply 17 of 45, by DonutKing

User metadata
Rank Oldbie
Rank
Oldbie

What I meant with socket hopping is when they introduce a new chip, the chip is released in a new socket which is incompatible with the older socket and then abandoning the new socket for an even newer socket when another "new" chip is introduced.

Yes thats what I was getting at 😀

If you are squeamish, don't prod the beach rubble.

Reply 18 of 45, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
DonutKing wrote:

What I meant with socket hopping is when they introduce a new chip, the chip is released in a new socket which is incompatible with the older socket and then abandoning the new socket for an even newer socket when another "new" chip is introduced.

Yes thats what I was getting at 😀

That's not what I was getting at. The difference with socket hopping is to create a chip that is incompatible with an older socket by choice, not because it has to be.
Intel's Tualatin isn't incompatible with Coppermine boards because there is some technical reason behind it, but because Intel WANTED the Tualatin chip to not work with older s370 boards.
Thats what I meant with socket hopping, choosing to make a new chip incompatible with an older one to prevent people with an older platform from having an upgrade path.
The jump from Socket 5 to 7 was not socket hopping as it provided a real benefit and the new chips could still work in the older socket.
Also during the LGA775 time Intel was not hopping sockets as the older netburst chips would still work in Conroe boards.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 19 of 45, by sliderider

User metadata
Rank l33t++
Rank
l33t++
Dominus wrote:

Quoting whole articles should be avoided. Wherever that was posted probably lives on ad income. You should only quote a small part and link to the original article. Full quote only makes this site look bad and vulnerable to copyright infringement claims...
People should be glad I'm not a mod 😉

Some people don't like clicking links that send them to unknown places, which was why I quoted in addition to providing a link. Some people are also too lazy to click the link.