5.21.2007

AMD`s Hype-per-Watt Leadership

From Digitimes:
“While the Puma platform looks set to include technological features which will improve performance and battery-life for notebooks; according to AMD hardware based on the new platform is not likely to hit the market until mid-2008, i.e. one year from now. While it is nice to get a preview of upcoming products and technology, with AMD's recent track record of confirmed (R600) and speculated (Barcelona) product delays, and poorer than expected performance (HD 2900 XT), it is questionable whether, when (or dare we even say if) it is launched, Puma will meet the hype AMD is currently trying to generate though these early announcements”.

A tide of frustration and scepticism is growing against AMD. This company is getting too good in making PowerPoint presentations while becoming worst in releasing products on time with the promised performance. AMD’s “proof to claims” ratio for the last 2 years is either a very low decimal number or zero. AMD may have achieved overall success is the last few years but by the way it has been behaving, it appears like it hasn’t put down the party hat. The exact opposite can be said about Intel with its product pull-ins. As if Intel is trying to compete with itself. We’ve seen the public benchmark of Penryn and soon, Harpertown – both 2008 releases while AMD’s 2H’07 product, one that they dreadfully need to survive is nowhere to be seen. With the unflattering product reviews of the 65nm Athlon X2, QuadFX, R600 & HD2900XT, it is easy to draw a conclusion as to why AMD is afraid to show any real benchmarks.

As AMD is desperately trying to come off as the innovator in the industry, they had no choice but to pre-empt Intel’s Santa Rosa product release with their very own “me-too!-Centrino-copied” mobile platform. The big difference is Santa Rosa was announced sitting inside an OEM laptop, Puma was announced on a PowerPoint slide. Luckily for AMD, they won’t hear Intel shooting its mouth about someone copying their strategy. It’s not because AMD copying Intel is status quo, it’s just that everyone knows being a year behind is embarrassing enough.

13 comments:

Axel said...

What's even worse is that Puma will apparently be based on K8, not K10:

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2992

http://arstechnica.com/news.ars/post/20070520-amd-sics-puma-on-intels-santa-rosa.html

So Puma hasn't a chance of competing with Santa Rosa in performance or performance-per-watt. This also means that AMD must intend to ramp K10 rather slowly, since there won't be enough K10 parts to go into Puma. That's probably a consequence of postponing the conversion of Fab30 to 65nm.

Heat said...

Another great article Doctor Roborat. Your blog is the guiding light in the sea of AMD fud. If only amd fanbois could put those power point slides in their CPU sockets and make them work.......

Roborat, Ph.D said...

axel said: What's even worse is that Puma will apparently be based on K8, not K10...

and most likely be at 65nm while intel introduces 45nm mobile quadcore. AMD is just too far behind on mobile.

Anonymous said...

"AMD is just too far behind on mobile."

While this is true - this is one area where they have been continuously eating market share from Intel (not sure about Q1), and if one is interested in seeing a healthy AMD (I'm not sure if I really care or not) - they are far better off working on mobile than desktop.

With ATI I think AMD has a chance to compete better in this area as CPU sales are far less dependent on CPU performance than most people realize. Cost, power, features, mechanical design of the notebook are nearly as important as the sheer processing power of the CPU. That said, it's not like AMD has advantages in those other or a clear roadmap to catch/pass Intel.

It is rather curious that AMD would base their next mobile product on K8 - what does that tell you about the K10 architecture? If it's 40% better than K8 they could have made a mobile chip 40% better at same power or kept same performance and gotten much better power #'s, no? As they are saying mid-2008, this couldn't be about the K10 schedule (unless when they started this project they were worried that K10 might be a year late!?!?), so I wonder why they chose this path?

Anonymous said...

CNET has an article today

http://news.com.com/AMD+shows+sunny+side+of+Barcelona/2100-1006_3-
6185526.html?tag=nefd.top

Exactly what is sunny about what was shown is a mystery to me.

Anonymous said...

Rumors about AMD winning over Toshiba as a mobile customer are starting to gather credibility.

Most of the growth is in mobile so it makes sense for AMD to try and capture a bigger slice of that pie.

Unfortunately as it has been pointed out, the Puma platform is a tweaked K8. Why AMD announced this one year ahead of availability is a mystery to most.

Anonymous said...

LOL. Here's a gem from the Cnet article:

'AMD did not test Barcelona's performance against one of Intel's quad-core Clovertown Xeon processors, but Allen said Barcelona "will be the highest-performing x86 chip out there. It will blow away Clovertown."'

Steady now lads: perhaps it may blow away Clovertown 'at the same frequency' and in one, at most two cherrypicked scenarios.

Fancy how Allen does not seem to have a clue (nor the reporter had the cojones) to ask about getting more specific about the rhetoric.

AMD BK Q2 2008

Anonymous said...

On the sunny side article (thanks for the link)- am I reading it wrong?

A next Gen quad core is about 2X faster than a current Gen dual core?

If this is true is this not a TERRIBLE result for AMD. Scaling a K8 from 2 to 4 cores with AMD's interconnect technology should almost double the performance to begin with, no?

Either the K10 clock was significantly lower than the dual core K8 they were comparing it to or the architecture isn't yielding much improvement...

Any thoughts on why AMD is dribbling out the Benchmanrks (if you can call them that)?

Does anyone else think there are still significant issues, whether it be clockspeed, power, erratic performance on some SW, or not so great performance gains? I'm certainly not buying the "we don't want Intel to copy the chip" BS that AMD is spouting...

Roborat, Ph.D said...

Does anyone else think there are still significant issues, whether it be clockspeed, power, erratic performance on some SW, or not so great performance gains? I'm certainly not buying the "we don't want Intel to copy the chip" BS that AMD is spouting...

Lower speed Barcelona at launch points to process issues. It looks like yields aren't very good, thats for sure.

Anonymous said...

"Lower speed Barcelona at launch points to process issues."

No, it's because the design was not optimized for the 65nm process.... Oh, wait I was accidently copy/pasting the 65nm K8 speed excuse.

On a serious note - 4 core clock speed degrades to lowest common denominator , 3 cores at2.6GHz, 1 core @2.3Ghz = 2.3Ghz part. Gotta love the "native" approach? I've mentioned numerous times the top bins will be tough because of this (and obviously the same thing happens with power and yield).

It may be more elegant and technically better than Intel's MCM approach - but elegance doesn't sell product.

Unknown said...

It has already been reported that both K8 and alledged K10 were running at the same frequency during the test.

It is known for POV Ray to mask any scaling inefficiency. i.e, 2 cores to 4 cores results in 2x performance, even with Intel's MCM approach.

That being said, one can conclude that K10's IPC is no better than K8's IPC under POV Ray. The true benefit of K10 (native quad core) will only be shown when running specific server applications. I'm wondering why AMD decided to use POV-Ray, which will not highlight K10's true benfits?

Assuming that K10's performance ins vastly superior than K8, then the only reason AMD uses POV-Ray is to produce fake image of K10, that Intel will not have enough time to devise a plan and catch up.
Even though this explanation doesn't really make sense either.

However, only time will tell as we get nearer to K10's official launch date.

Ho Ho said...

It is also interesting that for some reason Barcelona double speed SSE doesn't show its power in PovRay.

Anonymous said...

After the stupid hype they created they tried to lower expectations nonchalantly to something more reasonable, but it blew up in their faces. The problem is that K10 brings about 10%-15% IPC improvements over K8 on average. This is not enough to bridge the 20%-40% IPC gap with Core2 and they bin very little over 2.0GHz on top of that. The economics of the “naïve quad core” doesn’t add up.