7.15.2007

AMD's 2.0Ghz Barrier

More than year ago everyone was talking about Barcelona's from AMD that would clock to 2.3 and even 2.66Ghz in 2007. However, these speeds have never been released. And, rather than asking why, not a single review site has mentioned it. The common pretense today seems to be that AMD never claimed that it would top 2.66Ghz in 2007 which, of course, it did.

It isn't hard at all to find evidence of AMD's intention to have K10's clocked as high as 2 or 2.6Ghz in 2007.

AMD promises 2.6Ghz Barcelona
Clearly, everyone (including AMD) expected Barcelona's faster than 2.66Ghz to be released in 2007 yet these were never released. The common excuse given by AMD proponents is that AMD didn't release faster chips as originally planned because, “It didn't have to.” However, there is good evidence that the reason had more to do with temperature and overheating than a warm and fuzzy feeling of being safely in the lead.

It is very rare to get proper numbers for thermal testing of AMD cpu's. Typically, overclocking is done with premium cooling and testing with the stock HSF never thermally stresses the CPU. For example, when Anandtech originally reviewed X6800, they used a massive Tuniq Tower. This is, of course, nothing at all like what is shipped in the vast majority of computer systems. A term that often gets tossed around is "on air". However the truth is that high end air coolers today are as good as liquid coolers used to be four or five years ago. Thus the term "on air" now means very little. Cooling your CPU "on air" with something the size of a transmission cooler is not much an accomplishment. But, if you could overclock "on air" with the stock HSF, that would be an accomplishment indeed. Unfortunately, review sites seem to want to do thermal testing with a stock HSF about as much as they would like to feed a tank full of hungry piranhas by hand.

It appears that we only got some halfway informative numbers from Anandtech by accident when they were reviewing a cooler instead of the processor itself. I've talked about this information in an earlier article but I'm going to revisit it along with other information since people still seem confused about Barcelona's thermal limitations.

In Anandtech's Case cooling in the second chart: CPU Temperature Under Load, we have some data. Notice that at the stock clock speed of 2.93Ghz with the stock HSF, X6800 is reading 56 C. Now, the article says, "The stress test simulates running a demanding contemporary game. The Far Cry River demo is looped for 30 minutes and the CPU temperature is captured at 4 second intervals". Unfortunately, Anandtech's assumption is a bit off since Far Cry does not really thermally stress the core. According to the Barcelona Temperature Guide:

AMD provides a test program, Thermal Analysis Tool (TAT), to simulate 100% Load. Some users may not be aware that Prime95, Orthos, Everest and assorted others, may simulate loads which are intermittent, or less than TAT. These are ideal for stress testing CPU, memory and system stability over time, but aren't designed for testing the limits of CPU cooling efficiency.

Orthos Priority 9 Small FFT’s simulates 88% of TAT ~ 5c lower.

So, if FFT is 5C lower then Far Cry would be less than that. The Guide also clearly states Tcase Load should not exceed ~ 60c with TAT, and 55c with Orthos.

But, at 56 C we have already exceeded 55 C. And, that would be if we were running FFT. Since we are only running Far Cry the real temperature is probably closer to 60 C at full load. Now, someone will probably claim that that is okay because you would never hit full thermal load under normal circumstances. Unfortunately, that is already figured in. As stated in the guide: 50c is safe. 50c Tcase is a safe and sustainable temperature. 55 or 60 C are not safe or sustainable temperatures. Even though the maximum spec is 60 C, 60c is hot.

I have seen others claim that 60 C was fine because it didn't exceed the maximum rating. However, this is not the way the rating works. 60 C is the maximum for TAT only because no other program will ever reach this stress level. So, it is clear that X6800 with stock HSF will indeed exceed the factory cooling specs at stock speed. This is why AMD has not released a 2Ghz quad core.

But . . . what about dual core? Even if the quad core processor is running too hot with stock HSF maybe the dual core isn't. As I've already mentioned, these types of tests seem to be avoided like the plague by regular review sites. So, we need something obscure. However, as luck would have it, we do have something out of the ordinary at Digit Life where X6800 (2.93 Ghz) dual core is reviewed. These conditions are very unusual because he tests in his un-air conditioned apartment. In Moscow apartments, unconditioned for several days, the standard daytime temperature was within +25—+30°C. In this case the environment temperature was +28°C. 28 C is 82 F. This would not be unusual for Indiana in the Summer either. But, he uses the stock HSF. And, let's see what happens:

it didn't even occur to us that new AMD Barcelona processors could spring such a surprise with throttling...) Yes! It was throttling!

The chart is excellent because it compares the stock HSF to a better cooling solution. We can see that none of the common benchmarks really thermally stress the cpu. The one that stressed it the most was the Solidworks CAD & CAE benchmark. The chart only shows an 11% drop because it is a total score. However, The results of the overheated processor are very low in two applications out of three: SolidWorks 2005 and Pro/ENGINEER Wildfire 2.0. This is proof positive that even with a regular application that dual core C2D is busting the thermal limits when running at 82 F ambient. The author says that there was no thermal throttling at 72 F ambient.

Now, let's look at that chart again. Notice that with the 3D Shooter Games (F.E.A.R, Half Life 2, Quake 4, Unreal Tournament 2004) the drop is only 3% versus the 11% we saw with CAD. This again casts doubt that Anandtech was doing anything thermally stressful with Far Cry.

So, AMD did not release a 2Ghz processor in 2007 because they couldn't. Such a processor would have exceeded the factory's own limits when running routine applications at ambient temperatures common in the Summer. However, we also know that AMD has steadily improved the thermal properties of its C2D chips with each revision. It is possible that one of the newer revisions of C2D would be capable of going over 2.66Ghz without busting the factory's thermal limits. After all, one would assume that simply lowering the TDP would help and we know that TDP has come down. However, I'm not entirely sure that 65nm is actually an improvement. For example, if a given 65nm chip has the same TDP as a given 90nm chip then logically the 65nm chip would concentrate the heat into about half the die area. This would seem to be more likely to create hot spots rather than less.

I'm quite certain that AMD will indeed get chips out that exceed 2.66Ghz. And, I'm pretty certain that if AMD doesn't do it in Q4 of this year that Q1 08 should be reasonable. It might happen with 65nm as many expect but I can't see any reason why it couldn't happen with 90nm in another revision or two if 65nm wasn't up to at first. In other words, AMD would move to 65nm anyway to reduce die size even if it didn't make any major strides in power draw at first. However, the talk has been that 65nm at AMD is much better with power draw than 90nm. I guess we need to take that with a grain of salt since it was claimed in January 2007 that AMD had K10 samples hitting 2.0Ghz. It would be nice to have some up to date comparison information with newer C2D's to see if the thermal limits have been improving over time (which they likely have). But, since the information we do have about stock cooling is nearly accidental that seems very unlikely. We may just have to wait until Q4 and see if any 2 or 2.0Ghz 65nm chips appear.

16 comments:

Roborat, Ph.D said...

ah, the magic of Replace using Ctrl+F.

You may start to think i'm just fooling around but I think it's good to remind everyone that there's a much bigger clock freqency issue that really needs to be answered.

It's quite obviously that Intel's latest process revisions focused on TDP and maximum yields for both dual core and quad cores. I don't see any financial benefits in increasing clock frequency at this point when the only obvious way to maximize revenue is to ship more quads.

Anonymous said...

wtf Scientia LOL

Anonymous said...

Speaking of scientia, aka "thg and forumz sucks"... Let met cite thg and forumz when it benefits me, like a selective data picking fanboy lol.

If he still thinks that Intel is shipping products that are overheating lol, please cite Intel's own documentation
http://processorfinder.intel.com/details.aspx?sSpec=SL9S5

Anonymous said...

Why even bother reading any of Scientia's articles on process technology? He doesn't know anything in this area, has a ridiculous understanding of APM, doesn't seem to understand the difference between THEORETICAL die size area scaling between nodes and actual scaling and has no clue about transistor technology. Other than that he has a solid foundation.

Part of the problem is with the relative dearth of good news for AMD, he has to re-hash his old blogs. How many articles is he going to write about the "3.0 GHz" barrier...from someone who would preach performance and not raw GHz is really the important # anyway.... or perhaps he would prefer the 3.6GHZ P4's to the 2.93 Core 2's....

When performance of one's preferred company doesn't measure up - the only thing you can do is pick one indicator and focus on it, inan attempt to distract and bury one's head in the sand, hoping things will get better.

Why has he not yet written an article about AMD's FAILURE TO RELEASE A 65nm ATHLON AT EVEN THE SAME UPPER BIN SPEED AS 90nm? This is a company that has "nearly matched" Intel on process technology, yet can't get 65nm yielding well on high end parts.

Oh wait, my bad, they are focusing on "energy efficient" parts per customer demand (except of course Athlon 6000's and FX's...which BTW remain 90nm for some odd reason...)

Anonymous said...

Sharikou is apparently also jumped aboard the process technology fellow wannabe bandwagon:

"Intel is still alive for one single reason: AMD's complex 65nm SOI process is not producing the clockspeeds as expected."

I'm confused - a while back Sharikook told us that AMD/IBM presented data that 65mn was 40% faster than 90nm...now he's saying AMD released the process when it was SLOWER? I though AMD only ramped when it had things together?

And on a humorous note - I love it when people associate more COMPLEX with more ADVANCED. In manufacturing it is far better for things to be simpler and this is where INTEL excels... they don't need a more expensive and complex SOI process to hit 65mn tehcnology targets. They don't need 4 stress technologies. They don't need immersion litho technology on 45mn to print the features required.

To the average person, it would sound like this would mean Intel is much behind in these areas in terms of process technology. What most people don't realize is if you can get the same raw technical performance (I'm talking Idsat's, Ion/Ioff ratios, etc...) as your competitor with a far simpler, cheaper process and extend/reuse older equipmenmt beyond what most people think is possible, THAT is a sign of the more ADVANCED MANUFACTURING...

Intel can do the same technology as AMD except they do it earlier, cheaper, and with a simpler process. This is why you cannot look at the start of the technology node of Intel and AMD and say that is the lead (~1 year); Intel is YEARS (emphasis on the "s") ahead to those who have some education in this area. AMD's 65nm process is at a point where Intel would not even have released it (they target 30% transistor improvement over previous technology) - it's not even clear at this point if AMD's 65nm transistor is even 1% better than 90nm....

APM is a classic example of the misinformation... Wow AMD can feed forward info and fine tune process steps to eke out performance. What does this mean? It means the process window is crap and needs to be tweaked to get high performance - this is great if you are handcrafting a few wafers, but what happens when you need MILLIONS? Would you rather have a process that is stable and does not need to be tweaked wafer to wafer or one that does? Which process is more likely to have issues ramping or issues with excursions?

Intel is a MANUFACTURING company with solid engineering, AMD is an ENGINEERING company which they think equates into good manufacturing. This seems to fit the same mold as IBM - great theoretical/research/engineering capabilities just don't ask them to mass produce anything. PS3 shortages anyone? SiLK anyone? But hey, they do have those patent #'s....

Anonymous said...

Intel 45nm will bring 0.5 of 65nm, 20% better performance and 30% less TDP. Based on trends, I think a good estimate for AMD 45nm is 0.7 of 65nm, 10% worse performance, 10% reduction in TDP.

Anonymous said...

Roborat, I think you have a typo:

"However, we also know that AMD has steadily improved the thermal properties of its C2D chips with each revision"

I think you meant to write Intel instead of AMD in above :)

Anonymous said...

Do you really mean to say 'C2D' in all of the places in your article, assuming 'C2D' is Core 2 Duo?

Anonymous said...

It is rather amazing how someone with so little actual knowledge of process technology can take some random data points like temperatures and use that to deduce that Intel can't break the "3.0GHz barrier".

Process technology is all about tradeoffs....

He has postulated in the past that this is an issue with the 65nm process (mind you with no ACTUAL process data, just backward extrapolation with temps and TDP's). The problem remains (in addition to the backward logic and using OBSERVATIONS to "CONCLUDE" facts), he doesn't know what the limitations of the 65mnm process are - Intel could be focusing on keeping a wider process window in order to get better yields, factory inventory (WIP) turns or bin splits; or they could be focusing on TDP's instead of raw speed (by adjusting things like Vt implants, gate oxide performance, Halos and top implants...)

His stating that there is a barrier is like saying 80% of criminals watched violent TV when they were young, thus violent TV during childhood must make people criminals... the problem is he lacks an actual foundation on what the limitation is for 3.0 GHz parts (if there is any) and thus chooses to speculate on what paints Intel in an unfavorable light.

Similarly Scientia is ignoring the business aspect of things...

Intel has positioned the quad core as the "extreme" part... this could be for several reasons - to hurt AMD's manufacturing capacity by forcing more of the overall market mix to quads, to leverage Intel's current economic and time to market advantages with MCM approach vs AMD's "native" approach.

Regardless of why, from a marketing perspective alone it would be difficult for Intel to launch higher speed (and likely lower priced) dual core parts which for most applications will be faster than quad core 'extreme' parts that might be a bin or 2 down in speed. Until SW "catches" up, a slightly faster dual core dekstop part will generally outperform a slightly slower quad core part (with obvious exception of SW which is already designed to take advantage of multicore like encoding). This makes it difficult to call the quad core the "extreme" part.

Oh and his understanding of thermodynamics and where/how various chip temp impacts comes from (in terms of his die size, cache and process node theories) is UTTERLY LAUGHABLE. But for those with a background you already can see this.

Anonymous said...

http://www.sharkyforums.com/showthread.php?t=165292
Google the ScienceMark writers! Paid AMD shills.

Roborat, Ph.D said...

Roborat, I think you have a typo:

I'm amazed that some of you actually read the entire article because I didn't and I'm the one posted the article!

I swear all I just was just the following:

Ctrl+A (select all)
Ctrl+C (copy)
Ctrl+V (paste)
Ctrl+F (find)
Replace Tab
Intel with AMD
Conre with K10
C2D with Barcelona
3Ghz with 2Ghz
Replaced links
etc.
Clicked "POST"

Damn article is too long to read proof! sorry.

Roborat, Ph.D said...

anonymous said: "APM is a classic example of the misinformation... "

Agreed. APM is just a concept, a terminology or a manufacturing philosophy, or whatever you want to call it. It holds as much meaning as Intel's Copy Exactly! or Motorola's 6-sigma. It's not a single tangible item that you install in a fab and it makes things run better. It's a collection of tools and processes which at the end of the day, is just similar to what every advanced semiconductor company already has in their Fab, just named differently. It's just a catchy term AMD throws around during press conferences to make it sound more profound than it actually is.

I just smile when I see people talk about how APM gives AMD some advantage.

Anonymous said...

http://www.hwupgrade.com/articles/cpu/38/the-technological-future-of-amd-an-interview-with-giuseppe-amato_index.html


Hardware Upgrade: Can you give us some clues about the release dates of AMD’s new generation of processors for servers and desktops?


Giuseppe Amato: As for Opteron solutions, we have already made public the release date which is in September. At that time we will launch the new processors of the Barcelona family. As for the Phenom processors for desktops, the first samples of Phenom FX CPUs will be available by the end of the year, while the remaining versions of the family will arrive by March 2008.

Madness!

Anonymous said...

"I played with a Phenom!"

http://absolutevista1.blogspot.com/2007/07/i-played-with-phenom.html

Roborat, Ph.D said...

I played with a Phenom!

simulated phenom i presume...

Anonymous said...

Probably dual dual K8 underneath lol