Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
)! Now if only GPU manufacturers would get the same message about performance per watt as Intel has...Agreed.MikeC wrote: The Intel chip was an advanced yet unreleased model, while the AMD was a current, retail model. When Intel current best is compared w/ AMD's current best, then it will be a more relevant contest.
Yes, I have no doubt that Intel cherry picked those benches in some way, and I will take the result with a grain of salt. But it is true that those result may currently be inflated, it is also true that this is most likely an early sample. If this was the finished product, I am pretty sure that they wouldn't just keep it in their labs for another 6 months. Mass production would be starting much sooner than that. So what I am saying is, with further tweaks, the Conroe will most likely be a good Intel chip even after you strip whatever they might have done to those benches. Unless AMD has something up their sleeves, if I had to bet, my money is on Intel regaining their performance launch at launch.MikeC wrote: -- The comparison was interesting and exciting, but it's naturally biased.
Overclocked current model. AMD won't have a 2.8Ghz dual core out there in 3 months. They have bumbed the release of FX-62 to computex to 6/6/6. So that kind of levels it out.TooNice wrote:Agreed.MikeC wrote: The Intel chip was an advanced yet unreleased model, while the AMD was a current, retail model. When Intel current best is compared w/ AMD's current best, then it will be a more relevant contest.
This might explain something else: why does the AMD boot screen say "Unknown AMD processor"? Maybe it was an overclocked Athlon 64 X2 4800? (Which has smaller caches...) Or, it was an overclocked FX57, which is single core?rpsgc wrote:Too bad I can't remember the site, but they also did a little "benchmarking" much like Anand and guess what, their FX-60 beat the supposed FX-60 @ 2.8GHz. How's that?
Anandtech gave an explanation for this. Older bios didn't reconize the FX-60.NeilBlanchard wrote:This might explain something else: why does the AMD boot screen say "Unknown AMD processor"? Maybe it was an overclocked Athlon 64 X2 4800? (Which has smaller caches...) Or, it was an overclocked FX57, which is single core?
Anandtech wrote:The one item that a number of you pointed out was that the BIOS used on the DFI LANPARTY UT RDX200 (RD480) motherboard was in fact the first version released for this particular motherboard. Intel told us that the motherboard was purchased a little over two weeks ago and the BIOS used on it was what came with the motherboard, but we still agreed with you all that the system should be tested with the latest BIOS to remove all doubt of wrong doing.
Anandtech wrote:So we benchmarked Conroe; we previewed it, under the only circumstances we could. Intel setup the systems, Intel installed the benchmarks and Intel only let us run what it had installed. Given those circumstances we did our best to make sure the comparison was as legitimate as possible. We checked driver revisions, we checked hardware configurations, BIOS settings, and memory timings; we consulted device manager to make sure nothing strange was limiting performance. We did everything we could think of to make sure that the comparison we would present to the world was as transparent as it could be. But the one thing I ’ve come to understand and appreciate is that the AnandTech reader will always keep us honest; many of you came to us with questions and we spent all evening answering them.
Then again, the FX-60 is the current top of the line. Those chips tend to not have much headrooms (compared to lower end models).Erssa wrote:From a performance point of view it's exciting to know how much overclocking potential that Conroe has. Current 90nm FX-60s cannot be pushed over 3.0Ghz with air cooling, so the overclocking headroom of FX-60 isn't really spectacular.
That was exactly my point.TooNice wrote:Then again, the FX-60 is the current top of the line. Those chips tend to not have much headrooms (compared to lower end models).Erssa wrote:From a performance point of view it's exciting to know how much overclocking potential that Conroe has. Current 90nm FX-60s cannot be pushed over 3.0Ghz with air cooling, so the overclocking headroom of FX-60 isn't really spectacular.
Compare the Barton 2500+ and the Barton 3200+
The 2.4Ghz Northwood and the 3.2Ghz Northwood.
Opteron 165 and FX60.
If we use the Opteron 165/170 as an example, a moderate overclock is about 500Mhz, with a very good overclock hit nearly 1Ghz on good air cooling (suject to a good chip). Most overlock will be somewhere in between, which really isn't that bad actually.
Well apparently they will introduce 2+ cores in 2007, presumably this will be quadcore:Do their roadmap have anything interesting?
There is no way to confirm/deny what you write, it is in the realm of conjecture. I would disagree that the production Conroes most likely will be be better than the one was used in this test. Does anyone recall the speed race to 1GHz many years ago? The P3 that made it to 1GHz (or was it a bit higher?) initially was not available for many months thereafter, because Intel's yields were not good enough. This helped to accelerate the development & deployment of the P4. In other words, it's just as possible that this conroe sample is exceptionally good, and final production ones might fall lower on the performance scale. We'll learn soon enough.TooNice wrote:So what I am saying is, with further tweaks, the Conroe will most likely be a good Intel chip even after you strip whatever they might have done to those benches. Unless AMD has something up their sleeves, if I had to bet, my money is on Intel regaining their performance launch at launch.
The rumor is that AMD will skip the 65nm like they skipped 110nm. There shouldn't be coming anything smaller then 90nm before 2007.rpsgc wrote:It will, not 65nm yet, but an even more refined 90nm process.TooNice wrote:If the process is not improved, then I wonder what AMD will do. Do their roadmap have anything interesting?
Oddly, the p4 was made while the tualatin was going on. It came, and almost went, until someone like me saw that it went faster than a p4 and was cooler operating.MikeC wrote:There is no way to confirm/deny what you write, it is in the realm of conjecture. I would disagree that the production Conroes most likely will be be better than the one was used in this test. Does anyone recall the speed race to 1GHz many years ago? The P3 that made it to 1GHz (or was it a bit higher?) initially was not available for many months thereafter, because Intel's yields were not good enough. This helped to accelerate the development & deployment of the P4. In other words, it's just as possible that this conroe sample is exceptionally good, and final production ones might fall lower on the performance scale. We'll learn soon enough.TooNice wrote:So what I am saying is, with further tweaks, the Conroe will most likely be a good Intel chip even after you strip whatever they might have done to those benches. Unless AMD has something up their sleeves, if I had to bet, my money is on Intel regaining their performance launch at launch.
As for whether AMD has anything in the works, how could you think otherwise? The competition never stops, it's been a see-saw battle for years, and neither party simply sits still.
Netburst architechture was incredibly sofisticated with all the advanced branch predictions etc... and for a time it was the best. Intel must have been suprised, when the power consumption increased with the smaller manufacturing process. And then AMD hit the jackpot with the A64 architechture, HT etc... It's a good thing Intel knew to quit in time instead of pushing the P4 any further, which it could have done.~El~Jefe~ wrote:I dont like intel for pushing p4 for this long with high prices and bad bad bad cooling. while keeping Dothan needlessly expensive and not releasing chipsets and boards for it. Yet, now we are supposed to buy them.
Loyalty is hard to find these days. Why would you want to support the company that tried to stop AMD from satisfying your needs for several now?
I think that the BIOS did not support the FX-60, and I think the video driver was "cooked". Here's some thoughts on this situation:rpsgc wrote:Too bad I can't remember the site, but they also did a little "benchmarking" much like Anand and guess what, their FX-60 beat the supposed FX-60 @ 2.8GHz. How's that?
Edit: Here it is.
F.E.A.R.
640x480 0xAA, 0xAF
Conroe: 346 fps
Intel's FX-60 @ 2.8GHz: 247 fps
Bit-tech's FX-60 @ default: 253 fps
Bit-tech's Pentium XE 955: 213 fps
No, AnandTech went back and dispelled these rumors:NeilBlanchard wrote: I think that the BIOS did not support the FX-60, and I think the video driver was "cooked". Here's some thoughts on this situation:
http://voodoopc.blogspot.com/2006/03/if ... chine.html
Intel employ some of the world's smartest scientists and engineers. Increased leakage current at smaller transistor sizes (if materials remain the same) is covered in Microprocessor Fabrication 101. So no, they were not surprised at all; in fact the CTO predicted as early as 2000 that the constant heat increase was unsustainable.Intel must have been suprised, when the power consumption increased with the smaller manufacturing process.
I still refuse to believe that, they were 100% aware of how high the power consumption would rise. Leakage current doesn't exactly increase linearly and it hadn't been an issue with the earlier bigger manufacturing processes. Even the smartest scientist can make mistakes and miscalculations.jaganath wrote:Intel employ some of the world's smartest scientists and engineers. Increased leakage current at smaller transistor sizes (if materials remain the same) is covered in Microprocessor Fabrication 101. So no, they were not surprised at all; in fact the CTO predicted as early as 2000 that the constant heat increase was unsustainable.Intel must have been suprised, when the power consumption increased with the smaller manufacturing process.
Why would they use a video driver that was changed to recognize Conroe, on the AMD system? Why don't (some) of their results jive with other known FX-60 benchmarks?TomZ wrote:No, AnandTech went back and dispelled these rumors:NeilBlanchard wrote: I think that the BIOS did not support the FX-60, and I think the video driver was "cooked". Here's some thoughts on this situation:
http://voodoopc.blogspot.com/2006/03/if ... chine.html
http://www.dailytech.com/article.aspx?newsid=1185
In summary, they updated the BIOS and checked the video driver versions (they were the same), and re-ran tests and got about the same results.
Your questions are reasonable, but I trust AnandTech's judgement here. They said they didn't think anything funny was going on, and based on their reputation in the past, I tend to believe them.The performance picture with regards to Conroe hasn’t really changed all that much - on average we’re still seeing a bit over a 20% increase in performance over an overclocked Athlon 64 FX-60.
Because otherwise there would be piss and moaning for using a different drivers. The drivers were identical. They just needed an update to recognize Conroe.NeilBlanchard wrote:Greetings,
Why would they use a video driver that was changed to recognize Conroe, on the AMD system?
Could be many reasons, but hardly foul play. For example it could be because a different demo was used. The quake4 demo that anandtech normally uses in all of it's benchmarks was more favourable for Intel, then the Quake4 benchmark that Intel had chosen. Personally I'd have to say, that I hadn't seen any overclocked FX-60 Benchmarks on DFI RDX200 before these benches.Why don't (some) of their results jive with other known FX-60 benchmarks?
The difference was supposedly 7fps in F.E.A.R. it hardly affected other benchmarks. Remember also that in the first test merory timings were slower then the specified 4-4-4. They were running infact 5-5-5.And if the BIOS update dropped the difference from 30% down to 20%, then that hardly qualifies as "no affect", does it?
AMD will not really have anything to compete with Conroe. They are basically doing a die shrink, new socket, DDR2 and a speed bump with the release of the FX-62, and X2 5200+ in Q2 '06 and Q4 '06.Mariner wrote:Since we're not going to see Conroe for at least six months why get too excited about it now? Lets just wait and see what AMD has available in comparison when Conroe is released.
Well, this assumes that Conroe is released on schedule. We've no reason to believe it won't be but then we've no reason to believe it definitely will either!stupid wrote:AMD will not really have anything to compete with Conroe. They are basically doing a die shrink, new socket, DDR2 and a speed bump with the release of the FX-62, and X2 5200+ in Q2 '06 and Q4 '06.
I agree that it remains to be seen how much performance improvement we will get from the AM2 Athlons + the faster memory. Personally, I'd guess that AMD will struggle until they move their chips to 65nm because, in theory at least, the smaller process size should provide a large benefit. After the weak P4 it's almost a surprise to see Intel moving in the right direction again and I wonder if AMD will be caught resting on their laurels?stupid wrote:I think AMD will need to release their K-9 (or is it K-10?) CPU to become competitive again. But let's just wait to see how the production version of Conroe compares to the AM2 Athlons.