As an enthusiast, I am trying to see the good in all of this. In all the benchmarks it seems as though the intel Q6600 beats the Phenoms. To me, thats ok. From a cost standpoint, AMD is less expensive. Back when I read a compared of benchmarks between Intel QX6850 and AMD QuadFX...the difference was only 10%-15%. That was with the old 90nm. I am hoping that AMD will not dump the FASN8, and I will be running dual quads within the first half of next year....with SLI or CrossfireX. I don't think that Intel is going to added dual socket funtionality the Nehalem, so I guess my underperforming Phenoms will still out perform any Intel box. That is until Intel can role 8 cores into one chip. I applaude both companies for the products that they are launching, and working on...I think we are in store for a really good year. Even if AMD doesn't beat Intel.

I emailed you about that Theo and you said when the NDA was lifted then you would be vindicated, can you show us if that is now the case?


I'm pretty interested in seeing why Phenom's 3DMark numbers just rocket off as its Mhz increase beyond 2.4Ghz. There was some cryptic mention of this behaviour in the Anandtech forums. 

Another question that I'm wondering about now is whatever happened to the QuadFX (aka 4x4, QuadFather) platform that included Nvidia chipsets and AMD processors, which AMD was hoping will compete against Intel quad-cores. It was supposed to be even more spectacular when AMD quad-cores came out -- well AMD quad-cores are now coming out, so why no mention of it anymore? AMD seems to be concentrating on Spider, which seems to be fairly similar, minus the Nvidia chipset.

whether the sample theo got was hand picked or not is unknown, he asked about it and while the said it wasn't Chery picked it wasn't a plain old one, from what I understand its whole batch of ones that did really well.

soo of course they took one from THAT batch and gave it to theo.

It was clocked at 2.6ghz if you read the article, and he overclocked it to 3ghz on standard heatsink.

then with two 2900xt in crossfire, Oced to 825, rather small considering some are shiping that far overclocked now. 

and together the Phenom @3Ghz and the crossfire "supposedly" scored a little over 30,000 3dmarks. Unfortunately some one broke into his car and stole i think $3000 of his personal equipment. This included his laptop which was storing the bench results of the phenom setup. The only one he had in another spot was the 3dmark score.

So no the phenoms wont get 30,000 with two 2900xts , without being significantly overclocked. However the spider setup with the 3870 in crossfire x just might. 

I am interested to see what the benchmarks will be really.

i have a google news alert set for amd phenom and i get articles from all the big online tech sites,and i have to say the intel bias in this mag is just shocking.How much is intel paying you for this dribble?..the only reason intel has all the market is because they have bullied and bribed companies all over the world(in europe intel is facing several law suits),their chips have never been better than amd,quite the opposite actually.The enquirer forgets to mention or explain"native" quad`s vs fsb slowness either,or the fact that phenom is much cheaper than most intel chips!!

Hey Theo,

Where is the 30,000 3DMark Phenom that you blabbered about some time back?

Considering that the only component of a modern Intel package that's more than 65nm is the Southbridge (which uses at most 4W), you're being very misleading in saying that "This is quite important in terms of both die size and the thermal characteristics." No, it's really not important in terms of thermal characteristics. How about comparing the thermal characteristics of an AMD 65nm CPU to an Intel 45nm CPU at similar performance (you'll have to underclock the Intel part quite a bit), then see how the giant gap difference b/w the CPUs makes any motherboard gap pale in comparison. Come on, you can do better than churning out insignificant facts and then drawing incorrect conclusions from them.

@ leadsled - from what I've seen the new overclocking utility allows overclocking not of just the CPU but each individual core on split power plane boards. So your not limited by a sleepy core like on x2s.

@core2dude - the 30,000 3D mark was on a cherry picked 3Ghz quad that AMD demoed. I don't remember if it was CFX or it was a duel processor machine. Probably both.

Am I the only one that actually reads the articles and the references? As for the performance verses the cores', these figures seem pretty accurate to me. ( Since AMD had their hands slapped about benchmarks, it's to be expected though.) So we can expect basically the same performance from spider as the core2s' of the same clock speeds, except in multimedia apps, which is great news. ( You would be lucky to tell the diffrence if it were 10% in most cases anyway ) I just finished specing an upgrade for my Opteron workstations. I might just hold out a little longer after all and see what arrives in December.

I hope AMD surprises use all with the benchmarks of this quad system setup. It would be nice to hear some positive news on from AMD. 

Core2dude, i think your on about this thread (link below) its the Barcelona core the inq was on about, but yeah come test day these systems won't do over 30k on 3dmarks 06, i think that was a load of old quibble. Get the AMD hype going :)

http://www.theinquirer.net/gb/inquirer/news/2007/08/28/amds-3ghz-k10-to-break-30000-3dmark06