Sign in with
Sign up | Sign in
Your question

AMD's 3GHz K10 to break 30,000 3DMark06

Last response: in CPUs
Share
August 28, 2007 7:15:12 PM

take a grain of salt and read on


AMD's 3GHz K10 to break 30,000 3DMark06

K10: meet (Beverly) Krusher to the 10th degree

http://www.theinquirer.net/?article=41970

By Theo Valich @ the Inquirer: Tuesday 28 August 2007, 18:09
Quote:
LAST WEEK in Leipzig my kit was nicked, but before that happened we asked AMD if it would let us run memory benchmark scores on a system there. The reps gave us the company line and declined, so we decided to disclose the benchmark scores of our own K10 benchmarking here and now.

If you were wondering why AMD was hiding the scores of K10 so secretly, there were two reasons. The first might be that the CPU sucks badly and after AMD comes out, Intel's lads can start celebrating the death of AMD. On the other hand, there the was clear and present danger of the K10 significantly beating not just the current Conroe/Kentsfield generation, but easily out besting Wolfdale/Yorkfield. This statement warrants at least three hatemails from Intel's R&D lads, but all that we will disclose here are results we have in our possession. The pics are gone with my stolen laptop, though.

The particular piece of equipment that was briefly in our possession allowed us to run 3DMark06, the most important to overclockers worldwide, and Everest 4.0, our favourite memory benchmark.

The particular processor was none other than a single socket Barcelona or Agena FX, call it what you will. The reference motherboard containing RD790 chipset packed two HD 2900XT cards, and the memory installed was Corsair’s Dominator PC2-9136C5D, or the same ones we have been using ever since they came out. There was a Raptor hard drive, and that was about that. OCZ will like the fact that PP&C Quaddie CrossFire PSU was installed in the system.

Windows Me II (Vista) Ultimate was installed as the operating system of choice. For some odd reason, the 32-bit version was installed in a system with 4GB of memory, needless to say the system detected 3.24GB, and benchmarketing commenced.

When running at 2.5GHz, the 3DMark06 score ended at 23.768, so we were thrilled to see such a good score coming from two 512MB cards. This showed the clear potential of this four core processor marchitecture, but the helter skelter ride happened after we overclocked the processor to 3.0GHz.

When clocked at 3.0 GHz and equipped with two overclocked HD2900XT cards in CrossFire, Agena FX or single-core Barcelona smashed an index of 30,000 3DMarks 06. Yes, you’ve read it right - the barrier of 30,000 was passed to, barely, 30.031. We know that there are different systems floating around, and we have no doubt that leaks will start appearing after this article goes live. Knowing what was the overclock of Kentsfield processor, Intel Yorkfield has quite a big job to do. One thing is clear, though. Intel needs Nehalem, and it needs it badly.

To be clear with you, dear readers, two HD2900XT cards were overclocked to 830MHz for the core and 900MHz for the memory, but ominous Catalyst 7.7 drivers were used.

Memory wise, it is particularly interesting to see benchmark scores of DDR2-800 on AMD’s Athlon 64 X2 or now this Phenom FX something. When running in dual channel mode, a theoretical bandwidth of 12.8 GB/s can be reached, but that is a pipe dream - of course.

In the real world, our Intel Kentie 2.93GHz test bed will score around 7.5 GB/s, Athlon 64 X2 5000+ shoots around 8.3-8.8 GB/s, depending on the memory latency, and now Phenom scores even better – pretty close to a five figure range. But regardless of missing the 10 GB/s mark, it is still faster than any DDR3 memory on an Intel system, regardless of the clock achieved by the DDR3 memory. If you put the memory on 1066 MHz, 11GB/s bandwidth was smashed with read, write and copy tests and that was by quite some margin.

We expect the results to go officially live prior to Barcelona launch in September. µ


August 28, 2007 7:59:21 PM

it is the inquirer *takes bucket of salt*... but heres hoping.
August 28, 2007 8:04:42 PM

The Inquirer has been getting slightly better lately, but still taken with a Grain of Sodium Chloride.
August 28, 2007 8:07:35 PM

Ya I read that too. The only thing I'm thinking, even though it the Inquirer that this is way too close to the launch to be shoving your foot that far into your mouth! All I have to say is if this is true, Thank God!! We need the competition.
August 28, 2007 9:00:31 PM

So, they were able to run 3DMark06 twice, allowed to overclock it and the GPUs, but weren't allowed to run memory benchmarks.

Plus, the laptop they had pictures of all this just happens to get stolen.

Okay...other than that, totally believable article.

:pfff: 
August 28, 2007 9:13:43 PM

I hate you for beating me to this. I guess it looks like Phenom is the right name. Accordng to an article on TechARP, the SPEC numbers are already done but are under NDA. And AMD rep says that Barcelona is 20-30% faster on average and wins by as much as 170% on some benchmarks.

http://www.techarp.com/showarticle.aspx?artno=434

ALL HAIL THE DUOPOLY!!
August 28, 2007 9:14:08 PM

Am I dreaming? If this comes to realization and Intel losses their lawsuits, they might be the company in trouble.

I'm just excited about the technology. It would be wonderful to have these in desktops and laptops soon.
August 28, 2007 9:15:53 PM

Also, Valdis, it's copyright infringement to post the whole story. BTW, where'd you get that Phenom logo?
August 28, 2007 9:38:42 PM

I think the logo is custom-made.

That aside, those 3DMark numbers are great... but, by the time Phenom X4 chips make their way to market, we'll probably be talking about seeing 50,000 (possibly, or even higher) from an overclocked Yorkfield (possibly from Phenom X4 as well if clocked high enough) and nVidia's next-generation flagship GPU in SLI. I guess all we can say about current 3DMark scores is "stay tuned"...
August 28, 2007 9:38:45 PM

I don't get it. What does a 30,000 score really mean?
Lost planet will still play like crap, Vista will still have problems with the drivers, and the quad or eight core system will not be used to full potential unless doing 3d rendering or flopping...so what if barcelona is as great as this article claims...competition for Intel? bah
August 28, 2007 9:49:15 PM

bronking,

if it is faster than the Intel chips why wouldn't it be competition? Does speed not matter anymore? What is it about... price? If that is the case I guess AMD has already won.
August 28, 2007 9:58:12 PM

For the record, I call BS. I predict Barcelona will underperform.
August 28, 2007 10:03:41 PM

joefriday said:
For the record, I call BS. I predict Barcelona will underperform.




So it begins.... going on a hunch or past experience? The next two weeks may very well prove to be interesting. If for no other reason I would like to see it kick but to shut the haters up. Of course I am not directing this towards you joefriday.... I am speaking about the fanatics.
August 28, 2007 10:17:33 PM

I see you quoted yourself. :) 

As for explanation as to why I think Barcelona will stop short of expectations, my hypothesis is derived from observations of delays, paper charades, vague and highly controlled product demos (which never actually demonstrated high clocked cores nor outstanding performance), lack of third party benchmarks due either to a lengthy NDA or simple lack of engineering samples, Henry Richards exiting the company just prior to Barcelona launch, and finally, the author's well known bias towards AMD. I think the article was written by the author in the hopes that just writing it makes it true.
August 28, 2007 10:19:41 PM

What gets me is why Amd wanted to keep the cpu a secrete...Isn't letting the world know that their quad core is the fastest and best Intel by 30% not a good idea? I'm hoping this is true tho.
August 28, 2007 10:40:13 PM

I don't think anybody with such "readership" would spew B/S if he himself has "Sampled" a 3 Ghz Phenom and it's actually within it's speculated performance bracket. If he is truly B/S ing than it's a very cruel joke to all the readers, and he would lose loads and loads of credibility, if he has any.
a c 86 à CPUs
a b À AMD
August 28, 2007 10:41:12 PM

I don't see why claiming a 3DMark score is so special. When the 2900XT launched, it could be a 8800GTX in 06, but couldn't beat a GTS in a game. Drivers have changed that, but why show GPU results for a CPU? Show me superPi. Show me zip results. Show me what getting this chip will mean to me, not how fast a system with two video cards is overall.
August 28, 2007 10:48:17 PM

That was total score in 3dmark06 using two overclocked ati 2900s if this article is to believed, not cpu score alone.
August 28, 2007 10:59:21 PM

3DMark06 is special because it along with the latest and greatest Graphic Card performance gives you point lead with an amazing Overclock of CPU + GPU synchronously

http://www.futuremark.com/community/halloffame/
(Show's World Record 3DMARK06 @ 27039Points >Powered by ASUS 2900XT Crossfire 1200/1000MHz Mousepot rev4 cooled)

http://service.futuremark.com/compare?3dm06=2835078
Show's exactly what's been used for this world record

So a 2900XT maybe blowing trades with GTS in gaming performance, but in World Records the 2900XT bangs the 8800Ultra in a 4Ways on 3D Benchmarks.
August 28, 2007 11:11:15 PM

what is blowing a trade? Has ATI secretly been trying swap out the R600 GPU's with the G80's and not getting it done?

JoeFriday,

Thanks for bringing that to my attention.....
August 28, 2007 11:16:03 PM

Well, if these scores are even close to being true, I won't have to think to hard about whether or not to buy a K10, and maybe a couple 2900 XTs to back it up. Of course, maybe ATI will have an updated video card by the. But this does look exciting to me.

Then again, if its all a hoax, its back to crying in the beer and buying one of the new Intel chips. Oh well, I got some time yet, so no hurry or headaches over it all.
August 28, 2007 11:20:31 PM

BaronMatrix said:
Also, Valdis, it's copyright infringement to post the whole story. BTW, where'd you get that Phenom logo?

Photoshop :D  .whats the right way to post a story? will they sue me? or will i get banned? :ouch: 
August 29, 2007 12:00:04 AM

Valdis: Thanks for the article!!!

I TOTALLY believe that 30,000 is correct, my only theory is my old system (see sig) with an 8800GTS gets around 7 or 8,000 so a newer architecture with 4 cores and 2 gfx cards in SLI at 3GHZ? For sure I can believe that such a system is only 4x faster. Remember when AMD went 64 bit and Intel said nobody needs it? Look at the wall we hit with RAM requirements this fast... AMD thinks ahead, designs good stuff... Nobody seems to remember when AMD was performance KING before Core came out, so it is simple tick, tock we are seeing in a perfectly functioning duopoly. I feel the same way about AMD as some people feel about Apple products I guess, I'd buy Phenom even if it was slower than an Intel offering if I needed to upgrade. I just shudder at the thought of the crap products Intel would make and sell for top dollar if AMD went under...
August 29, 2007 12:00:27 AM

I'm Be-Fud-dled, K-10 going to suck, K-10 is going be the second coming, Yet nowhere is a reputable benckmark to be found. Just these type of statements with squat to back it up.

All we really seem to have in concrete is that K-10 is late, and not launching anywhere near 3GHZ.

August 29, 2007 1:09:52 AM

The last thing AMD needs to do is hype the new quad core have it underperform the Intel ones.
August 29, 2007 1:19:59 AM

itotallybelieveyou said:
The last thing AMD needs to do is hype the new quad core have it underperform the Intel ones.

Actually you're right. AMD's K7 and K8 Generation were totally written off as "useless" until the benchmarks after it's release gave it, it's Steroid shots into Stardom.

Most people around from that Era would totally remember this.

*That's why AMD isn't claiming 30K 3DMark06 Benchmark nor did it ever state anything about 3.0Ghz Phenom CPU, it's only demonstrated the 3.0Ghz K10 and Tri-Crossfired 2900XT(The same R600 GPUCore which holds the Current 3DMark 05-06 World Records)
August 29, 2007 1:40:45 AM

Well, put it this way... even if this article was total BS, how many people would have read the article and hesitated on buying a new INTEL rig... now consider that before launch if the propoganda mill kicks in full tilt, how many people again will wait till AMD launches their new chip before deciding what to purchase. Now put this to numbers... if 1 million people read this, and only 1% hesitate, that's already 10,000 possible customers that don't buy today... 10,000 x whatever INTEL chips would have been sold at for X price... and that's a lot of money / market share that would have been lost by the day.

Now if it IS true, then that would be great.

Now the big question is if the new AMD chips would be so great, why would they keep a lid on it? So they can keep an ace in the hole vs. INTEL? Doubt it. There's so much industrial espionage going on right now that I'm sure INTEL knows exactly what AMD has planned for the next 12 months and vice versa...

Just my 2 cents...
August 29, 2007 2:20:46 AM

datmantran said:
Well, put it this way... even if this article was total BS, how many people would have read the article and hesitated on buying a new INTEL rig... now consider that before launch if the propoganda mill kicks in full tilt, how many people again will wait till AMD launches their new chip before deciding what to purchase. Now put this to numbers... if 1 million people read this, and only 1% hesitate, that's already 10,000 possible customers that don't buy today... 10,000 x whatever INTEL chips would have been sold at for X price... and that's a lot of money / market share that would have been lost by the day.

Now if it IS true, then that would be great.

Now the big question is if the new AMD chips would be so great, why would they keep a lid on it? So they can keep an ace in the hole vs. INTEL? Doubt it. There's so much industrial espionage going on right now that I'm sure INTEL knows exactly what AMD has planned for the next 12 months and vice versa...

Just my 2 cents...


Friday, the Inq had an article talking about someone lifting all of their s***

http://www.theinquirer.net/?article=41909

It makes me think a couple of Intel guys stole their s*** so they wrote this fud with what datmantran said in mind.
a c 86 à CPUs
a b À AMD
August 29, 2007 2:29:09 AM

sailer said:
Well, if these scores are even close to being true, I won't have to think to hard about whether or not to buy a K10, and maybe a couple 2900 XTs to back it up.


You might have missed what I was getting at. When the 2900XT was released, it was fast enough to take the 3DMark crown. The 2900XT can beat the 8800GTX in 3Dmark. When it first came out however, it couldn't even beat the 8800GTS in most games. I don't care if they can release a 4GHz K10 that can get 75,000 3DMarks. The number of bungholio marks is meaningless if it can't do better then the quad chips of Intel in games/programs. Being able to set a new record is good, but its not the cats cigar if it can't beat Intels quads in most programs.
August 29, 2007 2:42:30 AM

4745454b said:
You might have missed what I was getting at. Being able to set a new record is good, but its not the cats cigar if it can't beat Intels quads in most programs.


No, I don't think I missed what you were getting at. Maybe I should have put quotations around the "if" part of my statement about the scores being so good. Yes, I think getting a good score is important. After all, if the score had been around 15,000, all the nay sayers would have a field day condemning the K10. Scores are important in that they show a potential of the system, but there are surely more things to be considered, as you point out.

When I said that I would think hard about it, I meant that I have been intending getting an Intel powered machine, as I mentioned in my second paragraph. Thinking hard, at least to me, means looking at the varying performance marks from a variety of things before making a final decision, and I have a couple months to do it before building, so I'm in no rush to run off and buy parts and build. If, that is "IF", the K10 performs well, I may then buy a K10 cpu instead of an Intel cpu based. But that's something I'll make the final decision on later.
August 29, 2007 2:50:26 AM

I still cannot see how this can be anything but fiction. The current world record holder in 3dmark06 has a score of 27,xxx using two X2900XT cards, (overclocked higher than the Inquirer's HD2900XTs), and a Kentsfield QX6850 at over 5 GHz. So a Barcelona core, at 3 GHz, can score more than 10% higher than a 5+ GHz kentsfield, despite having more than a 40% clock disadvantage and lower clocked cards? Not buying it one bit.

http://www.ripping.org/benchmarks.php?act=graphicscores...

August 29, 2007 3:00:51 AM

and why wasnt their "record" uploaded to the 3dmark06 servers?
August 29, 2007 3:12:25 AM

Quote:
The current world record holder in 3dmark06 has a score of 27,xxx using two X2900XT cards, (overclocked higher than the Inquirer's HD2900XTs), and a Kentsfield QX6850 at over 5 GHz. So a Barcelona core, at 3 GHz, can score more than 10% higher than a 5+ GHz kentsfield, despite having more than a 40% clock disadvantage and lower clocked cards? Not buying it one bit.


You've said it all.

This 30K score is too much higher than QX6850's to be true. There's no way they'd have this kind of trick up their sleeves without showing it to the world. No way.

This is just too "iffy" to be true. But hey, we'll know soon enough, I guess.
August 29, 2007 3:19:39 AM

skittle said:
and why wasnt their "record" uploaded to the 3dmark06 servers?


Because you are probably looking in the wrong place on the 3dmark06 servers. Odds are it was probably a simulated record on a simulated CPU running at an estimated clockspeed....so you need to look in the simulated scores section ;) 
August 29, 2007 3:26:11 AM

It would be all nice and good if the new Barcelona chips could compete in price with the budget Intel quads, however a 20% increase in performance (i wont give AMD more), isnt worth the premium people will be paying for their chips at launch. Sorry but computer building is about getting the best bang for your buck for most people.

Though I must agree, we NEED AMD to get back into this race with some force, or else the best bang for Intel quads will cost us all 1000's of the aforementioned bucks.
a c 99 à CPUs
August 29, 2007 3:40:08 AM

joefriday said:
I still cannot see how this can be anything but fiction. The current world record holder in 3dmark06 has a score of 27,xxx using two X2900XT cards, (overclocked higher than the Inquirer's HD2900XTs), and a Kentsfield QX6850 at over 5 GHz. So a Barcelona core, at 3 GHz, can score more than 10% higher than a 5+ GHz kentsfield, despite having more than a 40% clock disadvantage and lower clocked cards? Not buying it one bit.

http://www.ripping.org/benchmarks.php?act=graphicscores...


It might be true, it may be way off the mark- it is The Inq. But there are a few things that stand out to me:

1. The board uses a brand-new chipset, the RD790. AMD makes the chipset and the GPUs, so it's possible that the RD790 has features that allow better-than-previous scaling on the CrossFire setup.

2. The Barcelona at 3.0 GHz may very well perform better in a certain benchmark than a 5.0 GHz QX6850 does. Compare the performance in DivX 6.6a in a 3.00 GHz Yorkfield to a QX6850, and that's just due to basically one difference: SSE4. The overall performance of the A0 Yorkfield vs. the Kentsfield is really more like 5% for the Yorkfield, might be the same for the 10h vs. the Kentsfield.

We haven't seen much for 10h benchmarks besides a few that AMD demonstrated on very low-clocked Barcelonas on boards that they admitted had horrible HT issues. Since those were dual-socket chips, HT and NUMA really would kill performance if it isn't working correctly (see FX-70 series reviews) and thus wouldn't give much of an idea of what the chip can really do. So it's hard to tell if the numbers were real or just pulled out the posterior of Charlie.
August 29, 2007 4:08:26 AM

My questions about this starts with the orginal bench of 3DMark, which was done not only with a 5Ghz cpu, but also was super cooled. I heard a while ago that the 2900 would sing with a better cpu, now whether this transforms into better gaming is the question. I would point out tho, that at 5Ghz and oceed so high, the original mark certainly wasnt used for gaming. But if this is true, this setup could be used in this form, which of course looks quite promising
August 29, 2007 5:04:39 AM

turpit said:
Because you are probably looking in the wrong place on the 3dmark06 servers. Odds are it was probably a simulated record on a simulated CPU running at an estimated clockspeed....so you need to look in the simulated scores section ;) 



oh!
August 29, 2007 5:41:04 AM

Tell me how 30,000 3Dmark06 has anything to do with what matters, frame per second is what matters to me. I don't know about you but 30,000-50,000 means what? More salt?
August 29, 2007 11:36:57 AM

killer_roach said:
I think the logo is custom-made.

That aside, those 3DMark numbers are great... but, by the time Phenom X4 chips make their way to market, we'll probably be talking about seeing 50,000 (possibly, or even higher) from an overclocked Yorkfield (possibly from Phenom X4 as well if clocked high enough) and nVidia's next-generation flagship GPU in SLI. I guess all we can say about current 3DMark scores is "stay tuned"...


http://service.futuremark.com/compare?3dm06=2835078

the profile above is a score of QX6850 @ 5110MHz and scored 27039, but adding 2 more ATI HD2900XT to it might lift things over to 30,000 - 40,000 ...
August 29, 2007 1:35:13 PM

lets wait & see this 3Ghz barcelona ...how about penryn can it break the 30k mark06
August 29, 2007 2:37:48 PM

The only reason I can think of them not releasing reliable numbers for the processor earlier if it is great is that they need to keep selling the processors they have on the market.

If AMD releases some great numbers that completely overshadow their previous line of processors a month or two before they are actually released then 99% of the enthusiest community would not purchase their older parts. Even some vendors would likely put off ordering any of the high end processors until the new set came out. They can't go 2 months without selling much because everyone is waiting for something that isn't on the market yet.

We saw the same thing happen, to a lesser extent, simply when Intel announced their price drops on their slightly refreshed processor line. There was at least a month where a lot of people put off buying processors. Even the retailers, such as Newegg, weren't renewing their stock of processors until the new prices went into effect.


As for the benchmark of the K10, it at least shows promise. Hopefully the advantages they claim translate into real world gains. And not that the processor was designed around maximizing certain aspects of the processor benchmarks at the expense of normal application performance. Such as the HD2900 cards with had amazing shader processing numbers but cut down some of the other aspects of graphic rendering, since shaders is the most important number now, and hurt the overall real world performance of the card.
August 29, 2007 6:09:44 PM

The matchup we'll see by 10th September won't have any 3Ghz K10s at all, so this is all pointless anyway. In a few weeks, Intel will also launch the X38 chipset, and there's no hint of an AMD Desktop CPU launch based on K10 yet.

The first battle will probably be between a 5365 Quad-core Xeon @ 2.93Ghz or 3.0Ghz, now that we know that Tigerton will launch next week, and the 1.9Ghz or maybe 2.0Ghz Barcelona CPUs. It's kind of tricky, though: Tigerton has a quad FSB @ 1066Mhz, while Barcelona has a much more sophisticated HT interconnect; Tigerton has the highest clock. But one question stands out: Why are we discussing 3DMark06 scores, which are gamer-related, if this will be a server launch first?

OK, I know, we're trying to see how much potential this architecture has. But I think that posting these ridiculously high scores without actually allowing people to check these data with their own eyes is just spreading unjustified excitement.

It'll still take months before we even get to see desktop CPUs based on that tech. So let's not get overexcited and let's just wait.
August 29, 2007 6:43:19 PM

Personally, I hate Intel. I hope they go out of business. Most importantly, I hope thier executives and management become homeless so I can spit on them like they have spat on the consumer for over 20 years.
August 29, 2007 7:23:24 PM

erocker said:
Personally, I hate Intel. I hope they go out of business. Most importantly, I hope thier executives and management become homeless so I can spit on them like they have spat on the consumer for over 20 years.


Intel may not be perfect, but believe me, you don't want them going out of business. I find your comment rash and insensitive to the large number of employees who had nothing to do with the "debatable" business practices which provoked your little outburst.

How about we give Intel some breathing room. In the face of what plagues AMD, it escapes everyone that Intel has a great number of expectations to meet in the short and long term. They have commited to a tick and tock strategy that would put a great deal of strain on even the largest of research teams. Meanwhile, Intel has to contend with AMD's underdog persona, and as a result, is afforded little room for error.

"This message brought to you by Slap Those Fanboys, a USA company."
!