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

[cadaveca]

Um, yeah, cause you aren't running 2900's, nor do you havea phys-X, and obviously you set up a system, and keep it going for long periods.


I rarely keep a build together for a week before pulling it apart.

I bench. I play games. But i'm getting old, and my hands can't handle the games any more, so now i bench more.


BTW, you got Vista? You know Vista can overstep the 2GB memory address limit for applications, by virtualization? Memory mangement is not the same as it was before.

http://support.microsoft.com/kb/940105

(released today, BTW)

[CraptacularOne]

Um, yeah, cause you aren't running 2900's.

OK if you say so So your 2900s are somehow able to overcome the very principles of the way memory is handled in Windows? BS.....

[RyderOCZ]

BTW, you got Vista? You know Vista can overstep the 32-bit memory address limit?

Since when can 32-bit Vista react any different than 32-Bit XP? its still 32-bit and it doesn't seem to be coded like 32-bit Server2003 which handles it fine.

I deal with people every day that wonder where their ram went with 32-bit Vista.

[cadaveca]

Ryder, come now, you of all people should know that this is really dependant on the BIOS. PAE switch does nothing if you do not have bios support, correct? PAE oversteps the hardware system ram allocation, does it not?



Now, go get a P5K, dump in 4gb of your OCZ ram, load up vista with a single 2900, and tell me how much ram you have.

Now, add a second card.

Now, add a phys-X.

Still got 3070MB, don't ya?




Just because things should be so, doesn't mean they are. Ideals almost never happen.


Also:

To virtualize video memory, the video memory manager in Windows Vista assigns a virtual address range to every video memory resource. This range is conceptually similar to the copy that an application might create. However, the video memory manager manages the process more efficiently than the application might. The video memory manager uses the virtual address range to handle transitions or over-commitment of video memory. However, the virtual address range is typically unused on a system that has lots of video memory. As long as this virtual address range remains unused, no physical memory is allocated for it. In contrast, the system memory copy that is maintained in the older driver model is guaranteed to be fully populated with physical memory.

From the hotfix.

OK if you say so So your 2900s are somehow able to overcome the very principles of the way memory is handled in Windows? BS.....

Yes, because it's the bios that assigns that physical ram allocation, not windows. Seriously dude, check for yourself, with someone who REALLY knows what they are talking about, 'cause your sources are flawed.

[RyderOCZ]

I will see what I can do Cadaveca.

Let's get back on Topic and not argue about where the ram goes

[cadaveca]

I will see what I can do Cadaveca.

Let's get back on Topic and not argue about where the ram goes

Agreed. The point made by Theo is that the bios is not properly addressing the ram, and this in no way indicates that his info is false or fake, agreed?

I am merely pointing out that Craptacular's logic here is flawed, the ram allocation part really has no importance as to how the system performs, nor does it indicate how many components are installed, and in no way affects getting a system to score 30k in '06.

[CraptacularOne]

Agreed. The point made by Theo is that the bios is not properly addressing the ram, and this in no way indicates that his info is false or fake, agreed?

I am merely pointing out that Craptacular's logic here is flawed, the ram allocation part really has no importance as to how the system performs, nor does it indicate how many components are installed, and in no way affects getting a system to score 30k in '06.

I'm not saying that it how it "got" to 30k (which I still think is BS) in 06 or that it has anyrhing to do with the score itself but in every 32bit build I've had (and believe me I've had a few) there is a memory allocation that is identical to the amount of RAM on your video cards. I have only tested it with 32bit XP and not Vista since my Vista install is 64bit. If they have changed the way that the OS addresses memory in Vista 32bit then I was not aware of it. And I'm certainly not going to wipe my 64bit Vista install just to screw areound with a 32bit memory allocation issue.

Also, f I'm reading that article correctly, they are circumventing the memory allocation in Windows but they are still unable to overcome it 2GB max allocatable memory to a single given process. So that begs the question: What's the point of even being able to see more RAM then if you won't ever use it on a single app?

[xenolith]

There could be quite a loud thud on 9.10.07 if this does prove to be true... all of us fainting in concerted shock... and the price on everything that is Intel falling.

Nah, let's get real...

[dqniel]

I will only make one comment on this.


Intel crossfire = 8x8 PCI-E 1.0 + DDR2/3 MemoryControl on Northbridge.

RD790 Crossfire = 16x16 PCI-E 2.0 + DDR2 IMC.


Now, most comments about HD2900XT performance don't have the cards. I do.

Nobody has posted RD580(16x16 SKT 939/AM2) 2900XT Crossfire numbers.


BUT....

@ 3ghz K8 SKT939, RD580, DDR1 @ 250mhz 3-3-3-8, X1900Crossfire =10.5k '06.

http://service.futuremark.com/compare?3dm06=376335


@ 3ghz Core2, SKT 775, D975X, DDR2 @ 833mhz 4-4-4-8, X1900Crossfire =10.2k '06.


http://service.futuremark.com/compare?3dm06=639467


Now, although these numbers are NOT R600, they illustrate how limits imposed by PCI-E bandwidth @ 8x8 afect Intel's 3dmark scores.


Like it or lump it, clock for clock AMD really isn't that far behind Intel in '06. The only thing currently holding AMD back in benching is lack of high clockspeeds on K8. Add in IPC improvements on the new AMD core, higher PCI-e Speed, and greater overall bandwidth, even without the extra PCI-E speed...things don't seem too "fake" to me....they seem right where I expected, actually.

Thanks for the valuable comparison. This post brought me out of post hibernation because it gave me so much hope for AMD making a comeback

[FlawleZ]

Still your C2D score is not where it should be , ie 16x shouldn't beat 8x with lower CPU power in apples to apples comparison


Beat yes it is higly possible but can not be certain ,,, 30K is totally different league

And so is the P4 vs A64. Fugger is probably right, however hardly anyone actually runs 3DMark05 in seriousness anymore. It's all 06 it seems. Inquirer is notoriously BS, so I'm not believing one bit. But no one should think for even a minute that its an impossible situation. Improbable, yes.

Let's just sit back and wait a couple weeks, shall we?

[hipro5]

What does C2Q at 3GHz and 2900XTs (OC'd similarly) get at 640x480, 800x600, and 1024x768?

At least then we'll get some baselines (and yeah...might as well throw in default 1280x1024 in for kicks, though I SERIOUSLY doubt 30k was achieved at 1280x1024).

100% agree with you....
Maybe it was done at 800 x 600 or so.....just to see the cpu performance etc...

For example: a SINGLE VGA bench at 1150 x I don't remember , gained ~ 1800 more marks.....Now....IF we ran it at 800 x 600, I think that we will get WAY more marks......

Can someone test default bench and then again at 800 x 600?....

30K is a 2K5 score

Theo is mistaken

At which resolution Charles?....

[OBR]

I dont believe in this! I remember "magic scores" of HD2900 before official launch, and todays scores ...

It is only marketing massage, the same like with HD2900 ... nothing more ...

[xlink]

I dont believe in this! I remember "magic scores" of HD2900 before official launch, and todays scores ...

It is only marketing massage, the same like with HD2900 ... nothing more ...

last time I checked all thsoe magic scores which were popping up two weeks before the DNA was up were WORSE than today's...

[bing]

Nothing against AMD !

Its just about this thread, TEN PAGES arguments just about "vapour", unbelievable !

[LordEC911]

Nothing against AMD !

Its just about this thread, TEN PAGES arguments just about "vapour", unbelievable !

At least all this talk will end shortly...

What about all the more in-depth discussion/arguments about future architecture/marchitectures for CPUs and GPUs. Those seem to start at least a year or two before we will ever get the product.

[Lostfaith]

Nothing against AMD !

Its just about this thread, TEN PAGES arguments just about "vapour", unbelievable !

well not exactly "vapour" however I do think we should speculate less until the NDA is lifted on September 10th...

I do like all the mystery that is shrouded upon K10. Hopefulle AMD will strike back and we'll have dirt cheap C2Q's for a change :p

[xlink]

well it is inherant that we make a big deal out of this until the NDA is over.

see some people bought c2qs and c2ds lately and don't want to feel like they got ripped off. Others are looking to buy soon and are hoping beyond hope that it kicks ass so they can brag over all the people with "lowly core 2s"

[LordEC911]

So let's repost the possible explanations.
It's all K10 and it is a BEAST. (highly improbable)
RD790 + Pcie-2.0 (highly improbable)
3dMark05 scores. (highly improbable if Theo's email can be taken at face value)
Tri-Crossfire. (somewhat probable, rough estimates would put K10@3ghz around 6k CPUmarks plus Theo's email)
Not default resolution. (highly probable since no mention was made anywhere in the article)

[Lostfaith]

Hey guys, be it true or not completely true...
it's still probably a much better sign compared to the news about k10 we had up till now: At least now it's very likely it won't totally suck really bad

[Metroid]

Something else I picked out of this "article" to re enforce it's status as BS is...



Now, just how exactly is that possible? Anybody that knows how memory is handled will know. But is basically this: A 32bit OS only has 4GB total memory addresses. So with 2 512MB cards taking up the 1st GB that would leave Windows with an allocatable 3GB of system RAM, not 3.24GB.

This is BS and it did exactly what the writer intended it to do: Stir up controversy and draw attention to their sad excuse of a site (the INQ that is)

You never stop making wasteful posts, first have you understood what he said?. He said the system had 4GB of ram and the Windows detected 3.24 now tell me what is wrong with that?. Sometimes windows does not recognize more than it was detected, because bios or config differs with amount of hardware ram processing, listen man he just told us the truth to what happened in that occasion, as I pointed earlier that 3D mark calculator I distrust at all because this system contained many new improvements, features never seen or benchmarked before. So probably they will update it later as for now I distrust it.

Please stop with the nonsense, double posts, wasted sentences, crap speaking and the most important have some education, read it before you reply, stop saying all the time it is a BS, looks like you got a day off to be trolling this thread.

[naokaji]

Hey guys, be it true or not completely true...
it's still probably a much better sign compared to the news about k10 we had up till now: At least now it's very likely it won't totally suck really bad

hehe yep....

the problem is the comparison to current amd and intel cpu's....

k10 might even end up faster than whats around currently...
but...

(i'll just say one word i promise)

nehalem

[hollo]

shintai reckons that 32 bit vista can use all 4gb of ram (i think?). some other guy said "theo wasn't sure if it was tri-fire". and someone else thinks HD 2900s use 256mb of system memory.

so, the system was running tri-fire, with 4gb memory detected, and each 2900 using 256mb. usable memory = 3.24gb. plausible?

[Metroid]

shintai reckons that 32 bit vista can use all 4gb of ram (i think?). some other guy said "theo wasn't sure if it was tri-fire". and someone else thinks HD 2900s use 256mb of system memory.

so, the system was running tri-fire, with 4gb memory detected, and each 2900 using 256mb. usable memory = 3.24gb. plausible?

Video memory is different than system memory, but then who knows what really happened because many things are new at this very moment. It was a different config after all.

[StealthyFish]

Nothing against AMD !

Its just about this thread, TEN PAGES arguments just about "vapour", unbelievable !

Exactly, and from the Inq at that. We're from Xtremesystems, and we're taking from a dumbass from the Inq who calls Vista Ultimate, "Windows ME2 (Vista) Ultimate)? Are you kidding me? Wtf is ME2? Let's wait for someone more reputable to get their hands on the chip before we start making assumptions. Personally, I would have to say 30k my ass if the graphic cards are in 2x crossfire, but we'll see. Maybe I'll be proven wrong. Can't really prove or disprove anything right now.

You never stop making wasteful posts, first have you understood what he said?. He said the system had 4GB of ram and the Windows detected 3.24 now tell me what is wrong with that?. Sometimes windows does not recognize more than it was detected, because bios or config differs with amount of hardware ram processing, listen man he just told us the truth to what happened in that occasion, as I pointed earlier that 3D mark calculator I distrust at all because this system contained many new improvements, features never seen or benchmarked before. So probably they will update it later as for now I distrust it.

Please stop with the nonsense, double posts, wasted sentences, crap speaking and the most important have some education, read it before you reply, stop saying all the time it is a BS, looks like you got a day off to be trolling this thread.

And I have to say, you should take your own advice. 3dmark is calculated using a formula. They don't pull the numbers out of their ass every time you benchmark. There are set formulas. What you're hinting at is that futuremark decided to hide a bunch of features that wasn't to be benchmarked till later? I don't know that much about futuremark's 3dmark calculation, but I highly doubt it. The numbers all just input into formula. Just as you calculate the pythagorean theorum by inputting the values of "A" and "B" from a diagram of a triangle, 3dmark inputs the numbers it receives from the testing into the equation, regardless. If it was as subjective as you do hint, we wouldn't use it as a general benchmarking tool becuase if there's no uniformity, how can you compare systems? Your own post is wasteful in itself and you continue to repeat that you distrust 3dmark when your logic doesn't make any reasonable sense.

[kemo]

And so is the P4 vs A64. Fugger is probably right, however hardly anyone actually runs 3DMark05 in seriousness anymore. It's all 06 it seems. Inquirer is notoriously BS, so I'm not believing one bit. But no one should think for even a minute that its an impossible situation. Improbable, yes.

Let's just sit back and wait a couple weeks, shall we?

i am not talking about AMD vs Intel ,, i am talking about 2900 CF that wont break 30K on air ,, for me 30K on air is impossible even with Nehalem ,, much more CPU power will result in low gain in SM2/SM3 ,,, 50% more powerful CPU will only do 20% gain SM2 and 10% SM3 , slightly more slightly less
So a 4.5GHZ Qx run with 2900CF will do say 21K
9K SM2
9.5K SM3
6K CPU
give it 50% more CPU power , better drivers , PCI-e 2.0
12K SM2 33% gain
12.5K SM3 30% gain
9K CPU 50% gain
so even that scenario wont break 30K ,,, it is just impossible to do it with 2900XT and i am not talking about K10


any way it is only two weeks and people will buy whatever perform better after OC with respect to price ,,, so if 2900XT and K10 can do 30K even if 05 imagine what a 9800GTX will do !!!!!!!!!!!!!!!