The Fermi Thread - Part 3

[metro.cl]

Nut sure how real or accurate this are but we get the 1st leaked benchs.

http://www.hexus.net/content/item.php?item=23032

[annihilat0r]

these have been circulating for weeks now.

[japamd]

All i know, reviewers like SKYMTL and Benchzowner have hinted that GF 100 cards will have AWESOME gaming performance compared to competitor, even on todays games, that GF 100 will stomp RV 870 to the thrash bin where it belongs, so i think i will put more weight on their cues rather than Charlie the satan himself for this launch occasion. We will see in the next few days, who will be flat out right, and the one who talked out of their rear.

I'm tired of believing Charlie only to be mocked up by nVidia fanbois, so this time i'll swear by their side, and see which side is the correct one after all of this unveiled.

QFT

Yeah, I really wanna be surprised by NV!

[metro.cl]

these have been circulating for weeks now.

my bad i didn't know an saw it on hexus.

[Sushi Warrior]

All i know, reviewers like SKYMTL and Benchzowner have hinted that GF 100 cards will have AWESOME gaming performance compared to competitor, even on todays games, that GF 100 will stomp RV 870 to the thrash bin where it belongs, so i think i will put more weight on their cues rather than Charlie the satan himself for this launch occasion. We will see in the next few days, who will be flat out right, and the one who talked out of their rear.

I'm tired of believing Charlie only to be mocked up by nVidia fanbois, so this time i'll swear by their side, and see which side is the correct one after all of this unveiled.

SKYMTL has been very positive of Fermi over at HWC, he certainly is not setting up a bad tone for it. I really have my doubts he could be so positive knowing it is a fail... and Charlie is hardly credible, he can make all sorts of claims and hope he is right with some but the majority of the time he is wrong.

[LordEC911]

SKYMTL has been very positive of Fermi over at HWC, he certainly is not setting up a bad tone for it. I really have my doubts he could be so positive knowing it is a fail... and Charlie is hardly credible, he can make all sorts of claims and hope he is right with some but the majority of the time he is wrong.

Seeing as how he hasn't even opened the package from Nvidia, I don't see how he can be "very positive of Fermi."

[mindfury]

[Vipeax]

Unless my timezone causes me to become blind (4 AM), I can't see why the marked out parts are.... marked out.

Top Part:
Benchmark: PCGH-Save "From Hell's Heart"/Enthusiast details
Res./FSAA/AF: 1.920x1200/4x MSAA (in-Game)/No AF (Driver)
Date Published: 26 March 2010

Bottom Part:
Core i7 @ 3.5GHz, 175x20; SMT/TM off), Intel x58, 3x2.048MB DDR3-1400 (7-7-7-21)

Maybe it's just fake? I just can't figure out why those spots are marked out....

[annihilat0r]

Great blurring, I for one totally didn't see the 1920x1200, 4xAA, no AF settings

Crysis performance doesn't look too good, unfortunately. With a slightly overclocked new 5870 product ATI will be easily able to surpass it in Crysis.

[pentium777]

http://www.semiaccurate.com/2010/03/...tx480-benches/

They really should close down the website and come back as "rarelyifeveraccurate.com"

Likely they amongst a few other websites are just upset they'll never get review cards spewing all the BS they spew. Not trying to be a Nvidia fanboy just reading their comments and putting them into perspective.

[GoldenTiger]

Unless my timezone causes me to become blind (4 AM), I can't see why the marked out parts are.... marked out.

Top Part:
Benchmark: PCGH-Save "From Hell's Heart"/Enthusiast details
Res./FSAA/AF: 1.920x1200/4x MSAA (in-Game)/No AF (Driver)
Date Published: 26 March 2010

Bottom Part:
Core i7 @ 3.5GHz, 175x20; SMT/TM off), Intel x58, 3x2.048MB DDR3-1400 (7-7-7-21)

Maybe it's just fake? I just can't figure out why those spots are marked out....

Yeah, me either, especially with how lightly blurred they are...

[Vipeax]

Oh and actually I didn't even need to "try" to read it.... it's all the same as previously, apart from 16:1 AF to No AF :P.

[***Deimos***]

Man, calm down. You don't have to reply to everything that might be pro-Nvidia. TDP isn't going down, it's going up. I'm talking about 5870 here, its TDP is 188W but the E6 version is supposed to be 210 or 225W or something along those lines.

5890 can and thus will happen.

Virtually all 5870 (850Mhz) can max out overdrive at 900. Most cards do around 920-935 @1.15V. It doesn't take a genius to figure out that better binning or 1.2V could get 950Mhz. But to stay within 225W, probably binning.

Memory on 5870 (1200) overclocks to 1300. I doubt AMD will bother trying to source some rare uber high end GDDR5, since most of performance improvement is from core overclocking.

Therefore, "5870 OC" is probably gonna be something like 950/1250 2GB (with overdrive headroom up to 1000). Despite +12% clock, performance probably only +5%... you know, diminishing returns.

===========

if I recall correctly, 5870 is 4 phase, and 5850 is 3 phase. For 4890 (5 phase?), AMD made better than 4870 PCB.. maybe they'll reuse that idea..

[GoldenTiger]

Oh and actually I didn't even need to "try" to read it.... it's all the same as previously, apart from 16:1 AF to No AF :P.
[*img]http://e.imagehost.org/0779/Ati_Radeon_HD_5970_Grafikkarte_Benchmark_Crysis_Wa rhead_3_PNG.png[/img]

Lol, definite fake... then.

[Vipeax]

I'm wondering though (after looking at all the other reviews, Hemlock launch, Cypress launch, 5850 launch, 4890 etc): this is the only 1 where it says No AF (driver) instead of 16:1 AF (driver), as if they could not force AF with Fermi?...

[pentium777]

Did not see this posted, if it was my apologies. Talks about the launch for Fermi and that they will have 2x more than 5800 series launch did.

http://www.brightsideofnews.com/News...id=1240&page=0

[Blacky]

Did not see this posted, if it was my apologies. Talks about the launch for Fermi and that they will have 2x more than 5800 series launch did.

http://www.brightsideofnews.com/News...id=1240&page=0

Well they could have 2x more available at launch, but if the cards suck in the end doesn't matter

[ethomaz]

And this... GTX 480 sample with 512 SPs and possible 600Mhz (1200Mhz).

Be aware that the 5870 is set as reference card, so those are not FPSs in the graph below, but differences in %.

Tested games, 2560*1600:
Call of Duty World at War, 4xAA, 16xAF
Company of Heroes, Opposing Fronts
S.T.A.L.K.E.R. Clear Sky, NoAA, NoAF
Call of Juarez: Bound in Blood, Maxed
Enemy Territory: Quake Wors, 8xAA, 16xAF
Farcry 2, DX10, HDR On, 4xAA
Crysis Warhead DX10
Dawn of War 2, Ultra
Fallout 3, Ultra, High-Res Texture Pack
HAWX, DX10, SSAO Very High, 4xAA
Resident Evil 5, DX10, Maxed
Wolfenstein, 8xAA, 8xAF, Maxed
Batman: Arkham Asylum, NoAA, Maxed

The scores were a little higher for the ES card with the second set of drivers but nothing record-braking.

Core temp idle was 58C, core temp load around 86-90C, which was good.
It's laud, but not 5800Ultra.
In the tesselation tests, the GTX-ul won with +53% vs 5870, but this is as relevant as the 5-point parallelism issue where the The tesselation performance wasRAdeon wins with 350%+.

Idle power is at least 3 times that of the 5870 and the drained power at load was 283W. The official 480 will have less drained power because it will have a whole shader cluster disabled.



Source: http://forums.amd.com/forum/messagev...VIEWTMP=Linear

[GoldenTiger]

And this... GTX 480 sample with 512 SPs and possible 600Mhz (1200Mhz).
Source: http://forums.amd.com/forum/messagev...VIEWTMP=Linear

If this is accurate... O U C H!!!!

[ethomaz]

If this is accurate... O U C H!!!!

But real card has 480SPs @ 700/1400Mhz... does it make a difference?

[annihilat0r]

in that graph the 5870 beats 480 in Warhead, but from other benchies we know 480 is actually faster. So the 700mhz has definitely sped everything up, so we can expect somewhat higher than the scores there, still not record breaking.

also, this is obv. where Charlie was getting his estimates from. Looks like the increase from 600mhz to 700mhz will prove him wrong, not by much though

[GoldenTiger]

But real card has 480SPs @ 700/1400Mhz... does it make a difference?

By pure math...

32 shaders would be about 6.5% of the shading, lost...

but you'd gain about 17% clock rate... on the cores alone... and a good chunk more memory bandwidth.

You'd also be speeding up the ROP's whereas more cores don't do that alone...

Additionally the TMU's would be sped up as those are tied to the core clock and not shaders as far as I'm aware...

He also said the newer drivers added a couple % performance...

Also worth noting, this guy claims his Engineering Sample card had 512 shaders @ 600mhz, 1200mhz shader, and 2800 (GDDR5) RAM: full specs should be 480 shaders @ 700mhz, 1400mhz shader, and 3600-3700 (GDDR5) RAM.

So my guess is that with those #'s considered, it would be maybe 16-17% better performance for 700mhz @ 480 cores, vs. 600mhz @ 512 cores, taking into account the memory frequency as well. Couple that with a couple of extra percent from those tests adding 3% (he said a few) and we might see a graph looking a bit more favorable. Still, if this guy's accurate, it wouldn't be enough to make it what I would call a clean win for nV here.

[zerazax]

Oh and actually I didn't even need to "try" to read it.... it's all the same as previously, apart from 16:1 AF to No AF :P.

The drivers are different too

[flippin_waffles]

that pie chart reflects market share more than anything else. i have used both ATi and nvidia drivers. i do like having custom profiles with nvidia and in d3d apps i never had trouble with either. i use nvidia cards because i can run opengl apps that dont crash every 5 minutes from drivers.

I wouldn't say it reflects market share more than anything else. I'd say market share might be a small factor though. Regardless of how much, the end result is still that nv's drivers were inferior to ATi's, so the anti-ATi driver crusade has no merit.

Nut sure how real or accurate this are but we get the 1st leaked benchs.

When the earthquake hit Chile, you were the first one I thought of (i'm not familiar with anyone else from Chile). I hope things are OK down there for you and yours.

[japamd]

And this... GTX 480 sample with 512 SPs and possible 600Mhz (1200Mhz).

Source: http://forums.amd.com/forum/messagev...VIEWTMP=Linear

Where did you saw clock's reference?

As GTX480 power draw was about 283W, it could be 512 Cuda Cores card with 700 mhz.