Friday, March 19th 2010

NVIDIA Claims Upper Hand in Tessellation Performance

A set of company slides leaked to the press reveals that NVIDIA is claiming the upper hand in tessellation performance. With this achievement, NVIDIA is looking to encourage leaps in geometric detail, probably in future games that make use of tessellation. NVIDIA's confidence comes from the way its GF100 GPU is designed (further explained here). Each GF100 GPU physically has 16 Polymorph Engines, one per streaming multiprocessor (SM) which helps in distributed, parallel geometry processing. Each Polymorph Engine has its own tessellation unit. With 15 SMs enabled on the GeForce GTX 480 and 14 on the GeForce GTX 470, there are that many independent tessellation units.

NVIDIA demonstrated its claims in the presentation using the Unigine Heaven, where the GeForce GTX 480 was pitted against a Radeon HD 5870. In many scenes where tessellation is lower, the GPUs performed neck-and-neck, with the GTX 480 performing better more often. But in scenes with heavy tessellation (particularly the "dragon" scene, where a highly detailed model of a dragon needs to be rendered with densely tessellated meshes, the GTX 480 clocks nearly a 100% performance increment over the HD 5870. NVIDIA has been confident about the tessellation performance back since January, when it detailed the GF100 architecture. The GeForce GTX 400 series graphics cards will be unveiled on the 26th of March.


Images Courtesy: Techno-Labs
Add your own comment

145 Comments on NVIDIA Claims Upper Hand in Tessellation Performance

#1
KainXS
so they did drop a sm in the GTX480, meaning it only has 60 tmu's not 64
Posted on Reply
#2
btarunr
Editor & Senior Moderator
by: KainXS
so they did drop a sm in the GTX480, meaning it only has 60 tmu's not 64
Yup, 60 TMUs it is.
Posted on Reply
#4
human_error
hmm yeah i'll wait for some unbiased benchies running on the same (read: not nvidia specific) version of the euginie heaven benchmark. Still if this helps drive tesselation support up in games and gets both teams working on better tesselation then i'm all for it.
Posted on Reply
#5
Steevo
Even at stock I never went below 20FPS during the Dragon scene, At my current overclock i stay about 25FPS at the dragon scene, and that is before the new drivers (10.3a). Plus the benchmark Nvidia was using was version 1.1 according to many, that included up to a 30% performance increase during heavy tessellation by aggressive culling.


So their effective 100% boost can be cut down 30% by the use of the new Heaven benchmark, plus the new drivers ATI is releasing 10-20%, and the ability to overclock most 5870 by at least 20%.


So this is Nvidias fancy way of saying by using selective math we can bend numbers our way, all day. Just like I could say.

Available DX11 Video card manufacturers.

ATI



So ATI has the fastest DX11 video card on the market in any and all configurations.
Posted on Reply
#7
Kitkat
this news is 3 weeks old. they had this video on youtube for the longest time.
Posted on Reply
#8
erocker
I'm curious as to how much of a boost the 10.3a drivers from ATi gave the 5870 in relation to that graph. Anyone with a i7 at 3.3ghz and a 5870 know?
Posted on Reply
#9
crow1001
Is it really necessary to grab all the unsubstantiated fermi trash of the web, at least have a reliable source that can be backed up. The bloke next door could have compiled them graphs for all we know.

by: erocker
I'm curious as to how much of a boost the 10.3a drivers from ATi gave the 5870 in relation to that graph. Anyone with a i7 at 3.3ghz and a 5870 know?
NV were using an released version of heaven in their tests, with up to 30% tessellation optimization, so comparsions are not valid.

performance in heaven with my 5870 and 10.3's is now a lot smoother, never dips below 30 FPS.
Posted on Reply
#10
xaira
waw a 300 uhm, sorry 298watt gpu barely getting the upper hand on a 188watt gpu, thats really something to be proud of.
Posted on Reply
#11
Sasqui
Why compare one selective synthetic benchmark against a 5870, then turn around and provide a whole performance chart relative to a GTX 285?

Confused marketing.
Posted on Reply
#12
MarcusTaz
Did not Nvidia just release a driver that caused video cards to overheat and fry? ;)

I will stick with ATI as far as the eye can see. Nvidia will not get my business, not after their 7900GTX debacle vista driver and memory 2D sync issues and never fessing up to it. They took my money then and will not take it again. Plus everyone knows they cheat with benchmarks. :D

I am very happy with my HD5850 thank you that is green and runs cool...:)
Posted on Reply
#13
truehighroller1
I have the following setup and can run this if you guys like.

GA-EP45-UD3P Rev 1.6
Q9550 E0 @ 4.2~Ghz 1.37v~ 1.34v Under Load
Swiftech Water Cooling
2x2Gb OCZ DDR2 1150Mhz @ 1188MHz 5,5,5,18 @ 2.0v
XTX 5870 @ 1GHz / 1200
3x Western Digital RE3 320Gb 16Mb Cache SataII Drives in Raid0
Corsair HX 850w Power Supply
Antec 1200 Case
3dmark Vantage Score = 19049
3DMark06 Score = 23393
Posted on Reply
#14
mechtech
meh the only thing I have time for nowadays is an ocassional game of CS Source, and my 4850 pushes that beyond my monitor refresh rate anyway.

Why dont video card reviews use HL2 source engine for a gaming benmark, to give relevence for us running all the old classics??
Posted on Reply
#15
evillman
by: Steevo
Even at stock I never went below 20FPS during the Dragon scene, At my current overclock i stay about 25FPS at the dragon scene, and that is before the new drivers (10.3a). Plus the benchmark Nvidia was using was version 1.1 according to many, that included up to a 30% performance increase during heavy tessellation by aggressive culling.


So their effective 100% boost can be cut down 30% by the use of the new Heaven benchmark, plus the new drivers ATI is releasing 10-20%, and the ability to overclock most 5870 by at least 20%.


So this is Nvidias fancy way of saying by using selective math we can bend numbers our way, all day. Just like I could say.

Available DX11 Video card manufacturers.

ATI



So ATI has the fastest DX11 video card on the market in any and all configurations.
It's true, ATI have the most powerful card until now, but in a week the fastest graphic chip will be released.
Posted on Reply
#16
mechtech
by: evillman
It's true, ATI have the most powerful card until now, but in a week the fastest graphic ship will be released.
yep thats the way it goes, back and forth, nice to have competition to stimulate technology and keep prices in check.

And in another 6-8 months or so, it may be back to ATI again, if the 6000 series is out by then, time will tell.
Posted on Reply
#17
crow1001
In a week the fastest GFX card on the planet will still be the 5970..:rockout:
Posted on Reply
#18
phanbuey
by: truehighroller1
I have the following setup and can run this if you guys like.

GA-EP45-UD3P Rev 1.6
Q9550 E0 @ 4.2~Ghz 1.37v~ 1.34v Under Load
Swiftech Water Cooling
2x2Gb OCZ DDR2 1150Mhz @ 1188MHz 5,5,5,18 @ 2.0v
XTX 5870 @ 1GHz / 1200
3x Western Digital RE3 320Gb 16Mb Cache SataII Drives in Raid0
Corsair HX 850w Power Supply
Antec 1200 Case
3dmark Vantage Score = 19049
3DMark06 Score = 23393
Yes!!! I really would like to see that... also what is your GPU score for High in Vantage?
Posted on Reply
#19
DaedalusHelios
by: Sasqui
Why compare one selective synthetic benchmark against a 5870, then turn around and provide a whole performance chart relative to a GTX 285?

Confused marketing.
It was to show progression of the architecture rather than a comparison to the competition like the previous slides were comparing. It didn't confuse me. :laugh:
Posted on Reply
#21
Phxprovost
Xtreme Refugee
im sorry but dosnt your product have to be on the market in order to have an upper hand in anything? :wtf:
Posted on Reply
#22
DeathByTray
Is that a new bench or just rehash of yesteryear? ;p
Posted on Reply
#24
rizla1
wait for the gtx 460 . that will be a gud card. for middleclass
Posted on Reply
#25
nt300
Funny how those are Nvidia clips showing benchmarks. No worry ATI does same ;)
We wait and see real world benchmarks.
Posted on Reply
Add your own comment