• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Trubritar 480 benchmarks

Status
Not open for further replies.
Soldato
Joined
6 Oct 2007
Posts
22,261
Location
North West
Trubritar is a top reviewer and one of the most honest you can get so here is his benchmarks of the 480v5870 and 5970.

Right click on vid and watch in youtube if you prefer.

First up 480v5970 in crysis benchmark, surprised to see the 480 use more power than the 5970.


Next 480v5870 in the crysis benchmark, 5870 just beats out the 480


Next 480v5870 in vantage, all evens


Next 480v5870 in heaven 2.0 with extreme tessellation, 480 walks this and gives the 5870 a good beating.


Next 480 in TRI SLI in the heaven 2.0 bench with extreme tessellation and 8xAA, nice FPS.



Finally we have 480 in 3 way SLI (3DMark Vantage) shows impressive scaling in TRI SLI and impressive power usage...


GTX 480 Quad Sli Stone Giant Benchmark

 
Last edited:
Suspended
Joined
18 Oct 2007
Posts
1,862
I hope ATI can improve tessellation performance as the performance difference is massive, I realise it's not a game but jeez, basically twice the performance.

ATI need to close this gap, if only slightly.
 
Soldato
Joined
11 Oct 2008
Posts
3,834
Location
London
I hope ATI can improve tessellation performance as the performance difference is massive, I realise it's not a game but jeez, basically twice the performance.

ATI need to close this gap, if only slightly.
ATI's cards have a set amount of logic dedicated to tessellation, unlike the Nvidia cards which can use all of their shaders if they desire.

It's much less effort for ATI to double the performance of their tessellation just by throwing more die space at it. The debate will always be whether or not to give up that space instead of more shaders.
 
Soldato
OP
Joined
6 Oct 2007
Posts
22,261
Location
North West
Tessellation is not a performance killer in current titles for the 5870, and when we do get demanding DX11 titles even the 480 will be in the teens, by then we will have the next generation DX11 cards anyway, so buying a 480 for tessellation performance is kind of silly, unless you like your 25 FPS in the heaven 2 bench and play it all day.:p
 
Soldato
OP
Joined
6 Oct 2007
Posts
22,261
Location
North West
Well the 5850/70 gained a good 10 FPS in the heaven bench with a driver update, remember the dragon scene, older drivers were like 20 FPS, updated ones give 30 FPS so improvement in tessellation is there.
 
Associate
Joined
20 Nov 2009
Posts
2,050
Location
Haarby, DENMARK
Associate
Joined
20 Jun 2009
Posts
1,431
Location
Stockport
I won't be being this gen of cards i'll wait for revised version of the cards when the arti is unlocked prob end up with DX11.1 .. i made the mistake of spending £900 on to 8800GTX 768s when they launched luckly i sold them for £330 just before the ultra hit G92 came out.

Roll on 6xxx!.
 
Soldato
Joined
18 May 2003
Posts
4,884
I hope ATI can improve tessellation performance as the performance difference is massive, I realise it's not a game but jeez, basically twice the performance.

ATI need to close this gap, if only slightly.

If games don't use this level of tesselation then why would they bother?
 
Caporegime
Joined
20 Jan 2005
Posts
45,616
Location
Co Durham
If games don't use this level of tesselation then why would they bother?

Quoted for truth. Plus once overclocked the 5 series and the new refresh due soon will have enough brute horsepower to match the gtx480 in games and come close in Heaven benchmark.

There is going to have to be some killer tessellation games to warrant ATI doing anymore than this.
 
Soldato
OP
Joined
6 Oct 2007
Posts
22,261
Location
North West
It's OK doing decent FPS in a synthetic bench with nothing going on, in actual games it's a different story as the fermi shaders have to do more than just tessellate unlike in the the heaven 2 bench where it's all fairly static, ATI tessellation is dedicated so it's tessellation performance will be consistent in any game, games will be a different story as already said so purchasing a card on tessellation performance is a waste of time IMO.
 
Man of Honour
Joined
13 Oct 2006
Posts
90,807
How do you quantify "this level of tessellation"? as I said before if they just use it on the odd object it kinda defeats the purpose of tessellation and if they do implement it world wide arbitarily limiting it to a quality level also defeats the purpose - the same data is used to create normal quality tessellation as extreme quality.

Watching the heaven benchmark vid tho I'm a little suspicious, the ATI card was still struggling in mostly shader heavy scenes where it was flying in previous versions either that heavy tessellation is completely crushing it even when its not its primary workload (which isn't unlikely) or the benchmark has been opptomised to play to the nVidia card strengths and exploit the ATI card weaknesses.

End of the day tho if games come out that use tessellation as part of the main world geometry pipeline your gonna have to turn the tessellation level right down on the 5 series cards to get decent performance whereas the nVidia card will take a couple of levels higher even with a shader heavy game, 5 series refresh won't change this, tho the 6 series probably will as it should have ~8x the tessellation performance of evergreen if what I hear is true.
 
Last edited:
Man of Honour
Joined
13 Oct 2006
Posts
90,807
Could the 5870 use a portion of it's shaders for extra tessellation?

Its probably possible to driver level strap it - however you'd lose quite a bit of the performance advantages (I'm guessing you'd have to setup and transform every resulting polygon from scratch - which the tessellation unit doesn't have to do) and hit overall performance even more in shader heavy scenes.
 
Caporegime
Joined
20 Jan 2005
Posts
45,616
Location
Co Durham
How do you quantify "this level of tessellation"? as I said before if they just use it on the odd object it kinda defeats the purpose of tessellation and if they do implement it world wide arbitarily limiting it to a quality level also defeats the purpose - the same data is used to create normal quality tessellation as extreme quality.

I think it will be awhile before tessellation comes into it's own. It reminds me of the early days of AA and the hit you have to take with framerates. Same applies to physx.

It's only now that we have cards powerful enough to run with 4x and 8x aa and still have decent framerates.

I think by the time we have a game with "full" tessellation used we will have moved on one or two generations of graphics cards and they will be powerful enough to run these games at decent framerates.

So to me tessellation will be limited to a few effects in each game. It's hardly used in Metro even.

Both the 5 series and gtx 480 will struggle with these future games.
 
Man of Honour
Joined
13 Oct 2006
Posts
90,807
I think it will be awhile before tessellation comes into it's own. It reminds me of the early days of AA and the hit you have to take with framerates. Same applies to physx.

It's only now that we have cards powerful enough to run with 4x and 8x aa and still have decent framerates.

I think by the time we have a game with "full" tessellation used we will have moved on one or two generations of graphics cards and they will be powerful enough to run these games at decent framerates.

So to me tessellation will be limited to a few effects in each game. It's hardly used in Metro even.

Both the 5 series and gtx 480 will struggle with these future games.

True enough, tho decent tessellation performance shouldn't be that far off, if I understand the GF100 design correctly then beefing up the polymorph engines would quite easily double the tessellation performance without a massive hit on die size, while its not possible on 40nm unless the process is hugely improved it should be possible on even 32nm and its one of the things that has supposedly changed in the new ATI architecture.
 
Status
Not open for further replies.
Back
Top Bottom