I've recently bought a 'bog standard' 9800GTX for around the £180 mark and it clocks to 760core/2300mem, with no artifacts in ATITool.
I'd presume to get the mem as high as you did that BFG change the voltage that's supplied to the ram slightly (much like doing a 'pencil trick') which is probably why you managed to get it a little higher (2354/1177).
Cheapest 8800GTS I can see is around £150. 8800GTX is £172. 9800GTX is around £180.
Looking at the last 1920x1200 Crysis benchmark, going from 19fps minimum with the 8800GTX to 24fps minimum with the 9800GTX makes a fair amount of difference. I've noticed it's a fair bit smoother than than my old 8800GTS 640/G80 when I put some more of the graphics lubbliness on.
I think I'll stick with the 8800GTS for a little while longer. To be honest, nothing in nVidia's 9xxx line-up looks to be a big enough step up to warrant upgrading. If anything Im a little concerned about escalating power consumption on newer boards.
Tim can i ask why the newer graphics cards arn't testing on crysis's very high setting as these are the top of the line directX 10 cards so i would like to see if they are capable of running this years argueably most ambitious game with no story.
1280x1024 2xAA 0xAF, DX10, Texture/Shadows/Physics/Shaders/Water: High; All other settings: Medium
this really isn't showing the kind of information we need, we need to see if this card can handle the full might of VERY HIGH please could you explain to me why the high settings are still used?
Originally Posted by donnie Tim can i ask why the newer graphics cards arn't testing on crysis's very high setting as these are the top of the line directX 10 cards so i would like to see if they are capable of running this years argueably most ambitious game with no story.
1280x1024 2xAA 0xAF, DX10, Texture/Shadows/Physics/Shaders/Water: High; All other settings: Medium
this really isn't showing the kind of information we need, we need to see if this card can handle the full might of VERY HIGH please could you explain to me why the high settings are still used?
The answer is quite simple: it's because the cards can't run the game at very high at 30fps. To me, getting 12 or 13 fps at very high tells the reader absolutely nothing - apart from the fact it cannot run it. It doesn't tell you which card is faster at settings that you can play the game at. The reason we are not running very high settings in Crysis is because today's cards cannot play the game with very high settings enabled at a frame rate that you can actually play the game at.
I'm in the process of running some numbers for you to prove the point.
Originally Posted by donnie Tim can i ask why the newer graphics cards arn't testing on crysis's very high setting as these are the top of the line directX 10 cards so i would like to see if they are capable of running this years argueably most ambitious game with no story.
1280x1024 2xAA 0xAF, DX10, Texture/Shadows/Physics/Shaders/Water: High; All other settings: Medium
this really isn't showing the kind of information we need, we need to see if this card can handle the full might of VERY HIGH please could you explain to me why the high settings are still used?
The answer is quite simple: it's because the cards can't run the game at very high at 30fps. To me, getting 12 or 13 fps at very high tells the reader absolutely nothing - apart from the fact it cannot run it. It doesn't tell you which card is faster at settings that you can play the game at. The reason we are not running very high settings in Crysis is because today's cards cannot play the game with very high settings enabled at a frame rate that you can actually play the game at.
I'm in the process of running some numbers for you to prove the point.
I wasn't far wrong... an overclocked 3870 X2:
1280x1024 2xAA 0xAF DX10/Very High - 16.5 fps avg / 6 fps min
1680x1050 0xAA 0xAF DX10/Very High - 15.3 fps avg / 5 fps min
1920x1200 0xAA 0xAF DX10/Very High - 13.8 fps avg / 3 fps min
Originally Posted by ChaosDefinesOrder Is there any chance of an über multi-GPU comparison review?
3-way 9800GTX
3-way 8800GTX
2 x 9800 GX2
2 x 3870 X2
4-way CrossfireX
Would be interested to see how the results turn out, especially the Quad SLi vs CrossfireX
Yep - thats what buying these high end cards is all about, if the gx2 scales well in quad sli, then i might buy one, with the option of doubling it up later...
the addition of the sound passthrew so can output sound over hdmi cable is sweet, but to rich for my blood for that one bonus, would rather pick up 8800GT and if you require more performance get another and go sli
There are few cards missing from there.
At £260 we should be able to see different combos up to that money.
I bet 2x 9600GT SLI (£170 total both) will be close to the most expensive monsters at 1680x1050.
Originally Posted by Tim S It's something I've been working on (and am still working on) - it's in the balance at the moment given the amount of time it's taken.
Tim cheers for explaining it to me, strangely i've seen this guy called maxishine on youtube who gets some interesting scores but on a second look you were absolutely correct but i still think that the very high test should be run if only to show how it preforms.
Crysis is an odd beast as it doesn't require 30+ frames to run smoothly and has been seen running at 20's fairly nice but i agree that is far from ideal.
Agreed with Tim in that running Crysis at below at least 25FPS is just not worth it, even if 'Very high' looks lush. Nothing can do it, even Tri-SLI fails and dips down
Quad SLI: I'm trying to find out, but again I am thinking no
Is there anything wrong with the World In Conflict benchmark? My 9600GT at 1680X1050 all at VERY HIGH and open (4AA, 8AF) is giving me, 18 Minimum, 32 Average, 48 Max. Q6600 at 3.33, i650 motherboard.
Originally Posted by Panos Is there anything wrong with the World In Conflict benchmark? My 9600GT at 1680X1050 all at VERY HIGH and open (4AA, 8AF) is giving me, 18 Minimum, 32 Average, 48 Max. Q6600 at 3.33, i650 motherboard.
Do you mean the built in benchmark? The built-in benchmark is incredibly CPU limited - we're using real gameplay.
Comments 1 to 22 of 22
I would have thought benching against that would be logical as it's a pretty common card.
I'd presume to get the mem as high as you did that BFG change the voltage that's supplied to the ram slightly (much like doing a 'pencil trick') which is probably why you managed to get it a little higher (2354/1177).
Cheapest 8800GTS I can see is around £150. 8800GTX is £172. 9800GTX is around £180.
Looking at the last 1920x1200 Crysis benchmark, going from 19fps minimum with the 8800GTX to 24fps minimum with the 9800GTX makes a fair amount of difference. I've noticed it's a fair bit smoother than than my old 8800GTS 640/G80 when I put some more of the graphics lubbliness on.
1280x1024 2xAA 0xAF, DX10, Texture/Shadows/Physics/Shaders/Water: High; All other settings: Medium
this really isn't showing the kind of information we need, we need to see if this card can handle the full might of VERY HIGH please could you explain to me why the high settings are still used?
The answer is quite simple: it's because the cards can't run the game at very high at 30fps. To me, getting 12 or 13 fps at very high tells the reader absolutely nothing - apart from the fact it cannot run it. It doesn't tell you which card is faster at settings that you can play the game at. The reason we are not running very high settings in Crysis is because today's cards cannot play the game with very high settings enabled at a frame rate that you can actually play the game at.
I'm in the process of running some numbers for you to prove the point.
I wasn't far wrong... an overclocked 3870 X2:
1280x1024 2xAA 0xAF DX10/Very High - 16.5 fps avg / 6 fps min
1680x1050 0xAA 0xAF DX10/Very High - 15.3 fps avg / 5 fps min
1920x1200 0xAA 0xAF DX10/Very High - 13.8 fps avg / 3 fps min
Unplayable.
weak 'upgrade'
3-way 9800GTX
3-way 8800GTX
2 x 9800 GX2
2 x 3870 X2
4-way CrossfireX
Would be interested to see how the results turn out, especially the Quad SLi vs CrossfireX
Yep - thats what buying these high end cards is all about, if the gx2 scales well in quad sli, then i might buy one, with the option of doubling it up later...
It's something I've been working on (and am still working on) - it's in the balance at the moment given the amount of time it's taken.
At £260 we should be able to see different combos up to that money.
I bet 2x 9600GT SLI (£170 total both) will be close to the most expensive monsters at 1680x1050.
Encouragement +1
Would be awesome to see
Crysis is an odd beast as it doesn't require 30+ frames to run smoothly and has been seen running at 20's fairly nice but i agree that is far from ideal.
Thank you again for explaining it me.
Quad SLI: I'm trying to find out, but again I am thinking no
I fail to see the one for the user as well, but for the company it makes perfect sense due to much higher margins on the oc'd cards.
Do you mean the built in benchmark? The built-in benchmark is incredibly CPU limited - we're using real gameplay.