-
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.
TessMark
- Thread starter Raves
- Start date
Soldato
Still, amazing temps with a 1000 core clock. 
i guess its due to having a haf-x caseYou must have a very good airflow in your case
Ok, on the other hand 38% fan is pretty loud, I'd only run it like that if I was playing demanding games.
at 38% i can't hardly hear it over my cpu cooler which as 2fans on it..
at 30% it goes to around 71-73c and it is quiet
Last edited:
Oh well, my graphics is the noisiest component in the system tbh, can hardly hear the CPU fan in idle and cannot hear it in games at all (truth to be told it doesn't go above 1500 rpm and Zalman fan is pretty quiet without any whining noises coming from the heart). Only the GPU bakes the case and makes me wonder if someone's using a hair dryer
Watercooling it is.
Soldato
Nice...
I'll be getting a HAF-X delivered today, so I'm hoping for some high-airflow goodness.
I'm using a GTX480 though (will be two later today), so I can't expect quiet. They're pretty quiet at idle (can't hear it over my fairly quiet case fans etc), but in game the fan tends to climb to 60%, at which point it's pretty damn noisy!
Seeing how little this heats up ati cards i thought i'd give it a go. Scored around 8784 with gpu peaking at wait for it...26 degrees (ambient was 24.5). Makes me think there is a lot of potential to add more tesselation units to this architechture and keep temps down, if it is as simple as bolting them on. I know it was under water and the default benchmark doesn't run for long but as gareth's results on air show there is plenty more scope heat wise with this.
Last edited:
Soldato
Unfortunately, it's just not that straightforward. The geometry data has to be passed down through the rest of the pipeline; after all, tessellation must be performed before the vast majority of rendering as it is essentially a geometry operation. Using a more "classical" approach to the pipeline, as used in r800 or GT200, a massive bottleneck is present. It takes a complete redesign of the architecture, and a very different pipeline process, to remove this bottleneck.
Nvidia put a huge amount of development effort into doing just that, and have managed to produce a very efficient tessellation engine. But as I said earlier, it has cost them dearly in terms of delayed time to market, increased transistor requirement and die size, and also increased power requirements. Everything comes at a price - there is no such thing as a free lunch!
Last edited:
I think the current nvidia architechture will come of age with a die shrink, until then most of that tesselation power will more than likely goto waste as theres nothing that will use it, except some benchmarks. Thanks for the explanation about the hardware, thats where my knowledge is a bit thin 
Soldato
Very cool benchmark gave it a spin last night, was sweet! People wondering why the 400series do better than ATI cards?
Tesselation was the main feature of the Fermi core design. This is why. Not becasue oof physX, rendering textures has nothing to do with physX.
Tesselation was the main feature of the Fermi core design. This is why. Not becasue oof physX, rendering textures has nothing to do with physX.
Fixed results, was running the benchmark in the wrong slot (forgot about it months ago...) 
The temps lowered to 60*C, card is almost silent and scores improved by 3% ish.
BTW compare this result with gareth's score and clocks - 11% increase with 11% overclock - linear scaling = epic win.
And my 1000/1200 overclock at 1.25V:
The temps lowered to 60*C, card is almost silent and scores improved by 3% ish.
BTW compare this result with gareth's score and clocks - 11% increase with 11% overclock - linear scaling = epic win.
And my 1000/1200 overclock at 1.25V:
Last edited: