GTX 580 Reviewed

[Bosco]

Bosco, if you need my help in going another direction, I could help as well. DBA/Reporting is part of my expertise. I'd just need details of how the current reporting structure works, and how you want to change it, and I could at least give you some options.

 

Thanks for the offer I will let you guys know.

[ccokeman]

Regarding the Furmark/Kombuster thing. Have you tried renaming the .exe?

 

 

Yep Didnt work. Talked to a couple people today and they had the same issue.

[cjloki]

good review frank, good card....maybe when the prices drop a bit later on !

[Nyt]

I remember when ATI prevented the 4870's from running Furmark at full clock speeds because VRM's were being melted .

[ocre]

Yep Didnt work. Talked to a couple people today and they had the same issue.

 

I think it would be better to record the power draw while stressing the GPUs with a demanding game instead of using such programs. it really would represent actual real scenarios and thats really what is important. I think fermi cards being designed around computing can really pull some power on these type of test and it really dont represent the power draw the average gamer would experience. What i am saying is that if power is really important to a person who want to buy a modern GPU for gaming, those full stress scenarios arent a good measurement of what a gamer could expect. Often you see reviewers saying its "the worse case scenario" but when they review all the games they are giving exact data for exact fps in exact real games and the stress testers arent fitting when so much details are real but the power draw is simulated.

 

Easily you could monitor the power draw of the PC throughout the entire benchmarking process. While this data could easily be collected on every game without much worked you could limit it to whatever you would like. The data could be collected on a single game, in real time, while doing already necessary benchmarking of fps. All this without having to setup a simulated max power draw scenario with extra programs. I think with as much tedious work that goes into these reviews to produce exact and perfect data that represents real world games, it would only be natural to want the same real true numbers in the power usage data and the most beautiful aspect is that it would be based on the actual games people play.

 

Doesnt it seem more appropriate that way?

[FashionFish]

If I can find some time this weekend, I'll run some benchmarks on my 8800U SLI setup and compare against the single 580. Wonder what the performance of 2 x 8880Us will compare to 1 x 580. I'm thinking one 580 will blow my SLI setup out of the water.

 

Ouch, my 2 x 8800 Ultras in SLI are no match even for a single 460. Only scored 9373 in 3DMark Vantage's GPU test on high. Time for a new card or two

[ccokeman]

I think it would be better to record the power draw while stressing the GPUs with a demanding game instead of using such programs. it really would represent actual real scenarios and thats really what is important. I think fermi cards being designed around computing can really pull some power on these type of test and it really dont represent the power draw the average gamer would experience. What i am saying is that if power is really important to a person who want to buy a modern GPU for gaming, those full stress scenarios arent a good measurement of what a gamer could expect. Often you see reviewers saying its "the worse case scenario" but when they review all the games they are giving exact data for exact fps in exact real games and the stress testers arent fitting when so much details are real but the power draw is simulated.

 

Easily you could monitor the power draw of the PC throughout the entire benchmarking process. While this data could easily be collected on every game without much worked you could limit it to whatever you would like. The data could be collected on a single game, in real time, while doing already necessary benchmarking of fps. All this without having to setup a simulated max power draw scenario with extra programs. I think with as much tedious work that goes into these reviews to produce exact and perfect data that represents real world games, it would only be natural to want the same real true numbers in the power usage data and the most beautiful aspect is that it would be based on the actual games people play.

 

Doesnt it seem more appropriate that way?

I did just that with a card I just tested and found the thermal load imposed by Crysis warhead was just a significant as MSI's Kombuster utility. I had a higher current draw and the same temperatures. So what does that do for me. It shows me that in game I am getting the increased load on the CPU included on the meter reading so that I cannot show the game test as the deciding factor for current draw since the load on other components goes up instead of just the GPU if that makes sense.

[joel.monteiro]

Why do the colours keep cahnging according to the position of the card,cant u just give each card a permanent color throughout the review???,It will be mush easier reading the graphs,without checking the the colors assigned to the cards on every graph???

[Bosco]

Why do the colours keep cahnging according to the position of the card,cant u just give each card a permanent color throughout the review???,It will be mush easier reading the graphs,without checking the the colors assigned to the cards on every graph???

 

This has been addressed already. The new reviews don't do that anymore.

[ccokeman]

Yes it has ...finally !

[joel.monteiro]

just read the 6850 review nice.