![]() |
#31
|
|||
|
|||
![]() Quote:
SKUD |
#32
|
|||
|
|||
![]() Quote:
http://www.hardwarecanucks.com/forum...review-23.html Working SLI in CoD would make a fair difference to the 590 because you cannot turn off SLI. |
#33
|
|||
|
|||
![]() Quote:
I'm not certain whether the clock speed or the memory is the biggest contributor. Last edited by SKUD; 04-15-2012 at 06:11 AM. |
#34
|
|||
|
|||
![]()
SKUD,
In Sli you don't double the ram of one card unfortunately, so your 590 (2x580's on one card @ 1.5gb V/ram) shouldn't give 3gb of ram only 1.5gb. The 580 with 3gb therefore is the reason of better performance. We will have to wait and see if the patch is going to address the V/ram issue. I have 2x1gb cards in Sli and both are working to full capacity indicated by my precision tool. I think at the moment V/ram or lack of it is some of the reason that those of us with 1gb are noticing more problems than those with 1.5gb or more. My frame rates are very good but the stutters down low are a pain.
__________________
Rick Asus M4N98TD-EVO AMD Phenom2 965 x 4 3.4gig 8gig DDR3 Ram 2x GTS 450 Sli (1gig each) 1Tb HDD Partitioned x 5 700w Coolermaster single rail P/S 52a Windows 7 64bit 19" Samsung 931BW monitor 1280 x 960 Resolution |
#35
|
|||
|
|||
![]() Quote:
I believe if my 590 had 6GB (3GB each GPU) and same clock speed as my 580 it would get equal performance. Now the second GPU and the 1.5GB of memory it has is wasted. My precision tool also says both GPUs are going at it full bore. Trouble is I don't think they are splitting the work load, only doubling it. I believe SLI works by splitting the data stream sending half to each GPU and then recombining by interleaving the data after they have processed it. If the software isn't controlling that then both processors get all the data. Someone correct me if my perception is wrong. Last edited by SKUD; 04-15-2012 at 06:12 PM. |
#36
|
|||
|
|||
![]()
Hi SKUD,
Interesting that you say that your precision tool shows both your cards working full bore, mine says the same. I have tried disabling one of my cards and that shows zero use on the V/mem on that, and the fps drops to about 60% of what I get with both cards working but the V/mem still shows near max use on the working card. I have my clocks and voltages set to the same for each card even though they are from different stables and the percentage of use on each card only varies about 2% from one card to the other, that also varies with one card using more, then the other. Temps on both cards are within about 2--3deg of each other with card 2 always slightly lower than card 1 (this is what information I get from the different makers, the EVGA slightly hotter running than the Asus) and both are within the 50--60deg range. I wonder if others using Sli are getting the same results, it would be interesting to find out.
__________________
Rick Asus M4N98TD-EVO AMD Phenom2 965 x 4 3.4gig 8gig DDR3 Ram 2x GTS 450 Sli (1gig each) 1Tb HDD Partitioned x 5 700w Coolermaster single rail P/S 52a Windows 7 64bit 19" Samsung 931BW monitor 1280 x 960 Resolution |
#37
|
|||
|
|||
![]() Quote:
|
#38
|
|||
|
|||
![]() Quote:
![]()
__________________
Gigabyte Z68 Intel 2500K (@4.3 ghz)212 CM Cooler 8GB Ram EVGA 660SC (super clocked) 2GB Vram CORSAIR CMPSU-750TX 750W 64 GB SSD SATA II HD WIN7 UL 64BIT |
#39
|
|||
|
|||
![]()
Could you please expand your explanation. Thanks.
|
#40
|
|||
|
|||
![]()
Icarus,
Thanks for your information, from what you have said it looks like lack of V/mem isn't the cause of stutters with CoD that a lot of us are getting. My two cards are only 1gb each and with CoD they are both using between 990mb--1007mb each (cards @1024mb each), no other games that I have come anywhere near this. RoF is using under 75% on each card, so I'm hoping the patch will address CoD's V/mem demand.
__________________
Rick Asus M4N98TD-EVO AMD Phenom2 965 x 4 3.4gig 8gig DDR3 Ram 2x GTS 450 Sli (1gig each) 1Tb HDD Partitioned x 5 700w Coolermaster single rail P/S 52a Windows 7 64bit 19" Samsung 931BW monitor 1280 x 960 Resolution |
![]() |
|
|