3dMark 11

[MercuryDoun]

i removed on my of cards, and got P9416

 

http://3dmark.com/3dm11/1142092

 

im guessing even at 4.8ghz, the cpu just cant handle 2 6990's? :S hmmm.

 

Well this time it did properly read 4.8Ghz on your i7... maybe slap the other 6990 in again and run it another time??

 

I guess its possible that even a 4.8Ghz i7-2600k isnt fast enough for 2x 6990's. Maybe sell/return the 6990, and just buy a 6970? They tri-fire really well together. But something still smells fishy here, Even this time when actually being at 4.8Ghz my 4.4Ghz i7-930 still got a higher CPU score... There is no way memory plays THAT big of a factor. You should be breaking 12k CPU scores with your 4.8Ghz i7.

[seriousHUSTLE]

Well this time it did properly read 4.8Ghz on your i7... maybe slap the other 6990 in again and run it another time??

 

I guess its possible that even a 4.8Ghz i7-2600k isnt fast enough for 2x 6990's. Maybe sell/return the 6990, and just buy a 6970? They tri-fire really well together. But something still smells fishy here, Even this time when actually being at 4.8Ghz my 4.4Ghz i7-930 still got a higher CPU score... There is no way memory plays THAT big of a factor. You should be breaking 12k CPU scores with your 4.8Ghz i7.

 

ok so i put the second 6990 back in, even cranked up my cpu to 5ghz, and i get http://3dmark.com/3dm11/1142386

 

pretty sure the cpu just cant handle them lol. i searched around for peoples 2600k scores, and they seem to be getting around the 10-11k mark when its oc.

[MercuryDoun]

ok so i put the second 6990 back in, even cranked up my cpu to 5ghz, and i get http://3dmark.com/3dm11/1142386

 

pretty sure the cpu just cant handle them lol. i searched around for peoples 2600k scores, and they seem to be getting around the 10-11k mark when its oc.

 

Well currently the highest score for multiple GPU's on the OCC thread here. Is Praz: http://3dmark.com/3dm11/716403

 

As you can see, with him being at 4.8Ghz, he breaks 12k CPU score. Has a higher Graphics score and much higher combined score. Unknown what his memory speed is though.

 

Lets be fair... 2x 6990's should be hosing this... So this is why i believe you have some problem. Maybe some kind of issue where your northbridge(QPI) isnt getting enough voltage and its causing things to be much slower than they should be??

 

I dont know, its possible that 2x 6990's is just too much, Your PCI-E slots are running x16/x16 right? If there running x8/x8 that COULD be your problem right there... x8 bandwidth is fine for even a GTX580... but for any dual GPU card you really need the full x16/x16.

 

One way to test if 2x 6990's is too much, is just buy a 6970 like i mentioned before, and use tri-fire, if your score goes up... YAY! But it would be fairly clear that its a CPU bottleneck problem. If it stays the same... Then you have another problem, this would suck as it would be pretty hard to diagnosis. This type of test wont take very long so you should be able to return the 6970 if you dont want/need it. And if you decided to keep it, im sure you can sell either the other 6990 for a price that causes you to lose very little money, there brand new cards so how "Used" could it really be heh.

[seriousHUSTLE]

Well currently the highest score for multiple GPU's on the OCC thread here. Is Praz: http://3dmark.com/3dm11/716403

 

As you can see, with him being at 4.8Ghz, he breaks 12k CPU score. Has a higher Graphics score and much higher combined score. Unknown what his memory speed is though.

 

Lets be fair... 2x 6990's should be hosing this... So this is why i believe you have some problem. Maybe some kind of issue where your northbridge(QPI) isnt getting enough voltage and its causing things to be much slower than they should be??

 

I dont know, its possible that 2x 6990's is just too much, Your PCI-E slots are running x16/x16 right? If there running x8/x8 that COULD be your problem right there... x8 bandwidth is fine for even a GTX580... but for any dual GPU card you really need the full x16/x16.

 

One way to test if 2x 6990's is too much, is just buy a 6970 like i mentioned before, and use tri-fire, if your score goes up... YAY! But it would be fairly clear that its a CPU bottleneck problem. If it stays the same... Then you have another problem, this would suck as it would be pretty hard to diagnosis. This type of test wont take very long so you should be able to return the 6970 if you dont want/need it. And if you decided to keep it, im sure you can sell either the other 6990 for a price that causes you to lose very little money, there brand new cards so how "Used" could it really be heh.

 

Ive tested them on x16/x16 and x8/x8, i still have the bottleneck problem, just my gpu score is slightly higher.

 

I still dont understand how my cpu score is only 10.7k.. could the cpu be dodgy? im guessing not becuase everything runs fine, no random bsod etc.

 

eitherway, when only one card is in my combined score goes up, so im guessing i need to wait for faster cpus haha.

[MercuryDoun]

Ive tested them on x16/x16 and x8/x8, i still have the bottleneck problem, just my gpu score is slightly higher.

 

I still dont understand how my cpu score is only 10.7k.. could the cpu be dodgy? im guessing not becuase everything runs fine, no random bsod etc.

 

eitherway, when only one card is in my combined score goes up, so im guessing i need to wait for faster cpus haha.

 

Yeah i guess so, lol. Its possible your CPU score is lower just because of the fact that your 4 GPU's just require that much CPU usage even while idle...

 

Wait for 32nm Bulldozer... 8 real cores... YUM! Or possibly wait even longer for 22nm Ivy Bridge, but thats quite a ways off still, because they will be socket 2011.

[Water_Hazard]

got the water blocks installed yesterday and got a couple runs in last night. Here is the best I got so far.

 

P15995 i7 @ 4.3ghz / 3x GTX 580 @920/1840/2290 1.125v / ram @ 2154 10-11-10-30

[MercuryDoun]

got the water blocks installed yesterday and got a couple runs in last night. Here is the best I got so far.

 

P15995 i7 @ 4.3ghz / 3x GTX 580 @920/1840/2290 1.125v / ram @ 2154 10-11-10-30

 

Holy hell water... I guess that Hex-core i7 said F-U CPU bottle neck and took it like a champ! That is impressive!

 

Im surprised you didnt push your i7 and the 580's higher up though. Are you giving yourself alittle bit of extra head room for later down the road? Or is the addition of 3x 580's alittle much on your loop?

[Water_Hazard]

Holy hell water... I guess that Hex-core i7 said F-U CPU bottle neck and took it like a champ! That is impressive!

 

Im surprised you didnt push your i7 and the 580's higher up though. Are you giving yourself alittle bit of extra head room for later down the road? Or is the addition of 3x 580's alittle much on your loop?

 

Trust me, I was trying like hell last night to push them further. The cooling is definitely not the problem. I put the two water cooled cards on their own loop using an xspc 360 rad and swiftech mcr655 pump, card 1 max temp was 51c, card 2 max temp was 56c, and the Asus direct cu II was topping out at 68c. I am starting to think that the power supply is the problem. Seemed like no matter what I did, it would just reboot as soon as the first run would start(no bluescreen, no driver crashes, just complete shutdown), at anything over the above clocks on cpu or gpu.

 

I wanted to break 16000 SO BAD but finally at 1:30 I decided to give up for the night since I had to be at work at 8

 

Edit: also, I only have 2 of the cards water cooled. the Asus is a non reference design so the full cover blocks wont fit, but it does such a good job I don't think it needs it anyway.

Edited May 5, 2011 by Water_Hazard

[MercuryDoun]

Trust me, I was trying like hell last night to push them further. The cooling is definitely not the problem. I put the two water cooled cards on their own loop using an xspc 360 rad and swiftech mcr655 pump, card 1 max temp was 51c, card 2 max temp was 56c, and the Asus direct cu II was topping out at 68c. I am starting to think that the power supply is the problem. Seemed like no matter what I did, it would just reboot as soon as the first run would start(no bluescreen, no driver crashes, just complete shutdown), at anything over the above clocks on cpu or gpu.

 

I wanted to break 16000 SO BAD but finally at 1:30 I decided to give up for the night since I had to be at work at 8

 

Edit: also, I only have 2 of the cards water cooled. the Asus is a non reference design so the full cover blocks wont fit, but it does such a good job I don't think it needs it anyway.

 

Yeah wow comparing those temps on the Asus to the water blocked 580's. Whats the point in tossing a water block on it??

 

Dont you have a 1200w PSU?? I guess you could get a 2nd PSU in your case. They have those cool Aux PSU's that fit in a 5.25" bay and are like 400w or 500w. Pretty cheap too.

 

I assumed my PSU was the same problem ive been having with my GTX570. No BSOD no driver crash. Just BAM shutdown, then it auto reboots. VERY annoying... My PSU is a quad-rail and i was thinking that the relatively low amps on the individual rail's were holding me back. Paul was pretty convincing in saying it wasnt my PSU, but i cant get anything over about 1125mv to stay working on the card. Its so weird. Even at my 24/7 900mhz core clock, of which i only need 1050mv, if i set my card to 1150mv or higher it crashes instantly as well. I still really think it is my PSU. Let me know if solve this issue though yeah?

[El_Capitan]

Lol, so close to 16k!

[Speedway]

Trust me, I was trying like hell last night to push them further. The cooling is definitely not the problem. I put the two water cooled cards on their own loop using an xspc 360 rad and swiftech mcr655 pump, card 1 max temp was 51c, card 2 max temp was 56c, and the Asus direct cu II was topping out at 68c. I am starting to think that the power supply is the problem. Seemed like no matter what I did, it would just reboot as soon as the first run would start(no bluescreen, no driver crashes, just complete shutdown), at anything over the above clocks on cpu or gpu.

 

I wanted to break 16000 SO BAD but finally at 1:30 I decided to give up for the night since I had to be at work at 8

 

Edit: also, I only have 2 of the cards water cooled. the Asus is a non reference design so the full cover blocks wont fit, but it does such a good job I don't think it needs it anyway.

 

Man, that CUII only runs 12C higher than your 2nd card under water.....thats impressive! It does sound like a PSU issue with just a shut down like that, but with a 1200W running everything, it sounds weird for me to say that it is a PSU prob But, when you are running everything OCed to the max, that will def make a difference on the needed power, which seems to be your MO Water

[Water_Hazard]

Yeah wow comparing those temps on the Asus to the water blocked 580's. Whats the point in tossing a water block on it??

 

Dont you have a 1200w PSU?? I guess you could get a 2nd PSU in your case. They have those cool Aux PSU's that fit in a 5.25" bay and are like 400w or 500w. Pretty cheap too.

 

I assumed my PSU was the same problem ive been having with my GTX570. No BSOD no driver crash. Just BAM shutdown, then it auto reboots. VERY annoying... My PSU is a quad-rail and i was thinking that the relatively low amps on the individual rail's were holding me back. Paul was pretty convincing in saying it wasnt my PSU, but i cant get anything over about 1125mv to stay working on the card. Its so weird. Even at my 24/7 900mhz core clock, of which i only need 1050mv, if i set my card to 1150mv or higher it crashes instantly as well. I still really think it is my PSU. Let me know if solve this issue though yeah?

 

ok, I am going to pull my corsair 850w from my i7 860 rig tonight and hook it up to the bottom 2 cards to see if that changes anything. I know some MB's have molex connectors to help supply power to the pci-e connectors, wonder why mine doesn't , and wonder if it would help if it did... either way, this is giving me a headache