Sneakers_Matri Posted August 16, 2008 Share Posted August 16, 2008 Hey all. I have had some wierd inconsistent issues with my system since I got it. From time to time it has been flawless in the games I normally play, beeing Age of Conan and World in Conflict, both quite heavy on the PC. I got a second graphics card some months ago and I encountered problems during SLI, with wierd crashes and bsods as result. I got this system to work flawless aswell from time to time, but with problems comming back every now and then, and obviously I suspected that to be incompatability issues between Microsoft Vista updates + nvidia drivers causing the bsods and sometimes freezes during gaming. I naturally also suspected my PSU to be too weak to muster a SLI setup with my 8800 GTXs when I initially bought it, but I did some extensive research and found that it "should" handle it. Now some week ago I got my system to become rocksolid during sli play for about 10 days, then drivers changed, windowsupdates changed and I got issues again, with freezes and bsods. So now I turn to you guys to try and 100% cut the PSU as the cause from the list. My system is following: 2x 8800 GTXs, clocked to 630 clocks / 2000 memory ( a decent OC that is ). E6750 2.66 ghz clocked to 3.4 ghz (that to a decent OC ). Western Digital RaptorX 10k rpm. 2x dvd-rom/dvd burner (samsung standard 203 models ). Creative soundblaster Gamer soundcard ( pci ). 4x sticks of 1gb high end fabrique ram. Asus P5N-D 750i motherboard ( basically the same as the Evga 750i FTW mobo but with no unnecessary things ). I'm trying to power all this with a 620W HX modular PSU, wich gives out 50 amps over the three ( actually combined ) 12v rails. In my book this PSU should be able to do this without any problems at all. I have tried to verify this by 4-5 diffrent PSU calculators but to be frank they all give out diffrent estimations for what above system requires, also I know the wattage is the least problem here but the total amps delivered on the 12v rails that matter. Any comment or help in this matter would be higly appriciated. And it would be nice to hear what Ram Guy thinks about this, should the Corsair HX620 be able to power a system of this caliber or do I need to sell my PSU and move up to the 1000W one? :confused: /edit To make this even more of a fuss, I know for certain of people running 3 way 8800 GTX SLI on the 620 wattage PSU with problems only in Crysis ( wich pushes the gpus to very high loads over time ). Their comps hangup, wich in my opinion clearly shows of the PGUs drawing too much power in that game at 1920 resolutions, but that is to be expected since crysis is a really demanding game and a 50 amps psu shouldn't handle 3x g80 in an overclocked system. /edit2 To add something to the discussion ( I hope will arise @ monday :) ) here is an URL of a site that did tests running 8800 GTXs in SLI in a heavily OCed quad cpu system. http://www.legitreviews.com/article/462/1/ And here is a show of the system he used in this test: http://forums.legitreviews.com/viewtopic.php?t=8495 Link to comment Share on other sites More sharing options...
Corsair Employees RAM GUY Posted August 18, 2008 Corsair Employees Share Posted August 18, 2008 I would suspect the HX620 would have enough power for this system. If you want to make sure its not a power related issue, then I would remove/unplug any components which are not necessary in order to run one of the games you are playing (disconnect extra optical drives, hard drives, memory modules, sound card). Also I would reset the system to run at default values. Once the system is stable and you have found the issue, then you can reconfigure your overclock and make sure it is stable. If the system gives you the same problems with a reduced amount of power being drawn, then I would suspect drivers/software issues, however if you have the same problems after reducing the power, then there may either be a problem with the PSU, or with your overclock settings. On a separate note, it may be possible to run 3 cards in SLI on our HX620, but we would definitely recommend a higher wattage PSU to run 3 cards. Link to comment Share on other sites More sharing options...
Sneakers_Matri Posted August 18, 2008 Author Share Posted August 18, 2008 I would suspect the HX620 would have enough power for this system. If you want to make sure its not a power related issue, then I would remove/unplug any components which are not necessary in order to run one of the games you are playing (disconnect extra optical drives, hard drives, memory modules, sound card). Also I would reset the system to run at default values. Once the system is stable and you have found the issue, then you can reconfigure your overclock and make sure it is stable. If the system gives you the same problems with a reduced amount of power being drawn, then I would suspect drivers/software issues, however if you have the same problems after reducing the power, then there may either be a problem with the PSU, or with your overclock settings. On a separate note, it may be possible to run 3 cards in SLI on our HX620, but we would definitely recommend a higher wattage PSU to run 3 cards. Thanks for your quick response Ram Guy, much appriciated. As you say it could be my overclock, but since it is rock solid during 3dmark06 during sli, rock solid during prime95, rocksolid during cccp and the fact I only OCed my ram with 50 mhz, going from 800 to 850 with "very" loose ram timings, 5-5-5-16-2t @2.2 volts would indicate a fairly stable 24/7 system. Not leaving this out as a possibility, though I did have these crashes with the system reset to default bios settings, so I highly doubt it. The main areas I been suspecting, after doing what I could to rule out other possibilities have been the 4 gigs of ram cuppled with Vista + nvidia drivers during SLI. Suspecting that the OS handled the RAM I/O errorus. I have tested this in the past and still had bsods and crashes, though drivers have improved and Vista hotfixes have been released since. ( might be worth testing again to try and confirm/dissconfirm once and for all ). I been suspecting the PSU from time to time, of beeing to weak or having too much ripple, either by design or since it is malefunctioning, I have always doubted this theory myself though, beeing an electrician, I know 620 W is alot and there is always some swinging room, the amps is what is important. [Have seen tests of this PSU handling 840 wattage with no errors or ripple.] Then, to what I belived being the root of the problem, namly Vista + Nvidia drivers cuppled together, doing something they werent supposed to do :sigh!:, this again is very hard to prove, or trubbleshoot. Ah well, I'm gonna take your advice and dissconnect the dvd-rom, floppy, dvd-burner, soundcard and casefans to see if I can get it stable there ( might be that the PSU in my quite highly OCed system is on the limit already and when it gets too intense 3d rendering it fails to deliver steady power ). One more question. Do you know of any reliable PSU requirement calculator/formula or simular that might be of use here? I found 3-4 calculators that support my own theory of the PSU beeing sufficiend when it comes to wattage, but I also have seen results going as high as a req of 900 wattage minimum to run my system, wich is in my book extensive. Thanks again. :sunglasse Link to comment Share on other sites More sharing options...
Corsair Employees RAM GUY Posted August 18, 2008 Corsair Employees Share Posted August 18, 2008 Our PSU Finder recommends our TX750, but this is to be on the safe side. We have powered similar show systems in the past using our HX620. If you have the same problems with the reduced load on the PSU, then I would suspect drivers/software. But if the problem goes away, then there could potentially be a power issue, although I would suspect the HX620 would be enough to power your system. Link to comment Share on other sites More sharing options...
Sneakers_Matri Posted August 19, 2008 Author Share Posted August 19, 2008 Our PSU Finder recommends our TX750, but this is to be on the safe side. We have powered similar show systems in the past using our HX620. If you have the same problems with the reduced load on the PSU, then I would suspect drivers/software. But if the problem goes away, then there could potentially be a power issue, although I would suspect the HX620 would be enough to power your system. Cheers mate, I'll get back to you with my findings in a cupple of days :sunglasse Might be nice for others to know aswell, if the HX620 can power my system or not others can use it as a loose "reference". Link to comment Share on other sites More sharing options...
Corsair Employees RAM GUY Posted August 20, 2008 Corsair Employees Share Posted August 20, 2008 NP Please do let us know! Link to comment Share on other sites More sharing options...
Sneakers_Matri Posted August 27, 2008 Author Share Posted August 27, 2008 NP Please do let us know! Oh hey again mate. A small update. I said I would try to reduce the power usage, but I found that to be very subjective since I wouldn't know by how much I reduced it, and when we are talking over 600 wattage on the 12v rails and 50 amps on the 12v rails taking away a tiny amount of that might be alot of work for very little real experimental results. Anyway I will try this in a cupple of days anyway to see if it supports my other tests, wich I'm gonna explain in the next paragraph :).. Right, the test. My plan was to try and duplicate an extreme 3d gaming situation but without the problemetic games, to try and see if the PSU can handle MORE power strain instead of less, as suggested by you. Now I did load up the new DX10 graphics test 3dMark Vantage, from 2008. I forced x16 AF and x16 aaQ through the Nvidia control panel, unfortunaltly I couldn't change resulotions, but they were @ 1280x1024 ( where my main resultion is 1680x1050 ). Then I proceeded to load up Prime95, wich is one of the heaviest CPU and RAM stress tests there is. CPU load = 100% CPU temps = 59 and 61 degrees C, just borderline of beeing too high ( as any master OC you push it to the limit :x ) GFX clocks @ 630/2000 ( clocks and memory ), no artifacting. GFX temps rose to 70 degrees, wich is the most I ever seen them go up at, so would indicate alot of GPU stress, and close to max gpu loads. Now, the result after 5 tests in a row is no bsod, no crash, no freeze, no nothing. Works like a charm. This would tell me that the PSU itself can cope with the power preassure, since no game strains the comp as multiple and simultanious benches wich push ram usage to 75%+ and a constant 100% cpu load for over 30 mins. - I will conduct a Physx test aswell wich is 100% GPU bound and I'm gonna try and take out DVD BURNER+DVD-ROM+FANS+SOUNDCARD to reduce power usage, and put the sysem on default clocks in a day or so, just to see if the problem persists, wich I think it will. Might wanna add that I was using the newest Nvidia Physx + driver package : 177.92 during this test. Anyway, it seems to me that your PSU is as good as all other reviews showed arleady :D: /edit, Might wanna add that these tests have been done during SLI, and with SLI inducators on showing a near 100% utilization, wich isn't so surprising since 3d mark vantage is opted towards SLI performance. Link to comment Share on other sites More sharing options...
Corsair Employees RAM GUY Posted August 27, 2008 Corsair Employees Share Posted August 27, 2008 Now, the result after 5 tests in a row is no bsod, no crash, no freeze, no nothing. Works like a charm. This would tell me that the PSU itself can cope with the power preassure, since no game strains the comp as multiple and simultanious benches wich push ram usage to 75%+ and a constant 100% cpu load for over 30 mins. I would agree with you. If there were a hardware issue, I would guess that you would have had some sort of failure when stress testing the system. It would seem to me that there is a software/driver issue, causing your instability. Link to comment Share on other sites More sharing options...
Sneakers_Matri Posted August 27, 2008 Author Share Posted August 27, 2008 I would agree with you. If there were a hardware issue, I would guess that you would have had some sort of failure when stress testing the system. It would seem to me that there is a software/driver issue, causing your instability. I thats been my idea the whole time aswell, though I'm ofcourse abit bias since I don't want my hardware to be broken, so I would blame the software anyway - though the tests inducate that aswell. Anyway, I did some further testing today after the oh so joyful BSOD had a new little message for me this day, wich was random hardware error code msg. So I investigated further, and came to the conclusion my soundcard could possibly been overheating, sitting right ontop of the PSU ( wich you know blows air up ) and underneath the 2nd 8800 GTX wich blows air down, and also emmits indirect heat downwards to the soundcard wich is only ~ 1 mm away. and possibly 3-4 mm above the PSU. Anyway I was confident this was the problem and I had finally found the reason, so I jack it out and start testing, and I bsod anyway during SLI and rock solid during single gpu and SLI + prime95 and vantage ( simulatanously ). So I guess my first theory still holds :D: Link to comment Share on other sites More sharing options...
Corsair Employees RAM GUY Posted August 27, 2008 Corsair Employees Share Posted August 27, 2008 Do you have your onboard sound device disabled with the PCI sound card installed? Have you tried removing the sound card, the sound drivers and then use a driver cleaner, just to rule out a possible conflict with the sound card/drivers? I have had similar issues in the past with sound card drivers, and not until I actually used a driver cleaning program was the issue resolved. Link to comment Share on other sites More sharing options...
Sneakers_Matri Posted August 28, 2008 Author Share Posted August 28, 2008 Do you have your onboard sound device disabled with the PCI sound card installed? Have you tried removing the sound card, the sound drivers and then use a driver cleaner, just to rule out a possible conflict with the sound card/drivers? I have had similar issues in the past with sound card drivers, and not until I actually used a driver cleaning program was the issue resolved. You have a point there, and I did suspect the soundcard to be causing problems, but it feels like these problems would also occur during single GPU mode not only SLI, and also not only in games..since for example the Unigine ( http://unigine.com/ ) 3d benchmark uses EAX sound during benches. I ONLY have problems when I enable SLI and run games, SLI is working during benches regardless of cpu / gpu overclocking / underlocking, and regardless of soundcard beeing plugged or not. Today for example with Unigine 3d benchmark I got both graphics cards heated up to 74/77 degrees celcius, that is alot of heat and should indicate heavy GPU loads. The Unigine also utilizes dx10, shader 4.0, physx and both opengl and direct3d engines. I am by no means an expert but it seems odd that the PSU would be less stressed during these benches together with prime95 then in a 1 year old dx10 game as world in conflict. And since I never had any issue what so ever with my soundcard ( wich I know others have from the creative SB line ) it is hard to find juste cause that it should be it that is causing all this when there is no sign of instability during single GPU. Sad thing is I really need the SLI to be able to play WiC 100% smoothly at max settings on a 22 incher 1680x1050 dx10. /edit I'm actually running the OS sounds on the soundcard and using the onboard sound to direct voice coms through. Only reason I got a soundcard was so that I could have 5.1 sound + voice coms :x..nothing like big explosions in surround sound with a nice touch of bass~ Anyway this has been working for almost a year with single gpu ROCKSOLID, so thats why I never really suspected it to be soundcard drivers but more an overheating issue since psu blows hot air up and 2nd gfx card during sli blows hot air down on the soundcard wich has little airflow to begin with ( wich I proved it not to be by removing it yesterday and still had bsods during sli ). Link to comment Share on other sites More sharing options...
Sneakers_Matri Posted August 30, 2008 Author Share Posted August 30, 2008 Little update only. Gonna try now with an external PSU to connect to one GFX card, and if it still hungs up there, when one card has a dedicated 450 wattage 30 amps on 12v rails then it has to be something else then just lack of power comming from the psu. :sigh!: Link to comment Share on other sites More sharing options...
Sneakers_Matri Posted September 1, 2008 Author Share Posted September 1, 2008 Oi oi, just wanted to pop in and say I have solved the problem 100% now. Seems like Vista x64 cannot adress 2x GPUs with 4x1 Gig ramsticks at all during actual gameplay. with 4x1 ram + SLI = crash+bsod with 4x1 ram + NO SLI = work rocksolid with 2x1 ram + SLI = work rocksolid with 2x1 (tested all ram aswell, neither is broke ) ram + NO SLI = work rocksolid Also tested running tripple SLI with a friends 8800 GTX and it worked aswell with 2x1 ram, no problem at all.. the 620W HW psu is a monster, though according to pure mathematics it should be able to power it, since an OCed 8800 GTX runs @ 170 wattage each on 12 volts when peaking 100% load. Quick P = I*U shows one card to draw 170/12 = ~14-15 amps, that x3 equals to 45 amps on the 12V rail, wich 5-8 amps to spare ( 170 W is a high estimate aswell ). The motherboard and CPU will have enough wattage and amps to spare on the 12v rails aswell. Ah well it works flawless now, just wanted to let you guys know :sunglasse Link to comment Share on other sites More sharing options...
manta Posted September 1, 2008 Share Posted September 1, 2008 How are you running 2 8800GTX with only 2 6 pin connections available on that PSU? PLease let me know how you have set that up plus you dvd and HD? Link to comment Share on other sites More sharing options...
Sneakers_Matri Posted September 2, 2008 Author Share Posted September 2, 2008 How are you running 2 8800GTX with only 2 6 pin connections available on that PSU? PLease let me know how you have set that up plus you dvd and HD? Either you use PCI-E splitters or you use molex splitters, hook them up on the 12v rail(s) and then into the cards. Important is to spread the load of the 12v rails to all connectors so you avoid getting a bad load spread during idle ( wich means less room for lending amps between the rails ) For 3x 8800 GTX you need both pci-e splitters and molex since you can only run 2 cards with only pci-e splitters and need 2x molex splitters for the last card. Highly OCed 8800 GTX uses 170 wattage during 99% loads ( according to guru3d, wich I trust ) easy P= U*I formula shows one GTX can at peak use 14 amps, 14x3 = 42 amps used, and there is about 50 amps on the 600 w 12v rail, so that is sufficient ( in theory ) and by my own testing it worked in practice also, though we used a diffrent board since my 750i do not have 3x pci-e bus's :p Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.