Overclock.net › Forums › Graphics Cards › NVIDIA › Graphics cards not performing
New Posts  All Forums:Forum Nav:

Graphics cards not performing - Page 3

post #21 of 35
Thread Starter 
No I have only used MSI Afterburner and EVGA Precision X alternatively.
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
post #22 of 35
I'd be curious to see what happens when you run ati tool to check for artifacts.

How long have you had your cards for?
Cosmos
(15 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 EVGA 680i SLI A1 Radeon HD 7950 8gig Balistix 8500 DDR2 
Hard DriveOptical DriveCoolingOS
2x120gig vertex2 Pioneer D212 H60 Win7 
MonitorKeyboardPowerCase
Westinghouse 37'' Logitech G15 Antec True 650W Cosmos S 
MouseMouse PadAudio
Razer Deathadder Razer eXactMat X-Fi Fatality 
  hide details  
Reply
Cosmos
(15 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 EVGA 680i SLI A1 Radeon HD 7950 8gig Balistix 8500 DDR2 
Hard DriveOptical DriveCoolingOS
2x120gig vertex2 Pioneer D212 H60 Win7 
MonitorKeyboardPowerCase
Westinghouse 37'' Logitech G15 Antec True 650W Cosmos S 
MouseMouse PadAudio
Razer Deathadder Razer eXactMat X-Fi Fatality 
  hide details  
Reply
post #23 of 35
Thread Starter 
Quote:
Originally Posted by danewfie View Post

I'd be curious to see what happens when you run ati tool to check for artifacts.

How long have you had your cards for?

The Gigabyte GTX 670s I've had since August 2012. The EVGA GTX 660 Ti's I've had since last month or the month before.

This may make a difference. I swapped the cards back and forth between systems testing them as 1 single 670 with 1 660 Ti as dedicated physx card as well as had the 670s in one system and the 660 Ti in my test system and then swapped them. Though since then I've done a complete reinstall and format of the operating system and all kinds of driver uninstall, re install, and revert to older ones.

I'll grab that ATI tool and see about this artifacting.
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
post #24 of 35
Thread Starter 
I couldn't find an updated version that would work with my cards. Last version I found was for 2006. I'm thinking this will accomplish the same thing: http://www.evga.com/ocscanner/
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
post #25 of 35
Thread Starter 
So I've ran the OC Scanner and it doesn't present any of the problems I have in other benchmark software or games... Though looking at OCCT and monitoring the 12v it does show that its jumping back and forth between 11.95 and 12.05 every 10-15 seconds.
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
post #26 of 35
Quote:
Originally Posted by Apropo View Post

@hammong I haven't changed anything with my OS from when it was working perfectly until now - do you have a link on information of how this Hyper V affects my system and what will it do if I disable it as well as how to disable it?

Go to Control Panel -> Programs -> Programs and Settings

Click "Turn Windows features on or off"

Check to see if there's a checkmark next to Hyper-V. By default, it's disabled. I'd be really surprised if it was enabled and you didn't turn it on yourself. LOL.

Greg
post #27 of 35
Thread Starter 
it is not enabled tongue.gif
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
post #28 of 35
Quote:
Originally Posted by Apropo View Post

I tested the Gigabyte GTX 670 by disabling SLI and removing the SLI bridge and running the second card as a dedicated physx card. all overclocks have been reverted to stock clocks default.

I ran Hitman Absolution benchmark a couple times without MSAA and this is what the graph looked like:


When I turn MSAA x 8 - it seems to force the cards to try and run at their max boost level without dropping in performance.


Then I ran a test in Starcraft 2 on a custom map called Micro 2.0 with 50 zealots and 20 Medivac ships vs 50 zealots and 20 medivac ships and as you can see in this chart the performance DROPS when load gets put on the cards and then goes back up once the load has reduced.


The only thing that looks that odd in the first graph is the gpu usage being on the low side. I'm assuming you're saying that the usage % goes up when you throw on 8xMSAA, correct? I'm further assuming that you know to NOT be using v-sync when running a benchmark.

The primary reason for this particular behavior (barring v-sync) is *generally* ... a cpu bottleneck in effect.

I doubt that's the case here though, because Hitman is very gpu intensive and not especially CPU intensive, but this is going to be the most common reason for seeing low GPU usage. I'll have a test w/my 670 for you tonight as I have the game as well and a similar card. What resolution were you running that at? I assume it was maxed apart from MSAA?

AFA your 2nd chart, I cannot 'see' from the chart that the performance drops when the 'load' increases. Firstly, there's no FPS shown there, and I'm not looking at the game while studying the graph, so I don't know when the 'intense' parts are actually happening .. and to be honest, you might be thinking something happening is 'intensive' when, in reality, it's not really that intensive wink.gif. But SC2 is pretty well-known for heavy CPU usage, esp. in larger multi-player settings.

You also speak of usage problems with BF3 ... that's another game that, in multi-player mode, is absolutely FAMOUS for low GPU usage issues. Again, I think this has a lot to do with it being very CPU intensive and not having the greatest netcode.

I do know that these cards have a tendency to downclock (even below the 'stock' clocks) when the load is too easy (whenever gpu usage drops below 50%), and/or you have a CPU bottleneck going on (which can BE because the load is too easy ... this is especially common in SLI, when not running AA, at lower resolutions, with stock CPU's wink.gif).

I should also say ... these cards doing this particular thing is really normal when using v-sync, so if you are using v-sync, that's your issue.

Can you do a run of Heaven, maxed out, and show us how the graphs look? That's one test that I know for sure uses very very little CPU, and you should almost always see very close to 99% GPU usage throughout the test, esp. with only one GPU running. The Power Usage % on the Heaven test should be quite a bit higher than 40-50% though, even on a Gigabyte (the GB 670 cards always run MUCH lower power usage % than other 670's due to their extra 2-pin connections, e.g. 8-pin + 6-pin instead of 6+6). I usually see at around 60-85% POWER usage during that test.

It's also important to note the upping the power limit does not MAKE the card use more power. The power it uses 'is what it is' ... all you do by raising the limit is raise the power draw your card CAN pull ... without beginning to downclock/downvolt. It won't 'cause' your power draw to increase.

I could've told you not to bother with the PSU. This problem doesn't sound anything like the type of thing that could be PSU-related, so you can put that out of your mind wink.gif

Lastly ... if you don't have your CPU overclocked, you really aren't doing your rig any favors. 670 SLI is incredibly powerful, and a stock (insert any CPU here) is really nowhere near fast enough to avoid CPU bottlenecks in a LOT of situations, esp. if you're only at 1080p/1200p type of resolution.

I appreciate the fact that it's all very confusing becasue the problem seems 'new', but it *might* not really be as new as you think. Basically your rigs are way overpowered for 1080p resolution, especially with stock CPU, and you're likely going to NEED to run high levels of AA in order to keep your gpu usage level up in the more cpu-intensive games in order to avoid downclocking.

You might also want to look into Downsampling your image. Like, running at 2560x1440 and scaling the image down to 1080p. This is basically a primitive, resource-intensive, but VERY GOOD LOOKING form of AA. There's guides all over about how to do this thumb.gif
Edited by brettjv - 3/22/13 at 7:35pm
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
post #29 of 35
Thread Starter 
First I want to thank you for actually taking the time to come and try and help with this problem. By the way my CPU is normally clocked at 4.2Ghz - I saw no noticable change in performance between stock clock and 4.5Ghz so I lowered it to a frequency that didn't cause un wanted heat.

Quote:
Originally Posted by brettjv View Post

The only thing that looks that odd in the first graph is the gpu usage being on the low side. I'm assuming you're saying that the usage % goes up when you throw on 8xMSAA, correct? I'm further assuming that you know to NOT be using v-sync when running a benchmark.
Without 8XMSAA not just is the usage low, the core clock speeds drop bellow the 980Mhz stock frequency as well the voltage drops. Turning 2XMSAA the top (master) card acts speratic but the bottom one seems to perform as it did before all this started happening. Turning on 8XMSAA both cards are at their max boost, max Voltage, and what I know normal usage on this benchmark.
Now I have since removed both cards and tested each individually - the cards by themselves actually produce higher frames and work normally.
Quote:
Originally Posted by brettjv 
I doubt that's the case here though, because Hitman is very gpu intensive and not especially CPU intensive, but this is going to be the most common reason for seeing low GPU usage. I'll have a test w/my 670 for you tonight as I have the game as well and a similar card. What resolution were you running that at? I assume it was maxed apart from MSAA?
That is correct everything is maxed or as high as I can set it and MSAA is turned off.
Quote:
Originally Posted by brettjv 
You also speak of usage problems with BF3 ... that's another game that, in multi-player mode, is absolutely FAMOUS for low GPU usage issues. Again, I think this has a lot to do with it being very CPU intensive and not having the greatest netcode.

I do know that these cards have a tendency to downclock (even below the 'stock' clocks) when the load is too easy (whenever gpu usage drops below 50%), and/or you have a CPU bottleneck going on (which can BE because the load is too easy ... this is especially common in SLI, when not running AA, at lower resolutions, with stock CPU's wink.gif).
Well I've ran multiple machines with similar setups i5-3570k (STOCK) one system with exactly the same cards SLI Gigabyte 670 and it never presented this issue with AA turned off, its running a 120Hz gaming monitor. Another system is running Crossfire 7970s and again not presenting the same issue i'm having with yet again a stock i5-3570k. My system with a 4.2Ghz overclock up until last month, when this issue started happening, never displayed this problem of boost dropping gpu frequency. Mind you I have always played without AA, I really don't like it. Before this problem started the only time in game play I had ever seen my graphics cards drop in frequency was in a load screen and even then never under stock frequency of 980mhz.

Quote:
Originally Posted by brettjv 
I should also say ... these cards doing this particular thing is really normal when using v-sync, so if you are using v-sync, that's your issue.
unless my Nvidia control panel setting to have it off is broken, in both machines, I do know better then to have it on.
Quote:
Originally Posted by brettjv 
Can you do a run of Heaven, maxed out, and show us how the graphs look? That's one test that I know for sure uses very very little CPU, and you should almost always see very close to 99% GPU usage throughout the test, esp. with only one GPU running. The Power Usage % on the Heaven test should be quite a bit higher than 40-50% though, even on a Gigabyte (the GB 670 cards always run MUCH lower power usage % than other 670's due to their extra 2-pin connections, e.g. 8-pin + 6-pin instead of 6+6). I usually see at around 60-85% POWER usage during that test.
If you look at that first graph, which has the earth quake looking USAGE, that was the start of this problem, since then windows 8 has crashed a multitude of times with different drivers and I lost some of the Heaven runs I did that would show the usage - voltage - core clock frequency were not behaving normally. Or should I say behaving as the system did for the previous 6 months. Now that I have the new Power Supply in it seems to run Heaven some what normally yet the minimum frames dropped significantly but everything else seemed somewhat normal. (I'll post the graph at the bottom)
Quote:
Originally Posted by brettjv 
It's also important to note the upping the power limit does not MAKE the card use more power. The power it uses 'is what it is' ... all you do by raising the limit is raise the power draw your card CAN pull ... without beginning to downclock/downvolt. It won't 'cause' your power draw to increase.
yes yes this is known [/quote]
- I was just testing every possible slider to see how it affected the cards since they aren't working normally. Though one thing I probably should note - when I force max voltage on them their performance drops, the boost drops, and so does usage.
Quote:
Originally Posted by brettjv 
I appreciate the fact that it's all very confusing becasue the problem seems 'new', but it *might* not really be as new as you think. Basically your rigs are way overpowered for 1080p resolution, especially with stock CPU, and you're likely going to NEED to run high levels of AA in order to keep your gpu usage level up in the more cpu-intensive games in order to avoid downclocking.

I didn't respond until today because to be quite frank it came off as extremely condescending and it pissed me off. So I had to cool down before responding. You came off as telling me I really don't know my machine and or what I'm doing with a computer. Like I can't tell the difference in performance now vs when the system worked normally a month ago... Yes this performance change is confusing because it is a very new issue, I run my system playing these same games, same benchmarks with different graphics cards configurations all the time. I'm not confused because I'm some idiot who doesn't know what he's doing, and doesn't know the difference between how my system worked before this started and how its not performing now. When you've been working on computers for 10-15 hours a day for a very long time and something like this, without an obvious answer because there was such a drastic change in performance, it becomes confusing when you can't replicate the problem on another machine. Yes its happened on two of my machines but no others. So when I face a problem that I may be overlooking a simple solution I feel its better to get multiple opinions from people who know their stuff.
My rig is overpowered for 1080p to push a 120hz / 144hz monitor and was very capable of doing so until a month ago when now the FPS and performance has been worse then if I go back to using a single card. Again this was compared to multiple other systems with nearly the same setup but with a stock i5-. I've never ran AA before, I don't like running AA, and my system ran perfectly without running AA for 7 months. Now if I don't turn AA on the system can sometimes not be playable. So yes it is new to my system.

Yesterday I spent well over 12 hours swapping parts, testing, running benchmarks, running games after I installed a new power supply. Low and behold Battlefield 3 is nearly playing like it did a month ago. But performance in other games is still garbage compared to before. I did notice in OCCT that the 12vot with the old powersupply would jump from 11.95 to 12.05 every few seconds. the new power supplies 12v doesn't barely move and if it does its in a long drawn out wave and only fluctuates by .07.

Ok here is the new Heaven benchmark -


and graph -



When my system is running BF3 normal this is what it looks like - the Core clock, voltage, and what not barely ever changes even if usage is all over the place. (When my system is not performing correctly, the core clock and voltage drop when under load) and yes I know the difference. because when it happens the game isn't playable -



Now with heaven - that run looked semi normal, excluding the HUGE loss in minimum frames. When the system isn't working as it should even in heaven the core frequency and voltage act as if the cards are idle - though I couldn't replicate the issue since I installed the new powersupply... go figure. The problem isn't fixed but it seems to have stabled itself atleast in BF3 - So I'm still trying to figure out what could be causing this. If I missed something let me know, if I need to clarify or what not try a different test let me know. I'd like to figure this out before replacing any more parts. Gigabyte wants to replace the motherboard, and or they blame the cpu and suggest I try another i5-3570k. EVGA and Gigabyte both (mind you i've talked with sooooo many different techies from both those companies) state there is absolutely no way that a stock i5-3570k should bottleneck either of my SLI configurations pushing a 1080p 120hz / 144hz monitor. But hey most of the time they don't know what they are talking about either and are easily confused.
Edited by Apropo - 3/23/13 at 12:11pm
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
King
(14 items)
 
Queen
(12 items)
 
My Ultimate Rig
(8 items)
 
CPUMotherboardGraphicsGraphics
i5-3570k @ 4.2 Gigabyte Z77 UP4 TH EVGA GTX 660 Ti  EVGA GTX 660 Ti 
RAMHard DriveCoolingOS
Corsair Vengeance 8Gb 1600Mhz Samsung 830 120Gb SSD Coolermaster 212 Evo Windows 8 
MonitorPowerCaseMouse
Asus 24'' 1080p 2ms  Corsair H750 Phantom 410 Black /w White trim Logitech G9X 
CPUMotherboardGraphicsRAM
AMD FX-9590 Asus Crosshair V Formula-Z Sapphire Radeon HD 7990 Corsair Vengeance Pro CMY8GX3M2A1600C9R 
Hard DrivePowerCaseOther
Samsung 840 Pro Series 256GB XFX P1-1250-BEFX Corsair C70 Vengeance Corsair h100i 
  hide details  
Reply
post #30 of 35
Easy, man. Not trying to talk down to you ... I deal with A LOT of people with varying skill/knowledge levels every day on these boards, & I don't know you, so I only had so much to work with in terms of specs/graphs/games provided, etc ... From what I was shown and the particular games you listed, a CPU BN did not seem out of the realm of possibility (and YES those techs are completely wrong say you'd 'never' get a CPU BN with 2x670 + stock 3570 at 1080p resolution. You won't ALWAYS, but you most certainly CAN get one).

You also never mentioned that you're now getting better performance with one card than you do with two ... or if you did, I missed where you said that. That actually changes the equation quite a bit.

So, yeah ... don't get pissed off at me for saying ... your latest Heaven bench run actually looks way screwed-up. You should not be having those huge drops in GPU usage throughout the first 1/2 of the test like that. If this is representative of what's going on in these games, it's an issue I've never stumbled upon/seen. The fact that you have it on two different systems is REALLY strange. I see you have you tried reinstalling the OS on one or both machines, but I would try *starting out* installing an older nV driver that you know used to work 'right' on your system as the first driver after the reinstall. Maybe even revert back to an older mobo bios, assuming you've updated it at some point. Just to remove as many variables from the 'before vs. now' equation as you possibly can thumb.gif

Other thoughts ... do you have any other SLI bridges laying around that you can try? Have you checked your PCI-Ex bus (i.e. used the Render Test in GPU-Z, to make sure both slots are properly spinning up to PCI-Ex 3.0 when you put a load on)? Might want to also turn off the PCI-Ex power management function as well (turn it to 'max performance').

Are you on the latest BIOS for your board, and/or did you happen to update the bios on both your boards right around when this problem started up? Can you test these cards in any other system you have, sounds like you have lots of high-end computers ... & do both single cards behave the same way as what you see in SLi, regardless of which slot you put them in?

Lastly ... voltage flickering from 11.95 to 12.05 is perfectly normal. The time you worry about your PSU is when your 12V rail is being dragged down to like 11.25V, and/or your computer is just shutting off ... when it's put under a heavy load. Tiny fluctuations like what you describe are fine. Every PSU I've ever had does that wink.gif Bottom-line, I really, really doubt that this issue is PSU-related. IOW, if your BF3 seems 'normal' now, I'd bet $$$ to donuts that if you threw your old PSU back in, it would still be 'normal'. If not, then it'd be because of a bad power cable, or one of your GPU's has a flaky jack.

Good luck in any case brother. Try the suggestions above, if you've not already. I also really doubt it's the CPU, as long as it's properly 'waking up' and running at full turbo clocks when it's put under load, I'm sure it's fine. I'd definitely try a fresh mobo though.

Edit: One other thing ... it's not terribly uncommon (I think it's actually a 'bug') for Keplers to get 'stuck' in a lower-power mode once they start downclocking due to the load being 'too easy'. Like, if I run, say Half-Life 2, using v-sync, and the gpu usage is at like 25%, my clocks and voltage will both drop down to like 600MHz core and .975V (which is normal in this scenario) and kinda get 'stuck' there ... like, even if I shut that game down and try something else that's tough to run, the card won't fully power up anymore ... but this can always be fixed with a reboot. I bring this up just because ... well ... Kepler does have some issues with over-aggressive power management sometimes, for sure.

A Kepler owner kinda has to make the workload difficult enough that their gpu usage stays over 50% for these cards to really work reliably. Like I said before, you might want to look into downsampling, esp. since you have something against AA ... downsampling makes a FANTASTIC looking image, and will up your load on every game by quite a bit ... hopefully it will keep your cards from downclocking, due to the added load intensity. And I'd put your CPU back up to 4.2, cause you need ALL the CPU power you can get here ... esp. if you're wanting to pull down 120FPS.

It kinda seems like you're *trying* to run your system with too easy of a load ... For example: you're running Heaven at 1600x900 with no AA. Your graph showing BF3 ... you did that with settings so easy to run that your GPU usage is only around 50%. What I'm saying is, even if your computer didn't USE to have these problems... trust me, you can easily get some weird 'issues' with Kepler's power management system if you don't put a good load on these cards, and keep 'em well fed with data from the CPU. Their power management system is is really kinda buggy, honestly. It's a new thing, and it's way more elaborate than anything nV ever did. Next gen I hope it works better.

Also ... you've never tried flashing any of these cards to alternate (such as the 1.212V) bios, right?
Edited by brettjv - 3/25/13 at 1:25am
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › Graphics cards not performing