So I've just rebuilt my rig and was going to start overclocking it tonight, so installed Heaven and ran a benchmark to get a baseline figure while everything was at stock. I scored 848, some 600 less than i previously got before the rebuild. I did have my cpu overclocked to 4.3, but it shouldn't make that much of a difference on a largely graphical benchmark. I saw that the iGPU was listed in the graphics, so i disabled it in device manager, unistalled the intel drivers and installed the latest nvidia drivers that have just been released (a clean install). After a reboot or 2 i ran the benchmark again and scored 700. It was so sluggish and jerky. I'm using a powered pcie extender and thought that might be causing an issue. I'm unable to plug my graphics card in without it due to the watercooling loop, so tested it on the wife's rig. It's only got a core2quad and a 480 so lowered to settings to high. First run without the extender scored 824. Second run with the extender scored 834. I don't know what on earth is going on. It's possible the pci extender can cope with the 480 but not the much more powerful 970, i just don't know. Thoughts anyone?
^ Anything less than Titan X is a mid range compromise. Can you try the 970 in your wife's PC, with & without the extender? That aside though, it sounds to me like something isn't configured correctly. It's a new motherboard, right? Have you had a good look through the BIOS settings again?
I could try the card in the wife's rig, but i'd have to drain the loop and put the card back on air to do so. I looked through the bios settings and there was one where you could specify what gen the pcie slots should be running at and had set them to gen3. Could the power supply be interfering with the extender? the cable lies on top of the psu and with the powder coating being removed, there's not much in the way of shielding there. If that was the problem, what type of material could be used to shield the extender?
Wrap some tinfoil round it, single layer to start with, and see if it makes a difference. My money is on shielding.
Based on some benchmarks a user on the Overclock.net forums posted it shouldn't make much difference, other users later on in the topic have tested things like a R9 290X.
Oh hang on - and I've done this too - you are plugging the card (riser) into the correct PCI-E on the mainboard, no?
Yeh I've seen that before, which is why i thought I'd be ok this this. Yeh it's definitely in the top PCI-E slot. I'll give the foil a go tonight see how it goes with that. If that fails I'll redo the loop and plug it straight in the board. In case it makes any difference, this is the riser i got.
Well now things just got weird. Wrapped the riser in foil, 4 layers just to make sure. Plugged it back in, check the bios make sure there nothing I've missed and ran heaven again. Ran really smooth, scored 1441. Dead chuffed. Come out to check previous results and all is well. Happy bunny. Go back in to heaven, chug chug chug. I've pulled it out again as wanted to rewrap it as was a little messy and put a thin price of rubber down too. Unplugged everything out the back, went to move my case and almost crushed my usb stick that I'd put in the front case slot which had my old bench scores on. It occurred to me that I'm between the first goof bench and it going pants again, the only thing that changed was plugging the usb in. Is it possible that could be causing a problem? Will get it plugged back in and see what's what.
Might be a good idea to open up CPU-Z and GPU-Z to really verify everything is as it should be. Stick some screenshots up. Also what settings are you using in valley? EDIT: Settings were in your screenshot, nevermind.
Don't woryy Ramble. I'm an idiot too. ha ha So with sheilding redone and a strip of rubber on top of the PSU, I went again. Ran 2 benches without touching anything and scored 1431 and 1436, exiting and restarting in between to make sure there wasn't anything weird going on. Plugged in the USB drive and scored 1446! Ran again with it in the other USB slot and got 1435. God knows what happened before, but at least it seems to have worked itself out now. Another positive is I'm also getting a solid 1392 on the core without touching the clocks (but with a modded bios with more power), so that missing cap on my M5G was probably stunting performance. Massive thanks to jinqs for the foil tip. Would have been pretty gutted if i had to hide the gpu block again.
Ha ha. I'd forgot about that. I still feel sorry for the 2 old gnomes that used to look after that place. I'm sure we retired one of them we were that bad. lol
