|
Try another PSU first IMHO.
|
# ? Feb 12, 2016 14:20 |
|
|
# ? Jun 5, 2024 05:21 |
|
^ My thoughts exactly, especially with Furmark crashing in under a minute. Vostok, do you know how old and what wattage your PSU is?
|
# ? Feb 12, 2016 16:07 |
|
Check your VRM temps, bruh. You rightfully expect the card to throttle once the GPU temp hits a certain point but most BIOSes don't give a poo poo about VRM temps and VRMs start to poo poo out past a certain temp too.
|
# ? Feb 12, 2016 16:30 |
|
I had some similar problems and was going to blame my R9-290 staying at 95C all the time, but it was my 7 year old 500W PSU running an overclocked 2500K and R9-290 that was the issue. Replacing it with a modern 650W gold one fixed everything.
|
# ? Feb 12, 2016 16:59 |
I recently bought an Asus Geforce GTX 970 STRIX and the card refuses to go over ~50% GPU load except on benchmarks like Unigine Heaven 4.0 where it will reach a load of 90+% easily. I have a fairly weak processor (intel core i-3 4130) and 8 GB of Ram, so at first I assumed my CPU has become a bottleneck, however when I check my task manager for the CPU load it stays between 30-40% during gaming. My previous HD Radeon 7850 card never had this issue. I have tried all sorts of driver installs to no avail (using Display Driver Uninstaller to wipe everything clean and re-installing the latest driver (361.75) with different settings). Most games give me ~50 FPS when they should give me at least 80+ FPS, even with my weak processor as far as I can tell. Even more annoyingly all games have these weird short 'hiccups' once every 5-10 seconds, I assume it's because the driver is throttling my GPU even though I have set every setting I can think of, both in Windows 10 and the Nvidia Control Panel and my BIOS set to maximum power usage, disabled VSync, removed FPS cap, etc. GPU stats are otherwise completely normal (temp is at ~50C normally, 65C during benchmarks, clock rate is normal, etc.) except for GPU load. Some games like Alien: Isolation do reach ~100 FPS regularly but the GPU load is still the same and the annoying hiccup issue remains. From what I can find on the internet a fair number of GTX 700 and 900 series users have this problem, including people with much better processors and overall specs than mine. The few 'solutions' I could find to this problem is to artificially ram the requirements up sky high so that it forces your card to go for max load but I've had little success with this strategy (like forcing 4K downsampling through the DSR settings in my Nvidia Control panel). Nvidia has done nothing so far despite a couple thousand strong petition. Although I strongly suspect that it's a driver issue artificially throttling my GPU with no way of shutting it down, has anyone else with this issue been able to fix it? If I can't reasonbly fix this issue I'll most likely return the card coming Monday and just put my HD Radeon 7850 back in but I really don't want to, obviously. Ekster fucked around with this message at 17:54 on Feb 12, 2016 |
|
# ? Feb 12, 2016 17:49 |
|
Ekster posted:I recently bought an Asus Geforce GTX 970 STRIX and the card refuses to go over ~50% GPU load except on benchmarks like Unigine Heaven 4.0 where it will reach a load of 90+% easily. I have a fairly weak processor (intel core i-3 4130) and 8 GB of Ram, so at first I assumed my CPU has become a bottleneck, however when I check my task manager for the CPU load it stays between 30-40% during gaming. My previous HD Radeon 7850 card never had this issue. I have tried all sorts of driver installs to no avail (using Display Driver Uninstaller to wipe everything clean and re-installing the latest driver (361.75) with different settings). Most games give me ~50 FPS when they should give me at least 80+ FPS, even with my weak processor as far as I can tell. Even more annoyingly all games have these weird short 'hiccups' once every 5-10 seconds, I assume it's because the driver is throttling my GPU even though I have set every setting I can think of, both in Windows 10 and the Nvidia Control Panel and my BIOS set to maximum power usage, disabled VSync, removed FPS cap, etc. From what I remember, this was a common issue with early Geforce Experience. However that's been long patched... have you set your power settings to prefer maximum and all that?
|
# ? Feb 12, 2016 19:12 |
|
I bought a MSI brand 950 on Ebay during some $125 deal. Holy crap, kinda really impressed with it. Even at $160 (you can still find it at $130 with rebates) it seems like a good budget card with a low 90W power requirement. I pushed it's core clock to 1500 mhz boosted and it's memory clock to 7600 mhz and it's neeeeaarrrly within spitting distance with the 960. I booted up Fallout 4 with the settings it had for the 970 and it's still running at 60fps but with frame drops here and there. Automata 10 Pack fucked around with this message at 19:40 on Feb 12, 2016 |
# ? Feb 12, 2016 19:33 |
THE DOG HOUSE posted:From what I remember, this was a common issue with early Geforce Experience. However that's been long patched... have you set your power settings to prefer maximum and all that? Yes. The same problem occured when I reinstalled my drivers without Geforce Experience. I just checked that it's the latest version, 2.9.1.35.
|
|
# ? Feb 12, 2016 19:35 |
|
Ekster posted:Yes. The same problem occured when I reinstalled my drivers without Geforce Experience. I just checked that it's the latest version, 2.9.1.35. I would try a minor overclock with Afterburner. Other than that, it seems like you've done the normal stuff. If exchanging is easy, that'd be my next move. If two cards do the same thing then its likely related to your computer and you can start there. Is the motherboard stuck in PCIe 0.99 beta mode or something? penus penus penus fucked around with this message at 19:59 on Feb 12, 2016 |
# ? Feb 12, 2016 19:50 |
|
Ozz81 posted:^ My thoughts exactly, especially with Furmark crashing in under a minute. Vostok, do you know how old and what wattage your PSU is? Only eight months old, an EVGA supernova 550 GS. Got a pretty good review: http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story6&reid=438 Seamonster posted:Check your VRM temps, bruh. You rightfully expect the card to throttle once the GPU temp hits a certain point but most BIOSes don't give a poo poo about VRM temps and VRMs start to poo poo out past a certain temp too. When I get home from work I'll look at the VRM temps. Can I see that in MSI Afterburner or do I need something else?
|
# ? Feb 12, 2016 20:38 |
|
VostokProgram posted:When I get home from work I'll look at the VRM temps. Can I see that in MSI Afterburner or do I need something else? I think you can still use tools like GPU-Z to show you the temp of the VRMs at idle and load, it has a Sensors tab that shows what's monitored. If it's not the PSU or anything else the card might just be bad, might be worth swapping parts around if you've got a spare card to test with.
|
# ? Feb 12, 2016 21:23 |
|
obviously its a beta, but is overwatch really shittily optimized? One of the settings in High causes my 980ti to throttle and the fps plunges to 15 from high 70s. My cpu and ram arent hitting 100% but the GPU sure is
|
# ? Feb 12, 2016 22:47 |
THE DOG HOUSE posted:I would try a minor overclock with Afterburner. Other than that, it seems like you've done the normal stuff. If exchanging is easy, that'd be my next move. If two cards do the same thing then its likely related to your computer and you can start there. Just tried some overclocking with afterburner, still no dice. Is there a way to check the PCI motherboard thing? My motherboard bios (asrock b85m pro4) doesn't show anything about PCIe modes, but I'm not experienced with that kind of stuff. Is this a hardware issue or a driver issue, I wonder? Maybe I can get the shop I got it from to run some tests to make sure. I really wanted to have a Nvidia card this time around because games tend to have more support for them from what I've seen but I might trade this card for an AMD equivalent.
|
|
# ? Feb 12, 2016 22:57 |
|
Honestly except for maybe some launch day stuff that sometimes gives them a bit of a heads up on frames for a day or three, NV support hasn't been that much better than AMD in my recent experience, and if you go back to AMD and have to wait a day you can use it as an incentive to wait for games to end up on sale. It's exceedingly unlikely that it's with the 970 in general rather than some hardware or configuration thing though.
|
# ? Feb 13, 2016 00:19 |
|
Ekster posted:Just tried some overclocking with afterburner, still no dice. Is there a way to check the PCI motherboard thing? My motherboard bios (asrock b85m pro4) doesn't show anything about PCIe modes, but I'm not experienced with that kind of stuff. BIOS update perhaps? Not outside the realm of possibility that there could be a bug or something causing your issues. Otherwise it could be worth a try to have the shop test it out for problems. BIOS settings should have options somewhere for PCI-E where you can set the mode, usually Auto or 1.0/2.0/3.0 can be selected, could be worth a shot to see if setting it to the mode the card supports might help.
|
# ? Feb 13, 2016 02:26 |
|
Ozz81 posted:I think you can still use tools like GPU-Z to show you the temp of the VRMs at idle and load, it has a Sensors tab that shows what's monitored. If it's not the PSU or anything else the card might just be bad, might be worth swapping parts around if you've got a spare card to test with. GPU-Z doesn't show VRM temperature, nor does Afterburner. I think it's probably not exposed by the driver for this card. I guess I'll buy both a PSU and GPU and return whichever doesn't fix the problem. e: Power consumption scales linearly with frequency, correct? I underclocked the card to 510 Mhz and it still crashed around 70C. Does that exonerate the PSU? Yaoi Gagarin fucked around with this message at 05:20 on Feb 13, 2016 |
# ? Feb 13, 2016 04:53 |
|
VostokProgram posted:GPU-Z doesn't show VRM temperature, nor does Afterburner. I think it's probably not exposed by the driver for this card. I'm rather shocked you managed to get to 70C at 510MHz. In response to your question, power consumption generally scales superlinearly with frequency.
|
# ? Feb 13, 2016 05:57 |
Ozz81 posted:BIOS update perhaps? Not outside the realm of possibility that there could be a bug or something causing your issues. Otherwise it could be worth a try to have the shop test it out for problems. BIOS settings should have options somewhere for PCI-E where you can set the mode, usually Auto or 1.0/2.0/3.0 can be selected, could be worth a shot to see if setting it to the mode the card supports might help. I managed to find the BIOS settings (PCI-E speed link in my case) but unfortunately it didn't make a difference. My BIOS is at its latest version. I've tried everything I can think of so I'll return the card on monday, still not sure if I'm going to replace it with a different graphics card or if I'm just going to keep the money to upgrade my entire PC in one go in the future.
|
|
# ? Feb 13, 2016 07:51 |
|
I didn't see this anywhere in the thread but it's sort of interesting for AMD owners not looking to OC: http://www.tomshardware.com/reviews/msi-afterburner-undervolt-radeon-r9-fury,4425.html The results would seem to suggest that if you're not overclocking then undervolting a little lets PowerTune work a bit better and give you potentially more stable/efficient power use.
|
# ? Feb 13, 2016 11:46 |
|
Fabulousity posted:I didn't see this anywhere in the thread but it's sort of interesting for AMD owners not looking to OC: http://www.tomshardware.com/reviews/msi-afterburner-undervolt-radeon-r9-fury,4425.html The power efficiency gains GCN gets from undervolting paints an interesting picture of AMD setting voltage targets designed to make the bulk of production viable, rather then setting something more conservative that reduces yield but puts out even more impressive specs. I guess that's how they can potentially eat lower prices? Still, it should have opened up the door for better binned "E" products, as they don't seem to lose much in the way of performance from the undervolting but gain massively in power draw.
|
# ? Feb 13, 2016 13:02 |
|
I'm pretty sure the undervolting thing is already done for mobile bins. It also shows that Maxwell cards that aren't full size GM200 really aren't that efficient. Desktop GPU owners don't really care for power efficiency as much as they do overclocking and headroom though.
|
# ? Feb 13, 2016 15:38 |
|
Anime Schoolgirl posted:I'm pretty sure the undervolting thing is already done for mobile bins. It also shows that Maxwell cards that aren't full size GM200 really aren't that efficient. Yeah, Maxwell looks more like "this is literally the only thing we could do to improve performance on 28nm". But my point on potential "E" products (Like, a 390E, or 380E, etc) is that they also run much cooler as well, and underwater should clock higher. Just some basic testing, my 290X on air @ 1050/1250 with a drop in 50mv results 76° @ full load, while stock it's 95°. It's basically wasting energy away for no reason.
|
# ? Feb 13, 2016 17:54 |
|
AMD: If you manually fix our mistakes then it looks like Maxwell isn't so efficient after all! nvidia: what
|
# ? Feb 13, 2016 19:08 |
|
THE DOG HOUSE posted:AMD: If you manually fix our mistakes then it looks like Maxwell isn't so efficient after all! Again, I don't think it's a mistake so much as it's intentional to increase volume. Stricter voltage requirements means fewer chips to sell, and higher prices per chip.
|
# ? Feb 13, 2016 19:22 |
|
THE DOG HOUSE posted:AMD: If you manually fix our mistakes then it looks like Maxwell isn't so efficient after all! Gotta get those yields up. Still if you give Maxwell credit for overclocking you have to give GCN credit for undervolting.
|
# ? Feb 13, 2016 19:53 |
|
I know it's just the comments
|
# ? Feb 13, 2016 23:03 |
|
I got my new monitor hooked up and im having an issue. I cant get my secondary to be detected. I normally connect with DVI but it wont show up in my displays. It will show up HDMI, but its 2650x1600 seems to be too much for hdmi to display properly and there are black bars on the screen at the 2650x1400 I think was the resolution The primary monitor is display port. Before this I was using 2 monitors connected with dvi GPU is a g1 gaming 980ti, any solutions? Ok I fixed it somehow by plugging poo poo in and then reseting in various combinations. How it was fixed I have no idea. How do i know if gsync is on? Everything looks better and super smooth but i have no idea if thats the 144hz or gsync Fauxtool fucked around with this message at 08:35 on Feb 14, 2016 |
# ? Feb 14, 2016 07:58 |
|
Good inside look at how Digital Foundry measure performance: https://www.youtube.com/watch?v=fAVxmfNUuRs AMD's drivers don't come out looking too good. (Speaking of, why do the videos annihilate my CPU? Shouldn't my R9 270X GPU bear some of that brunt?) ufarn fucked around with this message at 11:56 on Feb 14, 2016 |
# ? Feb 14, 2016 11:52 |
|
what are my options when it comes to making 2 monitors look similar. Can a calibrator still help even if 1 is ips and the other TN?
|
# ? Feb 14, 2016 14:08 |
|
ufarn posted:(Speaking of, why do the videos annihilate my CPU? Shouldn't my R9 270X GPU bear some of that brunt?) Fauxtool posted:what are my options when it comes to making 2 monitors look similar. Can a calibrator still help even if 1 is ips and the other TN?
|
# ? Feb 14, 2016 15:10 |
|
Alereon posted:Yes, make sure hardware acceleration isn't disabled. For some reason that's been a popular tip for a few years, I can see they changed the default control centre, which is making a mess of things. Going to stick with "Default" as the Video Profile, since the monitors are calibrated, and change the Graphics Profile to "Image Quality". Y u do dis, AMD.
|
# ? Feb 14, 2016 17:18 |
|
I don't think youtube VP9 videos are hardware decoded unless you play it on a newer Intel integrated GPU. 1080p/60fps can be pretty taxing.
|
# ? Feb 14, 2016 17:47 |
|
but h.264 videos are hardware decodable by pretty much anything but just aren't in HTML5 mode for some browsers for some reason
|
# ? Feb 14, 2016 18:01 |
|
sauer kraut posted:I don't think youtube VP9 videos are hardware decoded unless you play it on a newer Intel integrated GPU. ufarn: Are you getting unexpectedly high CPU usage, GPU usage, or both?
|
# ? Feb 14, 2016 18:04 |
|
Alereon posted:Doesn't Youtube only use VP9 for fallback on systems without H.264? That video is H.264 @ 1080p60 on my machine, using HTML5 video playback in Firefox. I just notice my tabs refuse to load, until I select them and other things, and Chrome also just shows the CPU load. YouTube and Twitch in general. Maybe I've just noticed, since I got good Internet and a newer GFX and now run things in Source/1080p30/1080p60 when available.
|
# ? Feb 14, 2016 18:07 |
|
VP9 is the standard format, at least with Chrome. You can force H.264 with extensions like this one https://chrome.google.com/webstore/detail/h264ify/aleakchihdccplidncghkekgioiakgal Maybe try that and see it if makes a difference. If you watch a lot of twitch, download livestreamer and mpc-hc. Set the filter to DXVA2 (native) and enjoy 1080p/60 with 1% CPU load and not enough on the GPU to make the fans spin up. sauer kraut fucked around with this message at 18:16 on Feb 14, 2016 |
# ? Feb 14, 2016 18:12 |
|
Yeah, I use livestreamer for the most part, unless I'm lazy - which I've been lately - and it works pretty well. It's not easy with multiple streams running, though. I use it with VLC for reasons I can't remember, though. I already use h264ify, apparently. (I'll also enable DXVA in the madVR settings for MPC-HC for image scaling, which is the only place I can find it. Looks like DXVA2 is also enabled as the default video filter in MPC-HC.) ufarn fucked around with this message at 18:38 on Feb 14, 2016 |
# ? Feb 14, 2016 18:30 |
|
In the end the problem was with the GPU after all. 280X, you did good. ... Does anyone know a nice 970 OC guide they could link me to?
|
# ? Feb 14, 2016 22:21 |
|
VostokProgram posted:In the end the problem was with the GPU after all. 280X, you did good. Overclocking a 970 is stupidly easy, you just use MSI afterburner and bump up the core and memory clocks until you run into issues (usually the graphics driver will reset before you can do any damage) and you back the clocks off a bit until you're stable again. If you've got lots of thermal headroom to play with too you can even bump the voltage slightly to get more stability but I've never bothered.
|
# ? Feb 15, 2016 05:23 |
|
|
# ? Jun 5, 2024 05:21 |
|
VostokProgram posted:
Download MSi afterburner, push sliders forward and run Heaven. Look for artifacts and/or driver crashes. Rinse and repeat till you don't see any. The limits on the firmware of the card make it pretty hard to actually do anything risky beyond crashing the video driver without doing something excessive (for most people) liking flashing the BIOS. Don't push the sliders all the way up, that's generally WAY higher then most cards will go. Start at like 1/10 and go up in increments. My MSi 970 has been at +180 Core and +500 Memory for months now, to give you some idea. Mazz fucked around with this message at 06:35 on Feb 15, 2016 |
# ? Feb 15, 2016 06:31 |