|
wolrah posted:Without a doubt. 4GB is rapidly becoming unusable for just browsing the web. 8GB is a bare minimum for gaming to not be a horrible experience, and 16GB is generally a good idea. I think you missed that it's also _DDR2_, I can't even remember what DDR2 was like.
|
#
?
Dec 9, 2016 08:00
|
|
|
|
| # ? Jun 5, 2024 06:14 |
|
|
FaustianQ posted:Even assuming a RAM bottleneck, the rig I'm struggling with should perform better than that, but that setup currently outperforms mine (with 15-28FPS) Good luck with this because some things are just inexplicable e.g. my R9 Nano, Core i5 and 8GB ram run Doom on Ultra at over 200fps at 1360x768 but for some bizarre reason I can only get a whopping 21fps in Warhammer Total War (which according to every review of my graphics card is meant trounce).
|
|
#
?
Dec 9, 2016 08:09
|
|
|
FaustianQ posted:Even assuming a RAM bottleneck, the rig I'm struggling with should perform better than that, but that setup currently outperforms mine (with 15-28FPS)
|
|
#
?
Dec 9, 2016 08:14
|
|
|
New Zealand can eat me posted:Using NV's installer as an example of good or common UX is laughable at best. We're not talking about UX design though. We're talking about what two words Clean & Install mean when used together during the installation of a piece of software since like 2-3 decades ago.
|
|
#
?
Dec 9, 2016 12:11
|
|
|
KingEup posted:Good luck with this because some things are just inexplicable e.g. my R9 Nano, Core i5 and 8GB ram run Doom on Ultra at over 200fps at 1360x768 but for some bizarre reason I can only get a whopping 21fps in Warhammer Total War (which according to every review of my graphics card is meant trounce). What's your CPU?
|
|
#
?
Dec 9, 2016 12:26
|
|
|
Maybe the Overwatch on HD 6900 problems are related to old drivers? AMD hasn't updated drivers for pre-GCN cards in well over a year. I had an HD 6870 that couldn't play Superhot worth poo poo, even when paired with an i5-750 at 2.8GHz, but a slower-on-paper GTX 460 ran the same game on the same system like a champ. Youtube clips showing Overwatch running on an HD 4850 suggest other issues may be at play in your case, since those cards are running on drivers that are just as old.
|
|
#
?
Dec 9, 2016 14:27
|
|
|
Is it likely nvidia will find ways to boost the 1060s (and other cards) performance significantly with drivers as amd has just done with their cards?
|
|
#
?
Dec 9, 2016 16:03
|
|
|
Just an update from a few pages ago. I had does a resest of windows 10 with a fresh update of all my drivers and thought I had resolved the problem; I was able to play a few games of titanfall without a hard rest. I then played a battle in total war, and then entire battle went fine but I got a hard reset as soon as I hit the "end battle" button. When the PC restarted the background was a solid black even though it looked like my desktop settings were unchanged, and I could still pull up browsers and other windows. This resolved with a restart. It happened again when I tried another battle, but this time when the PC had the hard reset I came back to my desktop, but the bottom 1/3 of the screen or so looked like the image was corrupted (it basically consisted of a small tiled geometric pattern) but this went away when the next picture on my desktop screenshot loaded. These latest crashes make me really suspicious that my motherboard is just failing. I know that my power supply is supposed to be a particularly reliable unit. Anybody know about this motherboard? I had picked it in my build because a similar model was recommended at tomshardware, but now I'm concerned about a failing mainboard risking other components in my PC My specs again: EVGA supernova 650P2 PSU GIgabyte z170 HD3P i5-6600K EVGA GTX 1070 2x8 gig corsair (i think) DDR4 1TB SSD, 1TB HDD which is storage for media (sorry to bring this tech support poo poo back to the thread, but I had gotten a lot of good suggestions here and basically nothing in the haus of tech support)
|
|
#
?
Dec 9, 2016 16:16
|
|
|
Well there's a chance you got one of the bad EVGA 1070s unless it's manufacture date is sometime in November. What happens when you use the onboard (motherboard) gpu. Have you tried swapping out the 1070 for a different card to see if this stuff keeps happening.
|
|
#
?
Dec 9, 2016 16:24
|
|
|
Surprise Giraffe posted:Is it likely nvidia will find ways to boost the 1060s (and other cards) performance significantly with drivers as amd has just done with their cards? Why would they? That'd mean they can't sell a 1060Ti
|
|
#
?
Dec 9, 2016 16:27
|
|
|
I could try swapping in my old radeon card. I hadn't suspected my GPU because outside of the crashes, performance is great. Temperatures on the card have been appropriate (usually maxing out at 74C) and I'm not getting any frame rate slowdown
|
|
#
?
Dec 9, 2016 16:29
|
|
|
Pikey posted:I could try swapping in my old radeon card. I hadn't suspected my GPU because outside of the crashes, performance is great. Temperatures on the card have been appropriate (usually maxing out at 74C) and I'm not getting any frame rate slowdown A few 1070s have had black screen/weird artifacting issues, etc due to some bad components and that being possibly exacerbated by no thermal pads on the VRMs (prior to Novemberish manufacture date cards). VRMs stay within acceptable ranges but usually what happened was a capacitor or something will blow and the card dies since there was some bad capacitors or something like that that were used in a few shipments to retailers. It's still a good thing to check before getting rid of the motherboard. Either by putting in a different card and trying to trigger the problem or just plugging your monitor/a second monitor into the onboard GPU while the weird stuff is happening on the main gpu if windows hasn't crashed. MagusDraco fucked around with this message at 16:38 on Dec 9, 2016 |
|
#
?
Dec 9, 2016 16:35
|
|
|
Surprise Giraffe posted:Is it likely nvidia will find ways to boost the 1060s (and other cards) performance significantly with drivers as amd has just done with their cards? Generally speaking NVidia's drivers are mostly optimized out of the gate. AMD's drivers for GCN were initially terrible for the amount of theoretical power they have. They've been focusing on removing driver overhead for the last year or two and it's been paying dividends, especially on lower end systems.
|
|
#
?
Dec 9, 2016 16:39
|
|
|
PBCrunch posted:Maybe the Overwatch on HD 6900 problems are related to old drivers? AMD hasn't updated drivers for pre-GCN cards in well over a year. I had an HD 6870 that couldn't play Superhot worth poo poo, even when paired with an i5-750 at 2.8GHz, but a slower-on-paper GTX 460 ran the same game on the same system like a champ. I thought so too but I swapped between the HD6950 and HD7850 in the same system and nothing really changed (using appropriate drivers of course). No it appears the real culprit may be old microcode that's disabling a bunch of instructions on the Xeon processors, because I guess why the gently caress not? Could this really bring everything to a screeching halt? EDIT: Apparently the reason HD4800 cards will play Overwatch but not other DX10 cards is they're DX10.1 and support Shader Model 4.1. This means in theory the entire HD3000 and HD4000 line up will work, if you want to play settings limbo. EmpyreanFlux fucked around with this message at 17:02 on Dec 9, 2016 |
|
#
?
Dec 9, 2016 16:55
|
|
|
I don't know much about how microcode updates work, but I can certainly envision a situation in which your LGA771 proc goes into your LGA775 board but doesn't receive all of the microcode updates that LGA775 procs would because the board doesn't recognize it. I have an old HP desktop that came with a 1st gen C2D and earlier this week, I tried to swap a 2nd gen C2D in just out of curiosity. It didn't even work until I updated to the newest BIOS, but even after it booted it ran all fans at 100% and threw an error about not having the right microcode. I had to reapply the standalone microcode update after swapping in the processor to get it working normally. You could try reapplying your newest BIOS update just in case since usually microcode is packaged in with those, but with a Xeon who knows if it will make a difference.
|
|
#
?
Dec 9, 2016 17:21
|
|
|
lmao. AMD's new driver enabled "MultiView is AMD’s equivalent of NVIDIA’s SMP (Simultaneous Multi-projection) which greatly enhances efficiency when rendering stereoscopic scenes" on radeon 4xx cards, and they're trying to work it into older ones too. Wasn't nvidia saying they had to make this work with new hardware on 1000 series and thus no older cards will get it? 
|
|
#
?
Dec 9, 2016 18:10
|
|
|
God I hope VR has improved at the driver level somehow, I got a launch day Vive and sent it back within a week because the framerates and graphics were such loving garbage. It's so surreal to be able to count (flat, unlit) polygons on fingers, haven't done that since the 90s. And that was with a 980 and 6700k at 4.8Ghz, everything just looked like utter dogshit. And it looks terrible in a way that videos don't really convey, even at a super high resolution it's hard to tell just how ugly everything actually is inside the headset at that awful VR resolution/framerate/graphics.
Pryor on Fire fucked around with this message at 18:19 on Dec 9, 2016 |
|
#
?
Dec 9, 2016 18:16
|
|
|
Truga posted:lmao.
|
|
#
?
Dec 9, 2016 18:19
|
|
|
Truga posted:lmao. I get the feeling that Sony is probably giving AMD a hand with VR optimizations.
|
|
#
?
Dec 9, 2016 18:50
|
|
|
SwissArmyDruid posted:I think you missed that it's also _DDR2_, I can't even remember what DDR2 was like. Low latency!!
|
|
#
?
Dec 9, 2016 19:32
|
|
|
FaustianQ posted:I thought so too but I swapped between the HD6950 and HD7850 in the same system and nothing really changed (using appropriate drivers of course). If it does turn out to be a microcode/CPU issue, I've got a used Q9400 CPU in an old rig that works and has been collecting dust. Even running in a mini ATX board that supports DDR3 (2 slots up to 16GB 1333) and it's already got 8GB in it. Was going to set up a small file server/game server deal with it but had some money issues after major surgery last year so it never got done  hit me up in PM if you're interested and I can work a deal with you if you want any of the parts
|
|
#
?
Dec 9, 2016 21:43
|
|
|
There is information about patching microcode updates into off-the-shelf BIOSes if you look hard enough and are willing to roll the dice.
|
|
#
?
Dec 9, 2016 23:39
|
|
|
EVGA GTX 1070 SC came in. So this is what high intensity coil whine sounds like 
|
|
#
?
Dec 9, 2016 23:53
|
|
|
Even though a lot of the information I can find is just speculation at this point, what's a general idea for which month the 1080 Ti is actually going to be released? All the articles I can find seem relatively sure they'll be showing it at CES in January, but how long does it generally take from the showing until it hits the market?
|
|
#
?
Dec 10, 2016 00:42
|
|
|
NoEyedSquareGuy posted:Even though a lot of the information I can find is just speculation at this point, what's a general idea for which month the 1080 Ti is actually going to be released? All the articles I can find seem relatively sure they'll be showing it at CES in January, but how long does it generally take from the showing until it hits the market? If they were going to pull a "you can buy it NOW" thing, we'd have been seeing leaks of finished boards by now. nVidia historically likes to release top-end stuff in April, for some reason.
|
|
#
?
Dec 10, 2016 00:45
|
|
|
PBCrunch posted:There is information about patching microcode updates into off-the-shelf BIOSes if you look hard enough and are willing to roll the dice. Just did this, and it's somehow working (Although all the Nvidia chipset trappings are gone in place of generic Intel and MSI ones, wth?). I couldn't even get Windows 10 to properly work before and everything recognized the processor as some kind of 32 bit monstrosity, but now it's working and I'm moving into the testing phase, BRB with a trip report. May still take you up on that offer yet Booty-Ade.
|
|
#
?
Dec 10, 2016 01:18
|
|
|
I have a pair of AMD FirePro W9100s lying around but I'm hearing they're poo poo for gaming and my VR devs can't get any use out of them right now. If I get them running crossfire do I have a chance at some decent game performance?
|
|
#
?
Dec 10, 2016 01:57
|
|
|
NoEyedSquareGuy posted:Even though a lot of the information I can find is just speculation at this point, what's a general idea for which month the 1080 Ti is actually going to be released? All the articles I can find seem relatively sure they'll be showing it at CES in January, but how long does it generally take from the showing until it hits the market? It certainly makes the most sense to announce it during their CES keynote. With availability ~1 month after Vega launches.
|
|
#
?
Dec 10, 2016 01:59
|
|
|
Dali Parton posted:EVGA GTX 1070 SC came in. So this is what high intensity coil whine sounds like
|
|
#
?
Dec 10, 2016 02:08
|
|
|
TheRationalRedditor posted:There are worse models for it. It's always a good time to switch to headphones though I'm looking up on how to "fix" coil whine, and it seems like running a benchmark tool for X number of hours or over/underclocking it helps? Some of these seem borderline wive's tales but the only other option seems to be RMA'ing it. But even then I guess it can randomly appear and go away?
|
|
#
?
Dec 10, 2016 02:56
|
|
|
betamax hipster posted:What's your CPU? i5 4570
|
|
#
?
Dec 10, 2016 03:05
|
|
|
Dali Parton posted:I'm looking up on how to "fix" coil whine, and it seems like running a benchmark tool for X number of hours or over/underclocking it helps? Some of these seem borderline wive's tales but the only other option seems to be RMA'ing it. But even then I guess it can randomly appear and go away? Charging and discharging an inductor causes it to physically jerk a little bit. Same principle as moving the particle along a railgun, just weaker. It's enough to move air, and that's all it takes. Like a bell, an inductor has its own natural frequency and will resonate best there. By moving a little bit off that natural frequency somehow, you keep it from resonating. That's what's happening when you're underclocking/overclocking, limiting the framerates, etc. Your framerate or clockrate will affect how often that coil is charged and discharged and you can get it far enough off its natural frequency that it's not resonating. Running it for X hours sounds like BS to me though. I don't understand how that would change any of the factors involved in the resonance here. Maybe it's enough to knock the winings loose on the toroid? I'm not seeing that as a plus personally. The best way to fix it permanently is to track down which coil is resonating so badly and throw some hot glue on that fucker. But underclocking/overclocking a little bit is a lot easier. Paul MaudDib fucked around with this message at 04:25 on Dec 10, 2016 |
|
#
?
Dec 10, 2016 03:06
|
|
|
Killer Low Life posted:I have a pair of AMD FirePro W9100s lying around but I'm hearing they're poo poo for gaming and my VR devs can't get any use out of them right now. If I get them running crossfire do I have a chance at some decent game performance? Sell them. They'll never be useful for VR if you're not using them already. You're sitting on a few thousand dollars. I presented on the cost/benefit of getting those vs sticking with our old Teslas and it wasn't worth the additional developer time to rewrite their code from Cuda to the more generic Firepro language even with the raw performance and memory size benefits. You could probably sell both of them and get 2 Titan XPs which would be much more useful for most VR stuff and still come out with money left over. They're only really useful if you have a development environment that either is brand new so everyone can start from nothing, or is already working as a non Cuda environment.
|
|
#
?
Dec 10, 2016 03:49
|
|
|
Paul MaudDib posted:Charging and discharging an inductor causes it to physically jerk a little bit. Same principle as moving the particle along a railgun, just weaker. It's enough to move air, and that's all it takes. I shouldn't need to futz with voltage right? The 1070 SC is already factory overclocked and games run perfect with it, so I could start by underclocking it a little?
|
|
#
?
Dec 10, 2016 04:04
|
|
|
Truga posted:AMD's new driver enabled "MultiView is AMD’s equivalent of NVIDIA’s SMP (Simultaneous Multi-projection) which greatly enhances efficiency when rendering stereoscopic scenes" on radeon 4xx cards, and they're trying to work it into older ones too. You can implement SMP with standard geometry shaders. NV makes no secret of this, their own multi-projection sample app has a GS fallback that works on any DX11 card. Pascals "secret sauce" is that it can handle SMP in dedicated hardware and avoid the overhead of invoking the GS, which NV claims is pretty substantial. Perhaps AMDs more shader-endowed cards can get away with just brute-forcing it on the GS though. repiv fucked around with this message at 05:33 on Dec 10, 2016 |
|
#
?
Dec 10, 2016 04:09
|
|
|
Dali Parton posted:I shouldn't need to futz with voltage right? The 1070 SC is already factory overclocked and games run perfect with it, so I could start by underclocking it a little? I'd go for overclocking first for free performance, but up to you if it already performs well for you. Videocard overclocking is exceptionally safe these days since it's largely limited by sensors measuring voltages and currents and forcing a reset if something starts to seem wrong and less about hitting the mechanical limit of performance. I agree the "running for so many hours to fix it" is BS. One extra kink in fixing it, often you can't hot glue inductors on videocards because they're in cases. Poke all of them with a pencil eraser and you'll never get the ringing to stop because the inductor is ringing internally and not connected to the case. I wouldn't mess with the voltage slider, it's predefined so it is safe, but I haven't found pascals to benefit much from the extra voltage. They do get a lot hotter from it though, and get only maybe 13 more MHz over being left alone. I'd start out maxing out the power and temperature targets because that's not even overclocking, just sliding up the clock curve that's already flashed to it. As long as you have the case airflow and power supply to drive it, it's only increasing the TDP and should never cause a problem.
|
|
#
?
Dec 10, 2016 04:17
|
|
|
Took awhile, but while I do get noticeably better results I've basically come to the conclusion that TheRationalRedditor posted, it's entirely a memory bottleneck because it's not only DDR2, but MSI in their infinite wisdom made the board single channel. Incoming PM Booty-Ade.
|
|
#
?
Dec 10, 2016 04:20
|
|
|
If you have bad coil whine on a 400$ EVGA card just rma it, I know it's a hassle.
|
|
#
?
Dec 10, 2016 05:03
|
|
|
FaustianQ posted:Took awhile, but while I do get noticeably better results I've basically come to the conclusion that TheRationalRedditor posted, it's entirely a memory bottleneck because it's not only DDR2, but MSI in their infinite wisdom made the board single channel. Incoming PM Booty-Ade. Jesus H. Christ on a pogo stick with Mary and Judas. 
|
|
#
?
Dec 10, 2016 05:25
|
|
|
|
| # ? Jun 5, 2024 06:14 |
|
|
sauer kraut posted:If you have bad coil whine on a 400$ EVGA card just rma it, I know it's a hassle. Problem is I've also done 2 video card RMAs (for other reasons) in a row in the past two weeks. I'm just sorta tired of it. It looks like EVGA offers advanced RMA (where they ship the replacement to you first) but on top of wanting a credit card hold, they also want minimum 30 dollars for the service 
|
|
#
?
Dec 10, 2016 05:45
|
|