overclocking

[Monsieur Tirets]

damn it, i dont know whats up. i lowered the cpu clock and its still acting up. even outside of the game my mouse pointer is stuttering all over the place.

i dont know what the issue is, its sporadic and i ran a quick stress test and cinebench run as well as a gpu benchmark and everything is in line, so idk...

 

[Crystallas]

I only OC when I need it. otherwise I don't bother. I used to OC a lot in the 90s, but that's because you had improvements that were so big every 8 months, that you didn't care so much if something fried after a year. I had a SX2 486 33MHz at 105MHz for 3 years running before replacing that system, and it was fully functional even in the end. My final self-hosted bbs server before using a shell(early cloud computing LMAO), and I was nervous as hell it would die on me before having a backup plan.

Have fun experimenting. You'll go through the same path as most, OC now, then later come to the conclusion that it's smarter to save the difference in cost to do a proper OC, and just upgrade sooner.

 

[botfly10]

damn it, i dont know whats up. i lowered the cpu clock and its still acting up. even outside of the game my mouse pointer is stuttering all over the place.

i dont know what the issue is, its sporadic and i ran a quick stress test and cinebench run as well as a gpu benchmark and everything is in line, so idk...

If your GPU OC is unstable in a game, it can cause problems with your gpu driver, almost like a soft crash, sometimes without windows ever notifying you. Those problems can remain after the game is closed until you reboot.

Next time you have issues after you close the game, go to device manager > display adapter > find your gpu > r click your gpu > click disable > click ok > r click your gpu again > click enable

That will reload your gpu driver. If it solves problems with a laggy desktop, you will know that your OC caused a problem with the driver while the game was stressing it.

Laggy desktop suggests problems with the gpu and drivers, not really cpu OC.

 

[Monsieur Tirets]

yeah, i meant gpu not cpu.

i restarted the pc after it happened and desktop was still laggy, but then i shut the pc down completely and that seemed to fix things.

but now i know to try disabling the device and then enabling it if it happens again.

edit: well, while the desktop stuttering seems to be fixed, the witcher 3 isnt running as smoothly as it was. its stuttering and artifacting with the 1110mhz oc on the gpu even though i played for hours at the setting earlier with no problems, and even dropping it to 1100 it still isnt running smooth. i dont get it.

 

[botfly10]

Does it ever artifact with no OC?

 

[Monsieur Tirets]

not that ive noticed, but it doesnt run nearly as well as it was both before or after the oc. its like what ever caused the game to bug out last night has had a lasting effect, everything was good before that. ive even tried uninstalling and reinstalling the drivers. and just to see if it was the hardware i ran some more benchmarks to test whether or not the gpu is performing as it should, and it is, at least in benchmarks.

 

[botfly10]

It could be cause the part of the game you are in too. Like there could be a lot more textures, shadows, models, or whatever. Could be extra taxing on the card, idk. Shadows especially can vary a lot and cause inconsistent performance. I almost never turn them all the way up, even on my 980 ti

 

[Monsieur Tirets]

yeah, its happening in places the game used to run completely smooth as well, so idk. seems the gpu core clock isnt running at full speed in game either despite temps under 70 degrees. the games still playable, but it sucks that it went from running almost flawlessly to this.

 

[Monsieur Tirets]

hey, bot.... is it possible for a game to suddenly decide it doesnt like your gpu being overclocked? because ive set my gpu back to stock settings and the witcher 3 seems to be running better. its weird, it ran great with the oc, than bugged out and ran like shit, but now with stock settings is running smoother.

 

[botfly10]

idk man, that sounds weird.

I will say that with my 780s, if I OC'd at all, I would get freeze and crash during cut scenes. If you google that specific issue with cut scenes in W3, a lot of people had the problem and most were solved by lowering their clocks. Also, W3 and Tomb Raider were the first games to ever give me problems like this.

I know in my case, with those games, I was pushing my cards to their very max. And GPUs have a lot of different systems running on them. Some games can push different parts of the GPU hard. But my feeling is that W3 and Tomb Raider both push most or all of the systems on a GPU really hard. I think that is somewhat uncommon and can maybe cause problems when you OC cause secondary systems are already at their threshold. But I could just be talking out of my ass.

But at the end of the day, there comes a point where you sorta figure out what your hardware is capable of with a particular game and kinda have to just accept it and start turning shit down until you get a fps rate you can accept.

Also, your card's drivers are a whole other variable in the mix. To some extent, users are at the mercy of GPU manufacturers working with game devs to come it with good drivers and it doesn't always happen. And users are at the mercy of manufacturers continuing to tweak drivers as bugs are found. That track record isn't always great for both Nvidia and AMD. I know that when Crimson came out a lot of people were pissed off.

 

[Monsieur Tirets]

i was wrong, its still happening. my concern is whether or not its hardware or software related. and how to go about determining that. im guessing software, if only because all benchmarks fall exactly in line with were my hardware should, so...

 

[Crystallas]

I would try to RMA the card, you could have defective memory. I have had 4 GPUs in the last 12 years with memory issues that required RMA. Two nvidia, two ati. 6 if I count that I had to RMA an evga GeForce 6800 3 times before I just got a 7800 upgrade from them for my trouble. LOL I also buy a lot more hardware than most people, so this is still a somewhat small percent of total. For whatever reason, memory modules on higher end GPUs wind up being the weakest components.

 

[Ares]

I would try to RMA the card, you could have defective memory. I have had 4 GPUs in the last 12 years with memory issues that required RMA. Two nvidia, two ati. 6 if I count that I had to RMA an evga GeForce 6800 3 times before I just got a 7800 upgrade from them for my trouble. LOL I also buy a lot more hardware than most people, so this is still a somewhat small percent of total. For whatever reason, memory modules on higher end GPUs wind up being the weakest components.

He has 2 of them in SLI IIRC

 

[Crystallas]

He has 2 of them in SLI IIRC

I thought he had a single R9 390, anyways, AMD/ATI doesn't have SLI, but crossfire, so he could test the cards individually without pulling one out at a time.

 

[Ares]

I thought he had a single R9 390, anyways, AMD/ATI doesn't have SLI, but crossfire, so he could test the cards individually without pulling one out at a time.

Oh yeah crossfire vs SLI I forget.... I only ever buy Nvidia.

I thought he had 2 cards going, maybe I am wrong.

 

[Crystallas]

Oh yeah crossfire vs SLI I forget.... I only ever buy Nvidia.

I thought he had 2 cards going, maybe I am wrong.

Yeah, I'm surprised he didn't get an nvidia card too. Radeons are for more advanced applications while still being okay at gaming, while GeForces are for more advanced gaming, but still okay with advanced applications. Unless MT also does high throughput OpenCL engineering and drafting design, but just talks about gaming here.

 

[Monsieur Tirets]

i have one r9 390. and from what ive seen the r9 390 is the best gaming card on the market in the mid to low-high end segment. everything ive seen suggests it a better card than the gtx 970. hell, the gtx 970 is already being hindered by 4g of memory, only 3.5 at full speed to be exact, in some games.

the 390 is the better value imo. and ive never heard about the current amd cards being geared towards high end apps rather than gaming. everything i researched was based on gaming performance vs other cards in the segment. now, if i could have afforded a 980 ti, sure nvidia would have been the way to go, but for 300 bucks i saw no reason to go nvidia.

anyway, like ive said ive ran a number of benchmarks and the gpu seems to be functioning just fine. i know synthetic vs gaming is not the same but if it was the hardware wouldnt that show up in the benches? everything is right in line with where this card should be performing and gpu load is at 100% with clock and memory clocks functioning at full speed. and tomb raider runs at 60fps still except in the valley area where it seems to run shitty for most people. is it possible that when the witcher 3 bugged out the other night something in its files got corrupted?

 

[Crystallas]

Well, I wasn't trying to get at what is good or bad or fanboying(which is for idiots anyways). I'm running a Fury X. But what I was getting at, was the possibility of a bad memory module. If you're not having issues in most games, but narrow down the problem to just one, then I doubt overclocking did anything to it. And yes, AMD GPUs perform considerably better with OpenCL applications, to the point where fanboys who get into high demanding OpenCL applications wind up switching. It's a back and fourth game, to the point where the fanboyism BS is mind boggling and any smart consumer over a 10 year period should have owned both and not just the worst versions of one and the best of the other so they could confirm their bias.

 

[Monsieur Tirets]

well, id never buy based on brand, but rather which gpu offers the best value, and in this case that was the 390 for me. ive never owned nvida simply because this is the first time i built a pc and actually looked into components.

anyway, is there a way to test for certain whether or not the memory is functioning properly other than swapping in another card?

 

[botfly10]

I would try to RMA the card, you could have defective memory. I have had 4 GPUs in the last 12 years with memory issues that required RMA. Two nvidia, two ati. 6 if I count that I had to RMA an evga GeForce 6800 3 times before I just got a 7800 upgrade from them for my trouble. LOL I also buy a lot more hardware than most people, so this is still a somewhat small percent of total. For whatever reason, memory modules on higher end GPUs wind up being the weakest components.

Yeah, I have kinda been thinking about a RMA too.