fish don't fry in the kitchen so why can't nvidia fix this?
>https://forums.geforce.com/default/topic/939358/geforce-1000-series/gtx-1080-flickering-issue/26/
>26 pages
>390 posts
>first post was 06/02/2016 12:32 PM
>affects both 1070 and 1080 users
>Hello all,
>We are still not having any luck reproducing any flickering with the latest drivers. We could really use the following information from people still seeing this problem with driver 368.81 on the GeForce GTx 1080/1070 (make sure you reboot if you just updated to this driver).
> Posted 07/15/2016 06:46 PM
rofl really nvidia. over 390 pages full of 1070 and 1080 users complaining about it and some how you don't have a single computer in your billion dollar hq that can't reproduce it?
I have the issue on my 7970, but I understand the issue, its related to memory changing frequencies rapidly. and I've a solution for it, at the cost of idle power consumption (and temps). if you have a single monitor generally it isn't an issue, however, to understand it better, see pic related for the sake of my sanity. you need to understand p-states a little bit, but I'll try my best to explain.
I have a 7970, its max p#0 is 1040/1400, if I enable hw acceleration and watch a video, it'll switch to p#6 rapidly, stock p#6 on my card is 1040/1375 (lower mem!) with this I'll encounter this flickering the monitor "jumps" and then settles after a frame or two, but its very noticeable.
in the past for older cards (4890, 6970) the fix was to disable ULPS, and thus prevent the p#4 state, so p#5 was/is my idle, that would work if I had a stock 7970 (or use the bios switch) but it won't work on my cards default bios) because the memory will still jump between 1375 and 1400 in applications. (and don't even get me started on overclocking memory, since p#6 will override your shit unless you force p#0 ALL the time).
my solution currently, in the recent catalyst drivers (for the past 8 or so months) disabling ULPS via registry or in MSI Afterburner doesn't do anything, I presume this is because AMD stopped caring if you have multiple monitors (on 60Hz), since 70+ or higher automatically disables p#4, was to overclock both of my monitors to 72Hz (good anyway since I don't drop frames in movies anymore). now I idle with p#5, the other part of my fix involved flashing the bios and making p#0 memory speed that of p#6, since the radeon bios editor doesn't support edited p#6 for some reason.
fuckin lovely, having to waste so much time with this nonsense, so no, it isn't an nvidia specific issue, its a GDDR5(?X)-is-fucking-garbage issue
>>55626700
Can you explain how to fix thix to me? Keep in mind im retarded but this shit caused me to trade a friend my 290 for his 2nd 7970ghz only because my shit crashes less often with the 7970. Hed be happy to return it and I just want to understand how this works so I can try and un-downgrade myself.
>>55626224
My friend who has 1080 doesn't get it either, it's probably some random fringe issue that's caused by certain unorthodox hardware choices.
>Nvidia
>fixing issues
if anything its their job to pretend its not an issue, and all the dumb nvidyakydz eat it up
>>55627225
make sure all p-states have the same memory clock, if your memory shifts between like 1000 / 1375 or so, or even 1375 to 1400 like in my case, it'll cause the flickering.
either use software to force the clocks u want at all times, or edit the bios and then flash the card
>>55627433
Over 300 unorthodox hardware choices. Okay.
>>55626700
for whatever reason my 2 r9 390's ive had have an issue with one of the displayport slots where the screen flickers/stays black for a long time until the video ram is downclocked to about 1250 (1500 stock). if you go above 1300 it starts flickering a lot and 1350 is just black most of the time
the other slots work fine
i can run it with the 1500 memory but i have to lower res and refresh rate to way lower (monitor is 1080p 144hz)
im guessing this is a gddr5 issue? i rma'd one card because of it but then i realized the other slots worked well.
>>55629183
>Number of replies means that number of people posted in the thread
No - also hardware isn't the only thing that could be affecting it.
>>55629531
yeah there is some weird issues with displayport, though I use the displayport to mini active adapters which don't have that 1 extra pin for power, so I've sidestepped that issue. though it does sound like something that maybe RMA worthy, but if downclocking the memory works for you whatever, it isn't a big deal, memory speed doesn't do much for performance from what I've seen.
I hate dealing with that shit personally, at least with some online shops they can do advanced RMA where they mail you a new one and you put the old one back in the box and put their sticker on it.
>>55629641
>https://forums.geforce.com/default/topic/939358/geforce-1000-series/gtx-1080-flickering-issue/26/
didn't they release a driver a week ago to fix that?
Pls give me some fried fish
