Question AMD RX 570 OC

Nknpon

Prominent
Aug 5, 2019
30
2
535
hello folks,
i temporarily got this RX 570 https://www.mindfactory.de/product_...ulse-Aktiv-PCIe-3-0-x16--Retail-_1295291.html
I pushed it to 1500mhz@1150mV and 2250mhz@935mV - firestrike was runnig perfectly fine but i got a green screen after playing 1hour PUBG.
Now its stable on 1440@1140 and 2200@915 even on long sessions. I think i could push it even more but i like it quiet and still didnt change anything on the fans.
Temp was like 85C befor, now its on 75 max. But since im pretty new to this overclocking thing and specially to AMD cards i need to ask if this could cause me difficulties in the long run?
i thought i could burn my whole PC before watching this video, where a guy said: worst case you gonna get a black/greenscreen or a crash. Is this actually true? Whats possible the worst case that could happen by overclocking in general? Here my build if u want to know https://www.userbenchmark.com/UserRun/19005301
 

Nknpon

Prominent
Aug 5, 2019
30
2
535
Thanks guys,

yes i am using wattman and i really like it so far. Also tryed MSI afterburner before with my gtx 770.
But using afterburner with this ADM card kinda messes up my fan controll i feel like. Sometimes the GPU Fans werent runnig at all, so i deinstallend it and now it works fine.

like u said @Karadjgne, i was able to push it a bit more, but dont relly got any performance increase.

i got some more questions about overclocking the memory
I use 3Dmark Firestrike to test oc settings. Does a better Firestrike score always mean more performance ingame? I have heard many times that overclocking the memory isnt worth, but i acutally got a huge Firestrike boost by doing so. (from like 11300pts to 12700pts by pushing it from default 1750mhz@900mV to max 2250mhz@935mV.)
And it definitely pushed my FPS for Firestrike.
But someone told me u only get higher timings by ocing the memory and u dont want higher timings on gaming - Whatever that means, lol.
 

Karadjgne

Titan
Ambassador
It's a personal thing and specific to the games played. If you can't tell the difference in game, then while you have a better benchmark, it doesn't mean a thing. A benchmark is a tool, it's used for 2 reasons. 1) to tell you that there's been an improvement in the running of the card and 2) pto compare your card to others who have done the same.

But your changes are working on a nanosecond scale, your monitor will have limits, the cpu will impose fps limits, resolution will impose visible detail limits, your eyesight has limits. And your perception is only so tactile. So if you can't see the changes, then there's no difference.

Think about a 60Hz monitor. Best possible visible viewing is 60fps. If the fps output from the gpu is 80 minimum, then all the OC in the world to get that to 120 minimum won't mean a thing, because you only get 60. If that OC allows for a bump in detail levels while remaining at 80, then bonus, that's tangible results, but if that 80 is already at ultra, then the power to push 120 won't do anything for you at all visibly.

And that can change with a different game and different settings.
 
  • Like
Reactions: Nknpon

Nknpon

Prominent
Aug 5, 2019
30
2
535
Thanks again i really appreciate the answers,

i instantly got a greenscreen by turning it on "memory timing level 2" even with 0% load. I guess 2250@935 seems too high with this setting.^^
I turned it down again to 2200@915 and now it works fine. At the same time i undervolted the GPU a bit and turned down the clockspeed from 1440@1150 to 1400@1130. Yesterday i played like 2 hours pugb with stable 120fps on 71-74 degrees. Even tho i turned down the OC a bit i got an FPS increase between 7-15 by changing it to timing level 2 - thanks for that.

I could push it a bit more again and try how high i can get with the memory timing level 2 but i still got a 60fps monitor so there is no point overclocking it to the limit and its probably wastet time - thanks for the advice @Karadjgne
Plan was to get a new monitor when upgrading to 2080ti.
 
The memory gets the same voltage that the core is requesting. The memory voltage setting simply limits the card voltage so it doesn't dip too low for the memory to be stable.

With those timings (2200 from 1750) I'm not surprised that memory timing level 2 didn't work. Ultimately you're at the ragged edge of your memory frequency, so timing improvements would require frequency sacrifices. You'd have to test to see which variable improves your performance more.
I presume you tried level 1 also?
 
  • Like
Reactions: Nknpon

Nknpon

Prominent
Aug 5, 2019
30
2
535
I didnt test it with the same settings @memory timing level 1. I just assumed its better, because i changed it to timing level 2 and turned everthing else down a bit to make it run but still got an nice improvement.
Before changing it to level 2, everytime i turned up the oc i ran Firestrike tests, to make sure i wont lose performance.
But i didnt test the memory timing level 1 ingame with same settings, so u re right and i probably should do so today.