MSI afterburner giving unstable ocs

[DoomSlayer]

I found something rather strange... On my laptop which has a 1660ti, if I overclocked the core by 205 which would be around 2085 max boost, and would boost to around 2055-2070 in games - but... some games wouldn't be fully stable with the 205 overclock, and msi changes the clock curve randomly too... Max clock speed would jump form 2085 to 2100, and sometimes it'll drop to 2055?????? And this is with a manual overclock of 205+ on the core. Why is it changing the clock curve so much?

I decided to try out EVGA Precision X1 and these issues were fixed.

Max clock speed/boost speed is 2134mhz now (215 on the core now), and this time, EVGA didn't mess with the clock curve unllike MSI afterburner. Just played some RE2, maxed out settings etc, and it was completely stable, never dropped below 2000mhz, and would constantly boost to 2100. And power throttling for some reason, didn't happen as often.

This NEVER happened when I was using MSI afterburner. The GPU would constantly throttle with MSI afterburner, it would never stay at 2100 or 2080mhz under load. I would constantly see the clock speed dropping to low 2000mhzs and sometimes 1900s. EVGA kept it above 2050 90% of the time.

I don't know if what I said makes any sense, but what I mean is EVGA is giving me better, higher and more stable clock speeds. From 15mins ingame time, it was completely stable. MSI would crash within 15 mins if I had these clock speeds set in MSI afterburner.

Strange. Very strange... Never seen or had this happen to me before... Don't know how one software can differ from the other when its doing the same thing - overclocking...

[tacoshy]

well you do an OC of the normal clockrate. However the GPU will boost the GPu also addittionally according to your thermals. And there many different task/instruction a chip has to handle. if you face an intruction that is unstable it will crash. not all games will trigger this thats why stress testing is important to find out if it is stable in every case not just in most.

[Bad 💀 Motha]

Like any OC not all games will run faster while remaining stable. Some won't like an OC at all, like GTA4 for example.

Just leave the clocks on defaults and raise the power boost. The gpu will auto OC depending on loads and temps.

[DoomSlayer]

Yeah, I did stress test it- sort of- by using EVGAs 'test' feature? Score was "3" what ever that means, I quickly looked it up and couldn't find anything without digging around for while.

https://imgur.com/a/0cG3G2N (Screen shot of my settings)

Max boost is 2134mhz at 1100mv... My laptops GPU will run at 1065mv which would be around 2119mhz, at least thats what EVGA is telling me on the clock curve chart (or VF curve chart, once again no clue what VF is...)

MSI afterburner would have a max boost clock of 2100mhz, anything higher than that games would start to crash within 5 mins, and that isn't always at 2100mhz. It rarely hits 2100mhz, it more so stays around 2000 to 2050 sometimes dipping into the high 1900s with MSI afterburner yet THAT is somehow unstable with MSI which got annoying very quickly...

But with EVGA, somehow, gives me the ability to OC much higher than MSI while being way more stable otherwise. This time I would constantly see the GPU hitting the 2100mhz mark, if I were to switch software and use the exact same settings my game would crash, and or my card wouldn't run at the frequency EVGA gives it.

I am beyond confused now.... Ima try out some games just to see if I can get it to crash...

[Bad 💀 Motha]

Set msi Afterburner up better in its settings and try again, set to Kernel mode and tick all the unlocks.

[[☥] - CJ -]

Dont OC Core too much
Focus more on Memory less on Core
You'll tend to run into more problems when OC'ing the core too high compared to Memory
Not to mention you'll get more performance out of memory than the core anyway

[DoomSlayer]

Somehow EVGA allowed me to raise the voltage on the GPU... It never went beyond 1056, now it hit 1069... 20mv increase, this wasn't possible with MSI afterburner, the voltage bar did nothing.
https://imgur.com/a/qhwp2OA
Power target is still locked which is annoying, because that's the only thing holding back this card, is the stupid 80watt tdp.

[SeriousCCIE]

EVGA, in my experience, has good hardware but external limitations. I actually flashed my EVGA cards to have ASUS roms.

It's such that I now have to actually think about what the card actually is, because any benchmarks or system info etc all reports it was manufactured by somebody else.

I say this because I had to do so in order to raise my voltage cap. I don't know what options, if any, may avail themselves to you, but I found bios/firmware that allowed me to raise my voltages until it didn't work right (meaning--OC'd as far as I could reliably take it with hours of testing at the final speed I could reach) that was otherwise unachievable with the evga bios--with or without their weird precision stuff or msi afterburner.

As far as the precision tool is concerned, I thought that it stunk as a tool, and worse, the box says it comes with it and yet it required me to create a login and fork over an email address to get it? And I had to download it after logging in and confirming I'm me, rather than install off a CD? and it stunk anyway? screw that. not going to do that again.

msi afterburner didnt ask any of that and worked better for me.

[Bad 💀 Motha]

If ud read my post #4, that explains why msi AB had no voltage control. Enable it.

Also change the skin to a proper one that shows all the controls for newer gpus.

[Elon Mosque]

I can confirm there is something up with the custom curve in Afterburner right now. I have set 0.975mv and onwards at 2055Mhz. Which works flawlessly but sometimes it crashes and when I check the curve. Curve actually went up to 2100Mhz instead by itself and therefore causing instability. I will try the Precision though...

[DoomSlayer]

Bad 💀 Motha の投稿を引用：

If ud read my post #4, that explains why msi AB had no voltage control. Enable it.

Also change the skin to a proper one that shows all the controls for newer gpus.

Already did, voltage, power and temp is locked on my laptop (raising the voltage in Precision seemed to have a affect though, but power and temp is still locked). I need a modded vbios to flash to increase the TDP/temp limit, but theres nothing out there for the 1660ti...

Agnes Washington の投稿を引用：

I can confirm there is something up with the custom curve in Afterburner right now. I have set 0.975mv and onwards at 2055Mhz. Which works flawlessly but sometimes it crashes and when I check the curve. Curve actually went up to 2100Mhz instead by itself and therefore causing instability. I will try the Precision though...

Yeah, if I set my core offset to 200, sometimes the curve would max out at 2055. And then if I reapply the same offset, it'll jump up to 2085 maybe 2100 whatever it feels like doing...

Going to Precision fixed that issue, 215 offset is giving me a max curve of 2133 and hasn't change yet. I didn't have to enter my email or sign up to download the program either. Only that sucks is the OSD is hard to use and doesn't automatically lock onto a game, you have to manually add the games .exe file in order for the OSD to work.

[Bad 💀 Motha]

If it's a laptop, just leave the gpu alone. You really won't gain much in the end besides making it run hotter.

[DoomSlayer]

Bad 💀 Motha の投稿を引用：

If it's a laptop, just leave the gpu alone. You really won't gain much in the end besides making it run hotter.

I never broke over 65c on the GPU & CPU. If I was so concerned about temp I wouldn't be messing around with a brand new laptop. I had the ability to overclock the RAM as well, so I did that, a nice increase in RAM performance and CPU performance. 2666mhz now runs at 2932mhz. No increase in dimm temps, 45-50c under load. PCH temps are around 50-55C

[Bad 💀 Motha]

Ram generally doesn't get hot anyways, don't have to worry about your ram.

[SeriousCCIE]

I've found many people don't worry much about ram, but my experience has been there's a lot less 0x000000c5 and other 7a, d1, and other similar errors regarding ram or generically writing to disk from unstable ram... if the ram is cooled if overvolted or otherwise overclocked in some manner.*

If a box of ram comes with a ram fan, I'll use it. If there are heat spreaders on them, I try not to remove them to make something else fit.

I am the type of clown that accidentally peeled off an IC from a stick of ram while attempting to take off the thermal padding with a heat gun. I got lazy on about stick 7 since things took so long. Despite that INCREDIBLY EXPENSIVE mishap (I kept it as a reminder of how to be stupid) as a result of my getting impatient -- I'd still recommend for the enthusiast wanting more oomph or less crashes as it oomphs.. to not neglect cooling the ram.

It is not helpful that there is rarely a heat sensor on the ram or anywhere near it, and so most people really have no idea how warm their ram gets anyway, since most monitoring programs only report the type, speeds, or size of it, or how much free space is left, but few PCs have sensors near the ram to give an idea of if the stuff is baking in their shells or if they're at a similar temperature to the rest of the case averages.

*i tend to fill all the ram slots; that adds stress to the system as far as overclocking potential goes, and often an OC can be pushed farther for a CPU if there is less ram. I have to work around the other issues caused by a full complement of ram, so my experiences may not be similar to most people here.