3090 secret memory PerfCap

[Grim5]

What temps does your strix hit on the memory?

88c

 

[gpuerrilla]

88c

And when mining ETH for ~1hr?

 

[Grim5]

And when mining ETH for ~1hr?

never tried mining before, 88c is after a few hours of playing assassins' creed Valhalla

 

[deuse]

I've had a fan pointing at the back of my 3090 since day 1

The gpu sits at about 50C after gaming.

 

[Grim5]

I've had a fan pointing at the back of my 3090 since day 1

The gpu sits at about 50C after gaming.

is that the memory temp?

 

[deuse]

is that the memory temp?

No that's the GPU temp @2070hz
I have never found a way to monitor the memory.

If you tell me the software I will try it now.

 

[deuse]

Just checked.

I don't get a "Performance Limit- Thermal"
But I do get a "Performance Limit- power

 

[Grim5]

No that's the GPU temp @2070hz
I have never found a way to monitor the memory.

If you tell me the software I will try it now.

use latest version of HW info, its called Memory Junction Tempreture

 

[Saltank]

My GIGABYTE VISION 3080 OC starts throttling at 110C GPU Memory Junction Temperature. Meanwhile the GPU is 50C. I am reading that all GIGABYTE cards have these issues.

I've ordered some thermal pads and hope to fix this myself...

 

[deuse]

use latest version of HW info, its called Memory Junction Tempreture

It went up to 58.0 C
Is that no good?

 

[redkrptonite]

I get a prformance limit thermal yes when using realbench but memory junction temp only reaches 75c and gpu only reaches 52c but I don’t get it when I run timespy. Water cooled strix 3090. Not sure what’s happening.

 

[Jay-G25]

I get a prformance limit thermal yes when using realbench but memory junction temp only reaches 75c and gpu only reaches 52c but I don’t get it when I run timespy. Water cooled strix 3090. Not sure what’s happening.

Yeah I've noticed the same on warzone, watercooled 3090 Strix with gpu sitting at 51 deg and memory sitting at 72 deg but then check HWinfo and it says a thermal limit was triggered apparently

Leave it mining all night and memory will sit between 96 deg and 104 deg depending on how much the radiator fans are ramped up yet no thermal trigger .... I will also add due to EK backplate instructions being wrong i originally had 1.5mm thermal pads on the rear memory but turns out these weren't even touching and needed to be bigger ( 2mm+ ) as when i first tried mining memory would go straight to 110deg and throttle Since changing the pads the temps are better but still a touch hotter than i would like ... desperate for the watercooled backplate that is being released for the Strix... that should defiantly sort out the memory temps

 

[redkrptonite]

Yeah I've noticed the same on warzone, watercooled 3090 Strix with gpu sitting at 51 deg and memory sitting at 72 deg but then check HWinfo and it says a thermal limit was triggered apparently

Leave it mining all night and memory will sit between 96 deg and 104 deg depending on how much the radiator fans are ramped up yet no thermal trigger .... I will also add due to EK backplate instructions being wrong i originally had 1.5mm thermal pads on the rear memory but turns out these weren't even touching and needed to be bigger ( 2mm+ ) as when i first tried mining memory would go straight to 110deg and throttle Since changing the pads the temps are better but still a touch hotter than i would like ... desperate for the watercooled backplate that is being released for the Strix... that should defiantly sort out the memory temps

Where did you find out about the thermal pads being wrong? I had my ek strix backplates in one of the first shipments so I must have the wrong thermal pads on too.

 

[Jay-G25]

Where did you find out about the thermal pads being wrong? I had my ek strix backplates in one of the first shipments so I must have the wrong thermal pads on too.

I heard about the thermal pads potentially being wrong on the EKWB facebook group but when i fitted mine originally i must admit i didn't bother checking contact and assumed the instructions i used at the time were correct as they stated 1.5mm pads but now just been changed to 2.0mm ( https://www.ekwb.com/shop/EK-IM/EK-IM-3831109832479.pdf ) i assumed all was good as until recently we couldn't see memory temps on the strix but even then it wasn't till i started mining last week that i noticed an issue as the memory would go straight to 110deg even with stock memory clock so i pretty much knew the pads weren't touching the backplate ... pulled it apart last weekend and as expected not even touching 1 memory chip but was cooling an important screw

double stacked with an extra 1.0mm set of pads ( temporarily as was waiting for Minus 8 pads which have just turned up today ) and they do make contact and temperature has dropped as expected Would highly recommend you pull your apart and replace with 2mm pads .

 

[redkrptonite]

I heard about the thermal pads potentially being wrong on the EKWB facebook group but when i fitted mine originally i must admit i didn't bother checking contact and assumed the instructions i used at the time were correct as they stated 1.5mm pads but now just been changed to 2.0mm ( https://www.ekwb.com/shop/EK-IM/EK-IM-3831109832479.pdf ) i assumed all was good as until recently we couldn't see memory temps on the strix but even then it wasn't till i started mining last week that i noticed an issue as the memory would go straight to 110deg even with stock memory clock so i pretty much knew the pads weren't touching the backplate ... pulled it apart last weekend and as expected not even touching 1 memory chip but was cooling an important screw

double stacked with an extra 1.0mm set of pads ( temporarily as was waiting for Minus 8 pads which have just turned up today ) and they do make contact and temperature has dropped as expected Would highly recommend you pull your apart and replace with 2mm pads .

Just been doing some further testing. Ran an hour of heaven. Max junction temp of 66c. Thermal throttle thermal yes on both maximum and average. Even downclocking the vram by -502 still get thermal throttle. Only after lowering my core clock back to stock do the thermal limits remain a no. I think this thermal throttle limit for me at least is just the core clock lowering a step because of temps and nothing to do with my vram.

 

[redkrptonite]

I heard about the thermal pads potentially being wrong on the EKWB facebook group but when i fitted mine originally i must admit i didn't bother checking contact and assumed the instructions i used at the time were correct as they stated 1.5mm pads but now just been changed to 2.0mm ( https://www.ekwb.com/shop/EK-IM/EK-IM-3831109832479.pdf ) i assumed all was good as until recently we couldn't see memory temps on the strix but even then it wasn't till i started mining last week that i noticed an issue as the memory would go straight to 110deg even with stock memory clock so i pretty much knew the pads weren't touching the backplate ... pulled it apart last weekend and as expected not even touching 1 memory chip but was cooling an important screw

double stacked with an extra 1.0mm set of pads ( temporarily as was waiting for Minus 8 pads which have just turned up today ) and they do make contact and temperature has dropped as expected Would highly recommend you pull your apart and replace with 2mm pads .

Thanks I appreciate the information. I well can’t be bothered to take loop to pieces again. Ek should be sending out replacement thermal pads if that’s the case. Might contact them to see what they say.

 

[Greebo]

I haven't found any easy way. Maybe there is a way to generate error log files for VRAM, but if there is I have no idea how to do it.

The only reliable way we have is benchmarks. Set a base line score in 3d mark timespy with stock clocks - and then you can start increasing memory clocks speed, you'll know when ECC kicks in because the benchmark score will drop a lot. In 3d mark timespy I go from 21000 score at +200mhz on the memory to 15000 score if I go over that due to ECC.

Because of how ECC works, I believe it's beneficial for everyone to run timespy at least once with the card stock when they first get the card so I can set a baseline. Then in future if you suspect something is wrong with the card or you are just curious, you can run timespy again and compare against when you first got the card and if the score has dropped significantly then you know something is wrong. It remains to be seen if prolonged high tempreture on the GDD6X memory will cause the memory performance to degrade over time, but if it does degrade you will see ECC kick in and drop your performance so its a dead give away.

Here you go, tests the vram for error checking for you. Best to be running a benchmark in the background so the card and the memory is at in game temp.

https://easyupload.io/78ko7f

 

[redkrptonite]

Bit more testing. I only seem to get the thermal throttle limit yes when I’m running rtss. As soon as I close that the thermal limit issue goes away. Can anybody else test this or know what would cause this? Thanks Greebo.

 

[Spiritwolf]

Behold! my super getto cooling solution for vram junction of 3090 FE. Total cost 15pounds.

Heatsink best fit is at 100x100mm. It won’t thermal throttle now however still runs at around 100 degrees

 

[JohnG7]

Gelid Solutions GP Ultimate 15 W/mK Available in 0.5mm, 1mm, 1.5mm & 2mm thickness.

I bought 2 kits each of above as going to do my 1080TI FE and Titan Xp (2017) after I fit new fans (make annoying noise) then sell on EB, should make a good dent in the 3090FE purchase.