GPU with low performance under watercooling

[axman.com]

I bought a kraken g12 to use a rog strix lc 240 with an evga rtx 2080 super KO but now i have low framerate on minecraft

why?, I mean, before the upgrade i had stable fps at 60 or so.

Sometimes I also have a random but briefly black out in my screen

 

[Mark Kaine]

43 minutes ago, axman.com said:

stable

 

43 minutes ago, axman.com said:

or so

Oxymoron detected. 

 

So either way what are the fps now? 

 

Certainly strange, temps seen fine should be easy to reach 60fps in what looks like an extreme low requirement game. 

[-rascal-]

58 minutes ago, axman.com said:

I bought a kraken g12 to use a rog strix lc 240 with an evga rtx 2080 super KO but now i have low framerate on minecraft

why?, I mean, before the upgrade i had stable fps at 60 or so.

Sometimes I also have a random but briefly black out in my screen

 

 

Are you able to monitor the temperatures of the VRAM and VRMs?

Usually the cooler makes contact with the VRAM chips and/or VRMs, and provide them with active cooling -- so no JUST the GPU die gets cooled.

Now with an AIO over the GPU die, sure that will run cooler, but how is the rest of the GPU doing now?

 

What was the reason for going away from the EVGA cooler in the first place?

[axman.com]

I was worried the stock cooler wouldn't be able to handle load under deep learning tasks.

But i have to mention some things:

-1st when I first ran the benchmark, the AIO pump and the bracket's fan wasn't connected, but I never allowed for the gpu to reach a critical temperature

-2nd after connecting the gpu, but not the bracket's fan I ran the furmark benchmark, I thought, what could happen? (I lost the hdmi signal), so I decided to avoid using the gpu until I had a 4pin splitter "apparently" the signal lost wasn't too serious

-3rd even after reinstalling the stock cooler have the same stability issues

-4th the weirdest to me is that even changing energy motor control global config in nvidia panel, it still get stability problems with minecraft, even when the monitoring of the memory and gpu clocks look stable all the time

Pd. When I said stable fps before water cooling, yes, they were stable ranging from 59 to 61, but now I have drops to 28 fps and also, sometimes i get random signal lost from hdmi

 

[Mark Kaine]

54 minutes ago, axman.com said:

When I said stable fps before water cooling, yes, they were stable ranging from 59 to 61, but now I have drops to 28 fps and also, sometimes i get random signal lost from hdmi

 

Ok yeah that *was* stable then and is certainly cause for concern I don't really know how to fix this either, but I also immediately thought of the vrm cooling... Do they have sensors? And were there cooling pads on them? If yes are the cooling pads still on them? Could be overheating, but just a guess. 

 

4 hours ago, -rascal- said:

Now with an AIO over the GPU die, sure that will run cooler, but how is the rest of the GPU doing now?

So, before I google 10 different solutions... how do you typically cool the vrms when using an AIO? 

 

 

[axman.com]

4 minutes ago, Mark Kaine said:

Ok yeah that *was* stable then and is certainly cause for concern I don't really know how to fix this either, but I also immediately thought of the vrm cooling... Do they have sensors? And were there cooling pads on them? If yes are the cooling pads still on them? Could be overheating, but just a guess. 

 

So, before I google 10 different solutions... how do you typically cool the vrms when using an AIO? 

 

 

With an AIO the vrm cooling is usually done through a fan aiming directly with or without heatsinks on the vrms

[axman.com]

I've been thinking, do you know where can I send my gpy for repair?

[Bouncewasp]

58 minutes ago, axman.com said:

I've been thinking, do you know where can I send my gpy for repair?

I assume you mean GPU, not gpy? Either way I think this sounds like an issue with lack of cooling on the rest of the GPU after installing the Kraken, i.e. VRM's etc. You need to either make sure you've got good airflow onto the card from the case intake (can be difficult depending on case, fans, and not having high fanspeed all the time) OR to see if you can mount a fan directly underneath your GPU to cover the VRMs (assuming you GPU is mounted normally).

[-rascal-]

1 hour ago, Mark Kaine said:

Ok yeah that *was* stable then and is certainly cause for concern I don't really know how to fix this either, but I also immediately thought of the vrm cooling... Do they have sensors? And were there cooling pads on them? If yes are the cooling pads still on them? Could be overheating, but just a guess. 

 

So, before I google 10 different solutions... how do you typically cool the vrms when using an AIO? 

 

 

 

Usually there are sensors throughout the GPU.

One (or a couple) over the GPU die itself, one or two over the VRAM, another on the VRM area, etc.

 

You can buy copper / aluminum heatsinks, and mount them onto the VRAM chips, over the VRM ICs, etc.

I ordered a pack of these to fix the VRAM overheating issue on a ROG R9-280X Matrix Platinum.

[Mark Kaine]

1 hour ago, axman.com said:

With an AIO the vrm cooling is usually done through a fan aiming directly with or without heatsinks on the vrms

I see, yeah that's why these coolers usually have fan(s) as well I guess. So were there any cooling pads tho and are they still installed, and a fan is also present for the vrm cooling? 

 

Just asking because especially missing cooling pads are the only thing I could see making these issues, IF it is overheating issue. 

 

6 minutes ago, -rascal- said:

 

Usually there are sensors throughout the GPU.

One (or a couple) over the GPU die itself, one or two over the VRAM, another on the VRM area, etc.

 

Ok, my 2 MSI cards don't have this (1050ti and 1060) they only have 1 temp sensor in any monitoring software... 

 

My EVGA however :

 

 

 

Sensor overload galore.

[axman.com]

17 hours ago, -rascal- said:

 

Usually there are sensors throughout the GPU.

One (or a couple) over the GPU die itself, one or two over the VRAM, another on the VRM area, etc.

 

You can buy copper / aluminum heatsinks, and mount them onto the VRAM chips, over the VRM ICs, etc.

I ordered a pack of these to fix the VRAM overheating issue on a ROG R9-280X Matrix Platinum.

I'll try the heatsinks, at least the vrams get hot enough to leave a first degree burn if touched for long enough.

But got a question, did the heatsinks solve instability problems?

[-rascal-]

4 hours ago, axman.com said:

I'll try the heatsinks, at least the vrams get hot enough to leave a first degree burn if touched for long enough.

But got a question, did the heatsinks solve instability problems?

 

For the ROG Strix 280X Matrix Platinum?

Yup!

 

The GPU is known to have overheating VRAM issues, especially is running the fans at a lower speed. 

8/12 VRAM chips have a heatsink / heatspreader over them, while the remaining 4/12 is just bare...

On this card, because of the...CPU heatsink size copper heatpipes...I had to trim some of the copper fins off the heatsinks, otherwise they won't fit under the heatsink.

Spoiler

 

 

 

 

 

 

 

 

 

 

[axman.com]

On 11/12/2020 at 12:44 AM, Mark Kaine said:

I see, yeah that's why these coolers usually have fan(s) as well I guess. So were there any cooling pads tho and are they still installed, and a fan is also present for the vrm cooling? 

 

Just asking because especially missing cooling pads are the only thing I could see making these issues, IF it is overheating issue. 

 

Ok, my 2 MSI cards don't have this (1050ti and 1060) they only have 1 temp sensor in any monitoring software... 

 

My EVGA however :

 

 

 

Sensor overload galore.

actually, there were cooling thermal pads which aligned with some heat spreader from the stock cooler