MSI Radeon RX 6900 XT GAMING Z TRIO 16G - High Hotspot Temp/Large Delta

[HaKN !]

You big meanie.

I think we cracked it.

Good work there buddy ! MSI did a poor piss job at pasting/testing i guess.

 

[BlaezaLite]

Those temps are much more like it should be.

 

[LLezerus]

Here's temps at +12 power limit, 340W load. lol

 

[BlaezaLite]

Here's temps at +12 power limit, 340W load. lol

View attachment 279111

What did you use to get a 340w load?

 

[LLezerus]

What did you use to get a 340w load?

TimeSpy Extreme Stress Test, 20 loops.
here's the min-max-average

Hilariously enough, I got this achievement after running the standard run :')

 

[Space Lynx]

Here's temps at +12 power limit, 340W load. lol

View attachment 279111

I wouldn't be doing any sort of OC, 105 celsius is not good. If temps are good running stock, just leave it at stock. Glad to see the repaste fixed stock temps.

I think I might do a repaste now myself, you have inspired me (I already have MX-6 as well)

 

[LLezerus]

I wouldn't be doing any sort of OC, 105 celsius is not good. If temps are good running stock, just leave it at stock. Glad to see the repaste fixed stock temps.

I think I might do a repaste now myself, you have inspired me (I already have MX-6 as well)

It was more viscous than I expected to spread, seems like good stuff.

 

[Space Lynx]

I have some o-rings for a mech keyboard that are like close to 0.5mm, I wonder if I could use those as washers. lol

that and a mx-6 repaste... I am very very tempted to do a before and after temp comparison of tests now.

well I had nothing planned for tonight, so I guess I have plans now, thanks lol

 

[LLezerus]

I have some o-rings for a mech keyboard that are like close to 0.5mm, I wonder if I could use those as washers. lol

that and a mx-6 repaste... I am very very tempted to do a before and after temp comparison of tests now.

well I had nothing planned for tonight, so I guess I have plans now, thanks lol

my main scare was whether i had enough contact but i seem to have applied it well

 

[BlaezaLite]

I just ran the same test as you @LLezerus and maxed out the power (332.5w) and got a hotspot reading of 96C, so your card is still running hot, but it is within spec. My advice would be to leave the power stock and change the fan curve. I'm now slightly deaf after running Time Spy Extreme twice (1st time was at 290w) and having the fans ramp up to 3300rpm! I could of got the liquid cooled version of my card for the same price, kind of regret not getting that one now...

 

[LLezerus]

I just ran the same test as you @LLezerus and maxed out the power (332.5w) and got a hotspot reading of 96C, so your card is still running hot, but it is within spec. My advice would be to leave the power stock and change the fan curve. I'm now slightly deaf after running Time Spy Extreme twice (1st time was at 290w) and having the fans ramp up to 3300rpm! I could of got the liquid cooled version of my card for the same price, kind of regret not getting that one now...

im gonna try running it with what I consider “pre-diminishing returns” settings, +7 on power (so just above rage mode’s +6), 100mhz on core, +112 on memory with fast timings. I also found in my limited experience with the XFX 5700XT I mentioned that letting the paste “cure” by putting load for a few days has an effect on temps so I’m gonna see how we go. I know its considered an old wives tale in the community but I have faith in it.

 

[Space Lynx]

im gonna try running it with what I consider “pre-diminishing returns” settings, +7 on power (so just above rage mode’s +6), 100mhz on core, +112 on memory with fast timings. I also found in my limited experience with the XFX 5700XT I mentioned that letting the paste “cure” by putting load for a few days has an effect on temps so I’m gonna see how we go. I know its considered an old wives tale in the community but I have faith in it.

default fan curves are the worst culprit for higher temps, so if you aren't setting a slightly better fan curve than default you are doing it wrong imo.

i never go crazy with my fan curve, but I make sure my fans are at 80% when the card is at 75 celsius or so. i do a hard cap at 80% otherwise its too loud. but as the curve is aggressive in the early stages - temps still stay in check in my experience. its when cards idle hot and the fan kicks in that 80% too late that temps lose out in the end.

that's just been my experience though

 

[BlaezaLite]

im gonna try running it with what I consider “pre-diminishing returns” settings, +7 on power (so just above rage mode’s +6), 100mhz on core, +112 on memory with fast timings. I also found in my limited experience with the XFX 5700XT I mentioned that letting the paste “cure” by putting load for a few days has an effect on temps so I’m gonna see how we go. I know its considered an old wives tale in the community but I have faith in it.

Your card so you can do as you please, but I agree with @Space Lynx, set a custom fan curve. Glad that your temps have dropped and yeah, TIM can "cure".

 

[LLezerus]

default fan curves are the worst culprit for higher temps, so if you aren't setting a slightly better fan curve than default you are doing it wrong imo.

i never go crazy with my fan curve, but I make sure my fans are at 80% when the card is at 75 celsius or so. i do a hard cap at 80% otherwise its too loud. but as the curve is aggressive in the early stages - temps still stay in check in my experience. its when cards idle hot and the fan kicks in that 80% too late that temps lose out in the end.

that's just been my experience though

i got a custom fan curve but my limit is about 72% :’)

 

[Space Lynx]

i got a custom fan curve but my limit is about 72% :’)

i would try to bring that up to 77 at least and make sure the 77 is hitting at 75 celsius... and hard cap it at 77%... see if that helps... 5% doesn't sound like much but i think it does help. 72 is just a little too low for my personal taste.

 

[eidairaman1]

Good work there buddy ! MSI did a poor piss job at pasting/testing i guess.

No its been a chronic problem of MSIs, heatsink designs, they dgaf about giving out great AMD parts.

i got a custom fan curve but my limit is about 72% :’)

Later on find some plastic washers and do the mod like what was done on the VEGA VII, RX 5700XT, heck someone did it on a 1080ti and it helped.

 

[LLezerus]

i would try to bring that up to 77 at least and make sure the 77 is hitting at 75 celsius... and hard cap it at 77%... see if that helps... 5% doesn't sound like much but i think it does help. 72 is just a little too low for my personal taste.

I've toyed with the fan curve some more to get it comfortable but effective. I've also got a balance for power target and performance. I only get about 200 extra points at +12 on Timespy which is incredibly marginal.

No its been a chronic problem of MSIs, heatsink designs, they dgaf about giving out great AMD parts.
Later on find some plastic washers and do the mod like what was done on the VEGA VII, RX 5700XT, heck someone did it on a 1080ti and it helped.

I will try and acquire some. I went to some local hardware stores but couldn't find any small enough.

I have some o-rings for a mech keyboard that are like close to 0.5mm, I wonder if I could use those as washers. lol

that and a mx-6 repaste... I am very very tempted to do a before and after temp comparison of tests now.

well I had nothing planned for tonight, so I guess I have plans now, thanks lol

I actually decided to repaste my CPU with the MX-6 as well, it actually enabled me finally to get over 6800 on CPU-Z's multicore test.
Single core test is about 6.67% over stock, 10.66% over on multicore, beating out the 5800X which was more expensive. :')

P.S. The enclosed MX Wipes that came with the paste absolutely STINK but they are very effective. Can taste the limonene off them.

 

[AsRock]

I have some o-rings for a mech keyboard that are like close to 0.5mm, I wonder if I could use those as washers. lol

that and a mx-6 repaste... I am very very tempted to do a before and after temp comparison of tests now.

well I had nothing planned for tonight, so I guess I have plans now, thanks lol

Rubber decays over time.

 

[Space Lynx]

Rubber decays over time.

where does one buy washers for a gpu? any thoughts? USA

 

[LLezerus]

where does one buy washers for a gpu? any thoughts? USA

AMD Radeon VII 16 GB Review

The time has come. We're finally allowed to talk about Radeon VII performance numbers. The company's new flagship graphics card is the world's first to be made using a 7 nanometer production process. Also, it has the largest VRAM size of any card below $1000: 16 GB.

www.techpowerup.com

I see here TPU used 6x3x0.5mm washers, it seems you can get them from amazon, but only in large multipacks alongside other washers.

I found these in my toolbox, they look exactly right.

 

[Space Lynx]

AMD Radeon VII 16 GB Review

The time has come. We're finally allowed to talk about Radeon VII performance numbers. The company's new flagship graphics card is the world's first to be made using a 7 nanometer production process. Also, it has the largest VRAM size of any card below $1000: 16 GB.

www.techpowerup.com

I see here TPU used 6x3x0.5mm washers, it seems you can get them from amazon, but only in large multipacks alongside other washers.

I found these in my toolbox, they look exactly right. View attachment 279373

I found some right size washers in my toolbox as well, but I am wondering... the article he says to use plastic ones... but he uses metal in the pictures I think. Not sure if I should use metal or just order some plastic ones?

I have a ruler so I am going to measure just to make sure they are 6mm by 3mm by 0.5mm

 

[LLezerus]

I found some right size washers in my toolbox as well, but I am wondering... the article he says to use plastic ones... but he uses metal in the pictures I think. Not sure if I should use metal or just order some plastic ones?

I have a ruler so I am going to measure just to make sure they are 6mm by 3mm by 0.5mm

they said they recommend plastic, but they used metal anyway. probably what they had on hand.

Update: I am all washered up now, lets see what happens

Didn't go above 92C on the hotspot, the average was about 88C when the test ended,.

 

[Space Lynx]

they said they recommend plastic, but they used metal anyway. probably what they had on hand.

Update: I am all washered up now, lets see what happens
View attachment 279377

Didn't go above 92C on the hotspot, the average was about 88C when the test ended,.
View attachment 279383

so you want from 113 celsius to 92 celsius with paste change and washers? in same tests? side panel on the case both times?

 

[LLezerus]

so you want from 113 celsius to 92 celsius with paste change and washers? in same tests? side panel on the case both times?

From beginning to end yes, the paste made the most impact, taking the hotspot down to about 97, then the washers did the rest I suppose.
Same test, Timespy Extreme 20 loops, case closed.

 

[Space Lynx]

From beginning to end yes, the paste made the most impact, taking the hotspot down to about 97, then the washers did the rest I suppose.
Same test, Timespy Extreme 20 loops, case closed.

That's awesome. I didn't get around to doing it yet, but you have inspired me, lets see if I can bring my 94 celsius hotspot down below at least 90 with a paste change and washers.

W1zzard shows the washer going under the backplate, but you did it above with the screw top, does it matter which way?