RTX 3080 hashrate drop list

[lasttimei]

Just noticed if I set the fans to auto on my Zotac, at 32c ambient fans went straight up to 100%. Definitely on the throttling side. I deshrouded it and zip tied 2x 140mm fans and according to HWINFO, fans are now at 75%. Went down from 100% to 75%. Big help. Let's see if this holds at 35c summer temps. Also checked the pads of the Zotac Amp Holo --- they are 2mm. Will replace the pads later if this still doesn't hold.

Yeah tried that atm i've a ryzen cpu heatsink at the top and it works somewhat, but i think ima try the thermal pad mod , just hope that the 2mm works fine.

How does this work? I mean how did you mount the heatsink?

My 3080’s are not thermal throttling but i did notice the backplate is super hot where the memory and vrams are located so i put a 120mm case fan on there .

Now its just warm to the touch not burning anymore .

Maybe try that and see if it fixes ur problem before u open ur card .

Can you please post pictures? Kinda interested how you did it

i just dropped a cheap ryzen heatsink on top of it still not fixed but lowered temps a bit.

I think this would work better if you dont want to open your card.
This has to be a 3090 since there's no way a 120mm fan would fit, unless its smaller fan.
Got this image from a search,
https://imgur.com/gallery/tkLoZnt

I'm going to be trying something like this
https://i.imgur.com/ZkaRTbs.jpg

I'm not opening the front looks like too much work, just the backside.
unless this doesn't fix it.

My 3080FTW3 i think its probably the best card it can be really hot and it still pushes 100mhs.
Should be getting a few strix 3080 hope thats also good with memory cooling.
and another FE -_-probably selling it to a friend if i cant fix this one.

[Seph213]

Just noticed if I set the fans to auto on my Zotac, at 32c ambient fans went straight up to 100%. Definitely on the throttling side. I deshrouded it and zip tied 2x 140mm fans and according to HWINFO, fans are now at 75%. Went down from 100% to 75%. Big help. Let's see if this holds at 35c summer temps. Also checked the pads of the Zotac Amp Holo --- they are 2mm. Will replace the pads later if this still doesn't hold.

Yeah tried that atm i've a ryzen cpu heatsink at the top and it works somewhat, but i think ima try the thermal pad mod , just hope that the 2mm works fine.

How does this work? I mean how did you mount the heatsink?

My 3080’s are not thermal throttling but i did notice the backplate is super hot where the memory and vrams are located so i put a 120mm case fan on there .

Now its just warm to the touch not burning anymore .

Maybe try that and see if it fixes ur problem before u open ur card .

Can you please post pictures? Kinda interested how you did it

I would but i have no clue how to post pics on this site .

On ur backplate u have that spot where the mlcc/moscaps are showing . Well put a 120mm fan right on it and let it blow air on that area .

Card were running at between 44-48C during the day , now they at 41-44 but the main goal of this mod was to cool the memory and vram which i believe it did since the backplate is way cooler than before .

[Seph213]

Just noticed if I set the fans to auto on my Zotac, at 32c ambient fans went straight up to 100%. Definitely on the throttling side. I deshrouded it and zip tied 2x 140mm fans and according to HWINFO, fans are now at 75%. Went down from 100% to 75%. Big help. Let's see if this holds at 35c summer temps. Also checked the pads of the Zotac Amp Holo --- they are 2mm. Will replace the pads later if this still doesn't hold.

Yeah tried that atm i've a ryzen cpu heatsink at the top and it works somewhat, but i think ima try the thermal pad mod , just hope that the 2mm works fine.

How does this work? I mean how did you mount the heatsink?

My 3080’s are not thermal throttling but i did notice the backplate is super hot where the memory and vrams are located so i put a 120mm case fan on there .

Now its just warm to the touch not burning anymore .

Maybe try that and see if it fixes ur problem before u open ur card .

Can you please post pictures? Kinda interested how you did it

i just dropped a cheap ryzen heatsink on top of it still not fixed but lowered temps a bit.

I think this would work better if you dont want to open your card.
This has to be a 3090 since there's no way a 120mm fan would fit, unless its smaller fan.
Got this image from a search,
https://imgur.com/gallery/tkLoZnt

I'm going to be trying something like this
https://i.imgur.com/ZkaRTbs.jpg

I'm not opening the front looks like too much work, just the backside.
unless this doesn't fix it.

My 3080FTW3 i think its probably the best card it can be really hot and it still pushes 100mhs.
Should be getting a few strix 3080 hope thats also good with memory cooling.
and another FE -_-probably selling it to a friend if i cant fix this one.

Have u put thermal pads under those heatsinks ? I was thinking of doing that but the fan fix did it for me cause i don’t have an FE and the mlcc area is open so air gets through to the mem/vram

[btcshiner]

Apparently if you put your rig in a electric vehicle you can drive around and the air flow will reduce any thermal throttling....

https://www.tomshardware.com/news/the-fastest-rtx-3080-mining-rig-is-a-bmw-i8
 

[fenomenyaa]

HWInfo version 6.42 can now also display the GDDR6X temperatures

Guys pls share results..

[Seph213]

HWInfo version 6.42 can now also display the GDDR6X temperatures

Guys pls share results..

Would that be GPU memory junction temperature ?

[philipma1957]

HWInfo version 6.42 can now also display the GDDR6X temperatures

Guys pls share results..

Would that be GPU memory junction temperature ?

yeah I loaded 6.4.2 and don't see a number claiming to be ram temp.

but I have access to a 3060ti which uses gddr6


https://www.gigabyte.com/Graphics-Card/GV-N306TAORUS-M-8GD#kf

[jelome198959]

Yeah, just updated to HWINFO 6.42. It does indeed show memory temperatures. Just under GPU Core temp. But only for GDDR6X. It doesn't show GDDR6 temps, maybe not necessary as it doesn't run hot?

Here's mine at 28c ambient: (a bit cold today, some days temps reach 32c, will update later)
Asus TUF - 98c
MSI Ventus - 108c (near throttling)
Zotac AMP Holo - 96c
Gainward Phoenix - 86c
Palit GamingPro - 88c

All at +1000 memory except Palit at +700. The Asus, MSI and Zotac are deshrouded, and I have zip tied 120mm/140mm fans. Zip-tied fans are running at 90% speeds.
Surprisingly, the cheapest 3080s on my rig perform the best, and they're not deshrouded, fans at 70% for both.
The more expensive brands perform worse, even though they are deshrouded with high quality fans!

[ildotoress]

I've been playing with my MSI 3080Gaming X Trio for the days.
It can't do more than 86MH/s, no matter the temperature.
Issue is in the BIOS as when power reaches 205W it throttles.

I am using MSI Curve with minimum allowable voltage, 0.725 (in reality 0.731), and 1125MHz.
If I use normal PL level it does not do more than 82MH/s.
Also it must have a minus 120MHz on the memory, if I give it more it throttles instantly to 60-70MH/s.

Stay away from the Gaming X for minig!


PS. Yes, you loose warranty if: you change BIOS with any BIOS not destined for Gaming X (exception the 2 official ones= OC mode and silent mode - tested both, same 205W throttle limit), or if you remove the seal on the 4th GPU screw.

It's very strange, as I saw on some reviews, this card have TDP of 340w right?

Source: Toms Hardware: https://www.tomshardware.com/reviews/msi-geforce-rtx-3080-gaming-x-trio-review

I have an EVGA 3080 and I'm doing 98 stable, with 230w on software and 250w at the wall - I did fan curve to reach 1200Mhs on core clock with 718mv - 70% TDP
3080s are power hungry, and if I reduce even 1%, the card start to hash a lot less Mhs

I think your 1125Mhz on core clock is low, here I need exactly 1200mhz

But I don't know why your card only have 205w until throttle

Is it possible your card is having this throttle because of memory temperature?
Maybe after hit some temperature, the bios reduce the wattage to cool the board

I also tried 1200MHz, 1225, 1250, 1300, etc
The Gaming X is pure crap.

After seeing the VRM analysis I understood why. It has an underwhelming implementation that does deliver the necessary power but because it cuts corners in quality so it overheats massively in doing it.
And as a cherry on the top, the VRM and memory cooling is not efficient as the cooler does not have a coper base plate but direct contact width the heat pipes (and also a small micro radiator which is covered with a thermal pad so no airflow can flow over it  ).
That is the reason for throttling at exactly 205W.

So I sold it to some one else to use it for mining (he did not care about warranty and wanted to add the Suprim X BIOS).

Now I got instead for the same money a Suprim X which does:
100MH/s @ 0.725V with 1250MHz Core and + 1200MHz VRAM, 60% Fan (1800rpm).

Now I'm happy a camper  


PS. @agente, please remove the Gaming X from your no throttle list as that is a pure lie!

[Wotan Wipeout]

My personal experience with the Asus TuF 3080 (Watercooled):

max Ram +1149 = 98.6 MHs ETH
Going higher results in lower MHs.
So it might be other cards have faster ram timings.

VRAM Temp: 70 degrees.

[ShinyGG]

Updating everyone on the results after updating to HWinfo 6.42:

GPU temp followed by Memory Junction Temp:

1. GameRock: 66C, 96C
2. GameRock: 63, 98
3. GameRock: 55, 92
4. GameRock: 66, 106!?
5. GameRock: 52, 88
6. MSI Suprim X: 40, 90
7. GameRock: 43, 76
8. GameRock: 62, 94
9. MSI Suprim X: 51, 98
10. EVGA FTW3 Ultra: 62, 96

Clocks: -500 clock, +1000 Memory on all but number 7 & 8 sitting at +800.

Is this expected? Someone mentioned that the temps aren't for actual VRAM here, but reflect something else entirely? Im worried this isn't normal?

As for my room temps, it's around 20-25 I imagine. I dont have a thermometer here unfortunately.

[ShinyGG]

Updating the thread:


MSI Suprim X: Have 2. One will cause a crash at more than +1000 memory, hashes around 92 only. The other one is perfect at 99-100mhs, 35-40C on 1200 memory -500 clock, 60% PW 60% Fan
Palit Gamerock: 6 of them in total, all run 98-101 at 200W, by far the best card for mining. +1200 mem, -500 clock, 60%, 60%

Are both suprim x on the same bios ? I’m planning on making a 6 gpu rig with all suprim x . Don’t tell me they don’t all hit 99-100

I did not play around with the BIOS, I assumed I just didn't get lucky with the 2nd one. They can push to 99, but I'd rather prefer stability over the extra 2-2.5H. Running them at 96 and 94 respectively.

[agente]

PS. @agente, please remove the Gaming X from your no throttle list as that is a pure lie!

I moved to "Some throttle, some do not". Waiting more reviews..

[Seph213]

I've been playing with my MSI 3080Gaming X Trio for the days.
It can't do more than 86MH/s, no matter the temperature.
Issue is in the BIOS as when power reaches 205W it throttles.

I am using MSI Curve with minimum allowable voltage, 0.725 (in reality 0.731), and 1125MHz.
If I use normal PL level it does not do more than 82MH/s.
Also it must have a minus 120MHz on the memory, if I give it more it throttles instantly to 60-70MH/s.

Stay away from the Gaming X for minig!


PS. Yes, you loose warranty if: you change BIOS with any BIOS not destined for Gaming X (exception the 2 official ones= OC mode and silent mode - tested both, same 205W throttle limit), or if you remove the seal on the 4th GPU screw.

It's very strange, as I saw on some reviews, this card have TDP of 340w right?

Source: Toms Hardware: https://www.tomshardware.com/reviews/msi-geforce-rtx-3080-gaming-x-trio-review

I have an EVGA 3080 and I'm doing 98 stable, with 230w on software and 250w at the wall - I did fan curve to reach 1200Mhs on core clock with 718mv - 70% TDP
3080s are power hungry, and if I reduce even 1%, the card start to hash a lot less Mhs

I think your 1125Mhz on core clock is low, here I need exactly 1200mhz

But I don't know why your card only have 205w until throttle

Is it possible your card is having this throttle because of memory temperature?
Maybe after hit some temperature, the bios reduce the wattage to cool the board

I also tried 1200MHz, 1225, 1250, 1300, etc
The Gaming X is pure crap.

After seeing the VRM analysis I understood why. It has an underwhelming implementation that does deliver the necessary power but because it cuts corners in quality so it overheats massively in doing it.
And as a cherry on the top, the VRM and memory cooling is not efficient as the cooler does not have a coper base plate but direct contact width the heat pipes (and also a small micro radiator which is covered with a thermal pad so no airflow can flow over it  ).
That is the reason for throttling at exactly 205W.

So I sold it to some one else to use it for mining (he did not care about warranty and wanted to add the Suprim X BIOS).

Now I got instead for the same money a Suprim X which does:
100MH/s @ 0.725V with 1250MHz Core and + 1200MHz VRAM, 60% Fan (1800rpm).

Now I'm happy a camper  


PS. @agente, please remove the Gaming X from your no throttle list as that is a pure lie!

If thats the case why would the suprim bios make it go over 205W ? I think msi pwr throttled this card so it doesnt compete with the suprim X model .

I have the gaming trio X in my gaming pc mining at 100mhs .
Gpu temp is 48C , gpu junction temp is 108C

On open air frame

1th suprim X : 44C , 93-94C
2nd suprimX : 41C , 92C

[Seph213]

Updating the thread:


MSI Suprim X: Have 2. One will cause a crash at more than +1000 memory, hashes around 92 only. The other one is perfect at 99-100mhs, 35-40C on 1200 memory -500 clock, 60% PW 60% Fan
Palit Gamerock: 6 of them in total, all run 98-101 at 200W, by far the best card for mining. +1200 mem, -500 clock, 60%, 60%

Are both suprim x on the same bios ? I’m planning on making a 6 gpu rig with all suprim x . Don’t tell me they don’t all hit 99-100

I did not play around with the BIOS, I assumed I just didn't get lucky with the 2nd one. They can push to 99, but I'd rather prefer stability over the extra 2-2.5H. Running them at 96 and 94 respectively.

I have 2 of them and both running at 100mhs . Mem is +1300
Both on gaming mode bios. Didnt test silent mode to see if it reduces the hashing .

[jelome198959]

Gaming X Trio stock bios has some memory limitation as said by other users.

But it doesn't throttle because of overheating, it's throttling because of some bios limitation.

That's why when flashing Suprim X bios, it's fixed, as evidenced by Seph213... We also talked about this earlier in this thread.

[ildotoress]

Yes, there are two issues:
1.Lower quality VRM and lower efficiency cooling solution on the Gaming X compared to Suprim (resulting in higher overall AND Vram temp)
2.Lower BIOS limit for throttling, as I have complained initially and it was proven by the users who used the Suprim Bios on the Gaming X.

MSI does not want to give a different tuned BIOS for the Gaming X.
So the conclusion remains, avoid this implementation if possible.

[Erpelusas]

Galax / KFA2. GeForce RTX 3080 SG (1-Click OC).

HiveOs. Drivers Nvidia 455.45.01

Hash  96.62 MH
Temp 56°
rpm.  60%
Power Limit 225 W
Core  +150
Micron Mem HiveOs  1760 . Windows +880
Watts 225

No problem.

UPGRADE.:

Hash  96.58 MH
Temp 54°
rpm.  60%
Power Limit 220 W
Core  +100
Micron Mem HiveOs  1750 .
Watts 220

[Seph213]

The fan on the backplate pushing air on the mlcc area drops temps by 2C only .

Would need to see if putting thermal pads on the backplate+ heatsink with fan blowing on it would reduce those Vrams/mem temps .

Are they made to run this hot ?

[Wotan Wipeout]

For all 3080 buyers... think about watercooling.
This drops the temperatur to 70 degrees (VRAM).