By:
Filters
What's new
Carbonite

South Africa's Top Online Tech Classifieds!
Register a free account today to become a member! (No Under 18's)
Home of C.U.D.

GPU not boosting correctly (SOLVED but looking for advice regarding thermal pads)

FrankNStine

FrankNStine

Senior Member
Rating - 100%
33   0   0
Joined
Feb 20, 2020
Messages
443
Reaction score
97
Points
2,785
Age
26
Location
Durban
Hey there good people.

I'm having some weirdness with an RX580 4GB that I bought off Carb. Advertised clocks are supposed to be 1340mhz/1750 memory, but during a benchmark run in Heaven it drops after a few minutes all the way down to 1100mhz -ish. Temps are fine, well under 80C, plenty of power available to the GPU, so I have no idea why it's doing this.

Any professionals/really clever people have any ideas?
 
What temps are the other components of the power delivery?

It might not be the core overheating but maybe the VRAM or VRMs.

Edit: have you opened it up and looked at the thermal paste/pad application?
 
JollyJamma said:
What temps are the other components of the power delivery?

It might not be the core overheating but maybe the VRAM or VRMs.

Edit: have you opened it up and looked at the thermal paste/pad application?
Click to expand...
As far as I can tell, everything is well within spec. VRM temps are at 70C. I haven't opened it up yet because the temps are still pretty low and I literally only picked it up today
 
Still sounds like something is overheating.

Could be anything really.

Best way is to put the GPU on an open bench and use a temp probe to see what’s hot.

What software are you using?

HW Info shows more than just the GPU die temp.

I’d probably send the GPU back though.

Can the seller please confirm that it was running at full boost before they sold it?
 
FrankNStine said:
Hey there good people.

I'm having some weirdness with an RX580 4GB that I bought off Carb. Advertised clocks are supposed to be 1340mhz/1750 memory, but during a benchmark run in Heaven it drops after a few minutes all the way down to 1100mhz -ish. Temps are fine, well under 80C, plenty of power available to the GPU, so I have no idea why it's doing this.

Any professionals/really clever people have any ideas?
Click to expand...

Is it running at 100% usage ?
How many lanes (x16 or x8 or x4?) does GPU-z say it is using ?
What resolution and CPU ?
What brand of card ?

Also, I'd suggest trying it on a benchmark that isn't 13 years old.
 
Nyt Ryda said:
Is it running at 100% usage ?
How many lanes (x16 or x8 or x4?) does GPU-z say it is using ?
What resolution and CPU ?
What brand of card ?

Also, I'd suggest trying it on a benchmark that isn't 13 years old.
Click to expand...
Good point, I believe Time Spy is currently free on Steam.
 
JollyJamma said:
Still sounds like something is overheating.

Could be anything really.

Best way is to put the GPU on an open bench and use a temp probe to see what’s hot.

What software are you using?

HW Info shows more than just the GPU die temp.

I’d probably send the GPU back though.

Can the seller please confirm that it was running at full boost before they sold it?
Click to expand...
I definitely don't have an open bench or a temp probe 🤣 using Heaven cos I have it and it's simple enough. According to the seller, it was fine with him

eleventh said:
Lots of the old mining cards were underclocked and undervolted to conserve power.
Make sure you have the original firmware on it.
Click to expand...
According to GPU-Z, it has the correct BIOS and I don't have an iGPU toi flash the BIOS anyway

Nyt Ryda said:
Is it running at 100% usage ?
How many lanes (x16 or x8 or x4?) does GPU-z say it is using ?
What resolution and CPU ?
What brand of card ?

Also, I'd suggest trying it on a benchmark that isn't 13 years old.
Click to expand...
Yes, it's sitting at 100%
16 lanes
1080p, R5 3600
Gigabyte Gaming RX580 4GB
I use Heaven cos it's simple for a simpleton like me
 
I recall seeing JayzTwoCents using something that looked like GPU-Z and it showed him what the limiting factor of the GPU's performance was (When he undervolted too low, it would say it was power limited etc.). Does anyone know what software it was or what else I could use that would actually say why the GPU is throttling?
 
FrankNStine said:
I recall seeing JayzTwoCents using something that looked like GPU-Z and it showed him what the limiting factor of the GPU's performance was (When he undervolted too low, it would say it was power limited etc.). Does anyone know what software it was or what else I could use that would actually say why the GPU is throttling?
Click to expand...
HWinfo shows the reason that the GPU is throttling but it's almost always PWR/HEAT limit
 
Now I'm starting to think there's an issue with the card itself. I know this reasoning might sound weird, but I was just moving some windows around on the desktop and it was super stuttery. I'm using a 144hz 1080p monitor, so that stutter shouldn't exist... Does this mean F's in the chat?
 
Do yourself a favor
In MSI afterburner edit the voltage frequency curve.
You should easily take the voltages down to 1050 mv for the highest boost

Mine by default is screwed, so i'm suspecting yours is too.
Temperature Target is 75 °C according to techpowerup bios db
so it may be thermal throttling.

www.techpowerup.com

Gigabyte RX 580 VBIOS

4 GB GDDR5, 1340 MHz GPU, 1750 MHz Memory
www.techpowerup.com www.techpowerup.com
but thats for a gpu with Elpida memory, not hynix

vicadia.com

How to Overclock Radeon RX 580 GPU: A Step-by-Step Guide | ViCadia

Do you want to overclock your Radeon RX 580 graphics card, but don’t know how to do it? Here is our comprehensive guide on how to improve your card’s performance.
vicadia.com vicadia.com
 
Last edited:
Okay, so I just found something kinda strange. According to techpowerup, it's supposed to have a TDP of 145W. However, according to GPU-Z it says that the board power is 116W. I was told that the BIOS might have been flashed previously, so would I be wrong in saying that it might have a modded BIOS that's causing all my issues? When I set the power limit to +50% in Afterburner, it does boost up to 1340mhz as it's supposed to
 
FrankNStine said:
omething kinda strange. According to techpower
Click to expand...
FrankNStine said:
Okay, so I just found something kinda strange. According to techpowerup, it's supposed to have a TDP of 145W. However, according to GPU-Z it says that the board power is 116W. I was told that the BIOS might have been flashed previously, so would I be wrong in saying that it might have a modded BIOS that's causing all my issues? When I set the power limit to +50% in Afterburner, it does boost up to 1340mhz as it's supposed to
Click to expand...
You can download the polaris bios editor and view the bios file and confirm if its been modded.

Be careful when flashing a new bios onto the card, it needs to support the RAM that the card has.,

Get the bios directly from gigabyte
www.gigabyte.com

Radeon™ RX 580 GAMING 4G (rev. 1.0/1.1) Support | Graphics Card - GIGABYTE Global

Discover AORUS premium graphics cards, ft. WINDFORCE cooling, RGB lighting, PCB protection, and VR friendly features for the best gaming and VR experience!
www.gigabyte.com www.gigabyte.com
 
I'm absolutely stumped now. The BIOS seems to be the correct one, it's the same as the BIOS that's supplied on Gigabyte's website. I can't tell why this model calls for 116W according to the BIOS but the RX580 on Techpowerup calls for 145W. As far as I can tell, it's throttling because it's trying to keep the power usage down. As soon as I allow +50% power limit in Afterburner it goes to 1340mhz and fluctuates around there, and it draws 140W
 
Still having issues with the boost clocks on this GPU. I did a clean driver install today, and the problem is still there. It didn't even boost to its base clock, just sat at 1130-ish mhz. I've attached screenshots of the voltage curve as per MSI Afterburner and Radeon driver below, anyone have any ideas?

https://imgur.com/a/8aztktY
 
I just made a weird discovery. If I set the power limit to 0% in Afterburner, it's limited to about 116W and the voltage fluctuates between 900mV and 1131mV with the clocks at (more or less) 1120mhz. However, if I set the power limit to +50% then the voltage sits constantly at 1125mV and the power draw goes up to 140W, with the clocks at a steady 1340mhz which is advertised boost clock.

I'm really scratching my head with this thing.

Edit: Does anyone think it could be a P State related problem? The numbers seem to suggest that when the power is limited (ie power limit at 0%), it lives between state 2 and 3. IDK how that works, but the voltages and clocks align 🤷‍♂️
 
Last edited:
Okay. I'm here to claim my idiocy. This entire the time the card has had a BIOS meant for the Rev 1.1, until I disassembled the whole thing did I notice that the sticker says it's a Rev 1.0. So, as I hung my head in shame, I downloaded the F10 BIOS for the Rev 1.0, and whaddaya know, it's working as normal now 🤦‍♂️🤦‍♂️

I'm gonna leave the thread open for a few days for 2 reasons:
1. Roast me, it's well deserved 😂😂😂
2. What size thermal pads should I use? Best guess according to thermalpad.eu is 0.5mm for the VRAM, 1.0mm/ 1.5mm for everything else on the front, and 2mm for the backplate. If anyone knows otherwise, please let this noob know.

Thanks for the help peeps!
 
Last edited:
You must log in or register to reply here.

Users who are viewing this thread

Total: 2 (members: 0, guests: 2)

Latest posts

Top Donors

IOPS
NI∀ƆOƆ‾⊥ᴚ∩ʞ
eagle2k
Sp3c1al1zed
Confucius
nephilm
Athos
EELeat
Off-The-Chart
McDuncun

Share this page

Back
Top Bottom