• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Gigabyte 5700xt hot spot

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
Been using this card for about 7 months now and very happy with it. My only concern is that I see the gpu hot spot get into the 90°C to 92°c range when gaming. I see that it's good to about 110°C or so. Am I being worried over nothing?

I started messing with undervolting today but I got stuck flashing the new bios back to the card. Should I be going this route or should I stick with wattman? Any ideas on a starting point for this card? Right now I'm just kinda flying blind
 
Joined
Jul 25, 2020
Messages
30 (0.02/day)
Location
Germany
Processor i7-4790K 4.6GHz all core
Motherboard MSI Z97 Gaming 7
Cooling Arctic Liquid Freezer II 280
Memory 32GB DDR3 2400MHz 11-13-13-31
Video Card(s) 5700XT
Display(s) 144Hz 1080p
Power Supply Evga GQ 750W
Software WindOof and Linux
My only concern is that I see the gpu hot spot get into the 90°C to 92°c range when gaming.

The hotspot sensor (as one can assume) returns the hottest spot of the entire die. Since vega AMD has put some sensors inside the die, and of many this is the hottest spot. Nvidia perhaps has this feature too to control their boosting algorithm but does not show it to the user.

Am I being worried over nothing?

TL;DR
Do not worry about if it is below 100°C. 90-95°C is fine here if you only do gaming. For mining (or other 24/7 task) lower TDP for lesser temp, or undervolt. AMDs stock coolers take them for low level gaming beyond 110°C. Acually it stops there, but the GPU will throttle then.

Longer explaination:
The Navi 10 XT Chip (on 5700 XTs) is a very small chip 251mm² compaired to (for example) a vega 10 xt chip (on vega 64) 495mm². I got both so I can draw some conclusions:
Both have 240W TDP (both Sapphire Nitro+)
Navi10 chip has a power density of ~0.96 W/mm² | (Surface per watt 1.05mm²/W)
Vega10 chip has a power density of ~0.48 W/mm² | (Surface per watt 2.06mm²/W)

(Yes I know, not every watt is going to the GPU die, but a high amount of 150-200W if needed)

Therefore the coldplate from the cooler can collect the ermitted heat better (because of more surface and therefore more surface per watt). Vega64s hotspot is usually 10-15°C ahead from the normal GPU temp sensor, even at 300W TDP. (I also can not tell if the Nitro+ cooler on the vega is just better, it is however bigger)

RX5700XTs Hotspot is normaly 15-25°C higher than the GPU temp sensor. As said this mostly comes from the more compact chip.

Should I be going this route or should I stick with wattman?

I was afraid of this sensor too, so I found out that I can undervolt my GPU and still rase the GPU clock to wattman maximum. I do not like Wattman because it resets randomly. I recommend MSI Afterburner.
GPUZAfterburner.JPG

Stock 1200mV Vcore, 2070 MHz Core, 1800MHz mem. Do not touch mem! You can really brake the GPU by touching it!

I started messing with undervolting today
I recommend to try lowering voltage (-10mV steps) until the temperatures fits your needs (as said, everything below 100°C is fine for gaming). Use Unigine Superposition to test it (as example for a gaming-like benchmark) because you want to keep track of your performance as well. Lower core frequency only if it crashes. Some games however (Tomb Raider/ Unreal games) can still crash. Then just quickly lower core frequency a bit to continure gaming.

If you are done with optimising temperatures and you did not have to touch the core frequency, you can still try to increase it (slowly). But that is voluntary extra work.

I got stuck flashing the new bios back to the card.
I see no reason why to flash a 5700 XT. You have an unlocked GPU. Also flashing can destroy that thing quite fast.
 

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
Thank you for this info. I saw the TJ of the card was 110°C but people were telling me "that's way to high" so I kinda started to worry.

I've never been a huge fan of afterburner, but I know what you mean about wattman. Even running stock I would get messages that it crashed and reset.


Any other quality benchmarks I should look at. Just bought 3dmark/vrmark/pcmark 10 bundle off steam for $10. Figured it was worth the future investment.
 
Joined
Jul 25, 2020
Messages
30 (0.02/day)
Location
Germany
Processor i7-4790K 4.6GHz all core
Motherboard MSI Z97 Gaming 7
Cooling Arctic Liquid Freezer II 280
Memory 32GB DDR3 2400MHz 11-13-13-31
Video Card(s) 5700XT
Display(s) 144Hz 1080p
Power Supply Evga GQ 750W
Software WindOof and Linux
People usually use 3dmark Timespy or Firestrike. Also for the real maximum temperature use furmark or OCCTPT (this is a benchmark suite with CPU tests aswell). Both are can be called a power virus, for a good reason. Use only to test a stable undervolt for the maximum possible temperature (and not longer than 5 mins, It is just pointless). And also game a bit. Because benchmarks are great for compairision but if your undervolt is not stable (or at the edge of not being stable) you have nasty micro shutters. In addition not every game likes undervolts... You even may notice this after hours, or days running the game. Warframe for example does not like my undervolt and rarely crashes. Tomb Raider (2013) also crashes at undervolts. Some games may not care at all and run good.


What Gigabyte GPU do you have? (RX 5700 XT yes, but which cooler Gigabyte Aorus/ Gigabyte Gaming?)
 

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
Gigabyte gaming oc. What really got me into all this was Warframe over the past 2 weeks or so. The game and my pc didn't get along and had a lot of crashes thought my gpu was going bad.

When in the orbiter, looking out, I thought they were shooting stars,but I was getting artifacts everywhere. Turn to inside the orbiter, I'd get blue/green patches. Game would crash or freeze when someone would show up on the right side of the screen. Can't 100% confirm that but that's what it looked like.

Sent a dump file to support and they said it was my pc, not the game. Everything was stock except xmp. The optimization patch seems to have fixed it all tho.
 
D

Deleted member 193596

Guest
90°C is not good for any piece of silicon.

but it won't kill your card. since it's just the hottest measured point and not the average of the card itself.

you're fine.
 
Joined
Jul 25, 2020
Messages
30 (0.02/day)
Location
Germany
Processor i7-4790K 4.6GHz all core
Motherboard MSI Z97 Gaming 7
Cooling Arctic Liquid Freezer II 280
Memory 32GB DDR3 2400MHz 11-13-13-31
Video Card(s) 5700XT
Display(s) 144Hz 1080p
Power Supply Evga GQ 750W
Software WindOof and Linux
I was getting artifacts everywhere.
Long flickering triangles? But if it was fixed then nothing to worry about.

Hmm 190W TDP with that ticc cooler and still quite hot... Undervolting should increase your performance too, if you were TDP bound. If undervolting should not help you may remove your cooler to reapply more thermal paste there (you do not have to do this, but this is what I would do if nothing works). Or use teflon shims on the screws behind the die (usually -5°C but be careful).
Because this cooler looks very large for only 190W.

And also here from techpowerups bios archive:
Code:
Gigabyte RX 5700 XT 8 GB BIOS

Power Limit
Total: 190 W
GPU: 170 A  # ampere times voltage = watts, max amps on the die
[...]

Thermal Limit
  Edge: 100°C # normal GPU temp sensor trottle
  Hotspot: 110°C # trottle down at this temp
  Memory: 105°C
  [...]
  Shutdown Temp.: 118°C

Fan Speed
  Target Temperature: 90°C  # idk if hotspot or edge but mine has 91°C
  [...]
So everything is fine...

90°C are not good for every piece of silicon.
Yes true, silicon and especially GPUs do not like temps higher than 90°C. But for example a Nvidia GTX 970 would have 89°C as GPU edge sensor thermal limit. The hotspot would be much hotter at this temperature, perhaps 95-99°C (die has almost vega size but lesser power). So it is just that most people panic (including me) if they see the new sensor and forget about the other GPU edge we had for generations.
Furthermore I can not remember a 251mm² die with about 200W from past GPUs...

Edit:
-"fine"
 

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
this is what I got after about an hour of gaming. Still stock everything as far as the GPU is concerned
 

Attachments

  • Screenshot (6).png
    Screenshot (6).png
    57.2 KB · Views: 743
Joined
Jul 25, 2020
Messages
30 (0.02/day)
Location
Germany
Processor i7-4790K 4.6GHz all core
Motherboard MSI Z97 Gaming 7
Cooling Arctic Liquid Freezer II 280
Memory 32GB DDR3 2400MHz 11-13-13-31
Video Card(s) 5700XT
Display(s) 144Hz 1080p
Power Supply Evga GQ 750W
Software WindOof and Linux
Indeed totally fine. Fans ramp up quite a bit (114% fan speed ?!) but as long as this does not bother you change nothing.
 

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
that is what originally caused me some concern a few months ago. Wondering if that would cause a lifespan issue with the fans. Now I'm noticing my temps are climbing again to almost 100C at hot spot.

Also the artifacts are back in warframe. Superposition seems to run ok. Timespy though, I see pixilated boxes and screen will flash black and white boxes for a frame or so. Not sure what's going on at all.

When I did notice the high temps, for some dumb reason I reflashed my bios with the ones found here

Didn't do anything to fix the issues, but now seeing issues stated above in benchmarks that I wasn't before
 

Attachments

  • Screenshot (8).png
    Screenshot (8).png
    55.2 KB · Views: 326
D

Deleted member 193596

Guest
DDU'd the drivers? (including disabling auto driver installation)
reinstall LATEST drivers again.

does it still happen after that?
 
D

Deleted member 193596

Guest
yes 20.9.1

if you are not familiar with DDU use the "clean install" option in the installer.
 

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
I've used it last time I was trying different versions of drivers to see if I could find one that didn't have the issue

Used DDU and did clean install in safe mode. Rebooted back into normal windows and installed drivers 20.9.1. Went into game (warframe) and artifacts started showing up as soon as I logged in. Closed out game and tried to run Superposition, crashed almost right away. Tried to play borderlands 3, didn't get to main menu, got pop up saying drivers crashed. Everything worked fine yesterday, but not today
 
D

Deleted member 193596

Guest
I've used it last time I was trying different versions of drivers to see if I could find one that didn't have the issue

Used DDU and did clean install in safe mode. Rebooted back into normal windows and installed drivers 20.9.1. Went into game (warframe) and artifacts started showing up as soon as I logged in. Closed out game and tried to run Superposition, crashed almost right away. Tried to play borderlands 3, didn't get to main menu, got pop up saying drivers crashed. Everything worked fine yesterday, but not today

reseated the card?
is it overclocked?
 
Joined
Jul 25, 2020
Messages
30 (0.02/day)
Location
Germany
Processor i7-4790K 4.6GHz all core
Motherboard MSI Z97 Gaming 7
Cooling Arctic Liquid Freezer II 280
Memory 32GB DDR3 2400MHz 11-13-13-31
Video Card(s) 5700XT
Display(s) 144Hz 1080p
Power Supply Evga GQ 750W
Software WindOof and Linux
Do you always use the latest driver? I use 20.4.2 because all other drivers do akward things on my rig. And I need stability.

I also installed 20.9.1 and started Warframe. In the orbiter all my 3 monitors started fickering periodicaly at the same time. I did not move around.

Then I reinstalled 20.4.2 and problems were gone.

Went into game (warframe) and artifacts started showing up as soon as I logged in.
It would be really helpful if you could explain "artifacts" more precisely. I did never have artifacts in this game except those screen flicker ones described above. And I ran this game on different GPUs (RX5700XT/GTX1080Ti/GTX970/RX270) all with OC.
 
Joined
Feb 18, 2005
Messages
5,239 (0.75/day)
Location
Ikenai borderline!
System Name Firelance.
Processor Threadripper 3960X
Motherboard ROG Strix TRX40-E Gaming
Cooling IceGem 360 + 6x Arctic Cooling P12
Memory 8x 16GB Patriot Viper DDR4-3200 CL16
Video Card(s) MSI GeForce RTX 4060 Ti Ventus 2X OC
Storage 2TB WD SN850X (boot), 4TB Crucial P3 (data)
Display(s) 3x AOC Q32E2N (32" 2560x1440 75Hz)
Case Enthoo Pro II Server Edition (Closed Panel) + 6 fans
Power Supply Fractal Design Ion+ 2 Platinum 760W
Mouse Logitech G602
Keyboard Logitech G613
Software Windows 10 Professional x64
Sounds like AMD's legendary drivers strike again. Make sure ALL programs that could possibly interfere with the card (including Afterburner are uninstalled, then try 20.4.2 as @alxeus suggests.

And next time, if it ain't broke, don't try to fix it.
 

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
here's a quick video I captured with some of the issues present.


I DDU'd and installed 20.4.2 but just the drivers, not the radeon software. Still the same thing.
 
Last edited:
Joined
Jul 25, 2020
Messages
30 (0.02/day)
Location
Germany
Processor i7-4790K 4.6GHz all core
Motherboard MSI Z97 Gaming 7
Cooling Arctic Liquid Freezer II 280
Memory 32GB DDR3 2400MHz 11-13-13-31
Video Card(s) 5700XT
Display(s) 144Hz 1080p
Power Supply Evga GQ 750W
Software WindOof and Linux
Long slender triangle in the upper middle of the screen, kind of like a god ray but it does it all the time. more so when I'm in the orbiter but does pop up on missions as well. I'll get a couple for a few seconds then nothing, then like 10 in 5 seconds again.
My first 5700 XT Nitro+ had a cratered heatspreader (disassembly showed it therefore hotspot was always 107°C, 87°C edge, fans on max). Before I RMAed it I got black slender triangles in Total War: Warhammer 2 almost every frame. But only on campain map. Perhaps because battle mode is mostly CPU limited.
In superposition they started when everything was lifting up (black slender triangles on almost each frame).

here's a quick video I captured with some of the issues present.
Looks almost like my Warhammer 2 experience.

As the artifacts are still present with the 20.4.2 driver I would rather blame the hardware.

(I would test VRam, is there a VRam error tester that one can rely on? MemtestCL is unusable.)
 
Joined
Feb 18, 2005
Messages
5,239 (0.75/day)
Location
Ikenai borderline!
System Name Firelance.
Processor Threadripper 3960X
Motherboard ROG Strix TRX40-E Gaming
Cooling IceGem 360 + 6x Arctic Cooling P12
Memory 8x 16GB Patriot Viper DDR4-3200 CL16
Video Card(s) MSI GeForce RTX 4060 Ti Ventus 2X OC
Storage 2TB WD SN850X (boot), 4TB Crucial P3 (data)
Display(s) 3x AOC Q32E2N (32" 2560x1440 75Hz)
Case Enthoo Pro II Server Edition (Closed Panel) + 6 fans
Power Supply Fractal Design Ion+ 2 Platinum 760W
Mouse Logitech G602
Keyboard Logitech G613
Software Windows 10 Professional x64
By artifact you can see in the picture. Long slender triangle in the upper middle of the screen, kind of like a god ray but it does it all the time. more so when I'm in the orbiter but does pop up on missions as well. I'll get a couple for a few seconds then nothing, then like 10 in 5 seconds again.

The thing that looks like it's coming off the Earth?

BTW, which frame is that? I used to play Warframe but stopped as I found it got way too grindy.
 
D

Deleted member 193596

Guest
looks more like a hardware issue now.

if you can test the card in another PC to verify the issue would be great.

if not... make a backup, reinstall windows (just to make 100% sure that is not a software issue whatsoever) if the problem still exists i'd RMA the card.
 

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
BTW, which frame is that? I used to play Warframe but stopped as I found it got way too grindy.
Its Mesa with a custom color. There are parts that are very grindy. Good things to have that are like a 0.02% drop rates and such. If you're willing to spend a little bit of money when you get a 50% or 75% off tho its not as bad.

looks more like a hardware issue now.

if you can test the card in another PC to verify the issue would be great.

if not... make a backup, reinstall windows (just to make 100% sure that is not a software issue whatsoever) if the problem still exists i'd RMA the card.
I can throw it in my old pc and see what happens.

So on the suggestion from Alxeus, I downloaded OCCT. First time running 3D test and VRAM test, it gave me a bunch of errors detected, but doesn't tell me what they are.

Played the game in monitor mode, no errors, no issues at all with the game. Just doesn't make sense.

This would be the 2nd time I've completely reformatted and installed windows to try and fix this issue.

I'm almost starting to think that maybe the gpu wasn't using all its vram and causing the issue? Im gonna restart again and see what happens.
 

Attachments

  • OCCT-20200921-111639-GPUMemtest.png
    OCCT-20200921-111639-GPUMemtest.png
    172.1 KB · Views: 199
  • OCCT-20200921-112050-GPU3D.png
    OCCT-20200921-112050-GPU3D.png
    153.5 KB · Views: 203
Joined
Jul 25, 2020
Messages
30 (0.02/day)
Location
Germany
Processor i7-4790K 4.6GHz all core
Motherboard MSI Z97 Gaming 7
Cooling Arctic Liquid Freezer II 280
Memory 32GB DDR3 2400MHz 11-13-13-31
Video Card(s) 5700XT
Display(s) 144Hz 1080p
Power Supply Evga GQ 750W
Software WindOof and Linux
So on the suggestion from Alxeus, I downloaded OCCT. First time running 3D test and VRAM test, it gave me a bunch of errors detected, but doesn't tell me what they are.
Just told you to use this or Furmark for maximum heat test. To be honest, I did not had the newest version and therefore did not know of the memory test. However at GPU Memory I do not have a single error (on both, Vega64 and 5700XT). And I think the artifacts are coming (for some reason) from the memory.
(Ignore displayed GPU usage, this is from my vega)


GPU 3D mode however displayes errors (or sometimes not, very random). And as the program does not tell me what these errors mean (games run good) I would ignore them. (Maybe little complains about my undervolt)


(I asked a friend of mine to run this on his nvidia GPU as prove if this can be trusted)
 

Attachments

  • OCCT 3D 5700 XT.JPG
    OCCT 3D 5700 XT.JPG
    143.1 KB · Views: 148
  • OCCT-20200921-182847-GPUMemtest.png
    OCCT-20200921-182847-GPUMemtest.png
    202 KB · Views: 307
Last edited:

Renrut23

New Member
Joined
Sep 18, 2020
Messages
11 (0.01/day)
heard bad things happening to GPUs from Furmark so always been iffy on using it. Seems to run fine, now i can run superposition as well without it crashing. very strange as to why bc i havent changed any settings since i sent those pic or vids of whats going on
 

Attachments

  • Screenshot (15).png
    Screenshot (15).png
    105.8 KB · Views: 168
  • Screenshot (17).png
    Screenshot (17).png
    185.1 KB · Views: 226
Top