[SOLVED] Red CPU light on MSI B450 tomahawk with 3080 FE and 2600x (on high loads)

Dec 23, 2020
18
0
10
Hi all,

I recently bought the 3080 FE and replaced my 2070 (no super), did the DDU to uninstall and and install new drivers.

Everytime with high loads (timespy, or games like cod:warzone, cyberpunk 2077) the display output is cut, and the CPU light on my debug lights red. I have to restart the PC.

It boots up perfectly fine and no problem on POST. On lower loads (browsing, video, spotify, discord etc.) there is no issue.

To check if it's due to high temps in my case, I readjusted my fan curves aggressively, so it is kinda louder now but the CPU is at load only reaching to 65-70° C, but when idle the temps are a bit high, 45-55° C.
When I just boot my system up and go to bios, the CPU temp is around 39-40°C but after win10 boots up, it shoots to 45-55°.

In games, for example on warzone, before the system halts (with no video output) and the CPU debug light on MB lights up, the temps were, CPU: 65, GPU: 75
However the FE 3080 heats up my case pretty fast and I reach following temps quite fast while gaming
System: 40°C, PCH: 60°, MOS: 50°

I have two exhaust fans on top and one to the rear. The radiator is sitting in front (no place on top due to MB heatsinks) with its fans on the inside, and also exhausting, so somewhat negative pressure case (I know it's not optimal)

CPU: Ryzen 2600x (stock settings, no extra oc) PBO is on AUTO but I tried with off, too. No difference, I get cpu red light again.
Cooler: Arctic Freezer 240 II
MB: MSI B450 Tomahawk (the regular version, no MAX)
RAM: 2x Corsair vengeance lp 8GB 3000, cl15
GPU: Nvidia FE 3080 (previously it was Gainward 2070)
PSU: Sharkoon Silentstorm Cool Zero 850W (previously it was Kolink KL-600M)
Case: Kolink Stronghold
M2: Sandisk Extreme Pro 512 GB
SSD: Crucial MX500 1TB
HDD: Western Digital Red 3 TB
Case fans: 3x Alpenföhn Wing Boost 3 120mm

I will try following steps today:
Reseat the CPU
Reapply thermal paste
Adjust the rad fans for intake rather than exhaust
Format and install new win10


Can you help what could be the reason for CPU red light while gaming / high loads?

I didn't have any issue until I replaced the PSU with 850W and installed the FE 3080.

Any help is much appreciated.

Thanks!
 
Solution
Any other ideas what could be the reason?
getting the 3080 really ruined my Christmas trying to fix my PC.
I just realized I think I got your post confused with another post. Looking over everything, your problem is likely the PSU. Even though it's advertised as an 850watt unit and it's new, it may just be a poor quality PSU and unable to properly power the system and the GPU. I've never even heard of Sharkoon brand and you should consider that you might need to return it if still possible and get a different PSU.
Dec 23, 2020
18
0
10
Make sure you used the 4+4 pin or 8 pin CPU power cable and that it's actually installed on the motherboard. PCIE 6+2 has the same pin shape on the connector but not the same wiring as the CPU connection. Also make sure the modular cables are fully plugged into the PSU.

I've checked the PSU connections, everything's correct, CPU power is 4+4 pin. All modular cables are fully clicked into place into the PSU as well as MB end / GPU

I also changed the CPU power cable at the PSU end to another 8 pin that's labeled CPU. Still no difference, still got black screen / cpu debug light under load. Temps are not crazy high when it happens, max 72 °C.

Power to the GPU is also going via 2 seperate PCIE 6+2 cables into the mini 12-pin Nvidia connector, so that's also correct

I've formatted my boot drive and did a clean reinstall, still experiencing the same issue.

RAM's are running in their correct XMP settings, had no issues with them since 2 years so far.

Since my CPU doesn't reach high temps when the crash/debug light happens, I don't think temperatures are the reason, but the case is running hotter than I like.

I've also tested my PSU before the 3080 arrived, with my old 2070. And there I didn't have any of this issue.

Any other ideas what could be the reason?
getting the 3080 really ruined my Christmas trying to fix my PC.
 
Any other ideas what could be the reason?
getting the 3080 really ruined my Christmas trying to fix my PC.
I just realized I think I got your post confused with another post. Looking over everything, your problem is likely the PSU. Even though it's advertised as an 850watt unit and it's new, it may just be a poor quality PSU and unable to properly power the system and the GPU. I've never even heard of Sharkoon brand and you should consider that you might need to return it if still possible and get a different PSU.
 
Solution
If temperatures are good, I would suspect either the PSU or the GPU itself. The one review I found on your PSU from profesionalreview.com wasn't too bad. If you have a friend that wouldn't mind testing your new GPU, I'd suggest that to eliminate the possibility of the GPU being the culprit.
 
Dec 23, 2020
18
0
10
I'm pretty sure the GPU is fine, since the error is CPU-bound (according to the MSI debug lights). And yes, a friend indeed bought the FE 3080 for himself too, so he can test my card, just to make sure

I think something's wrong with either MOBO, CPU, or PSU or poor case ventilation.

I'm considering testing my old 600W PSU (it's decent) with my current setup (2600x and rtx 3080) and if I encounter any similar issues.
Do you think that's a good idea? or the total power will trip my PSU?

I'm getting a Noctua thermal paste tomorrow, so I will also try to reseat CPU/reapply thermal paste and try to improve airlow a bit.

I guess I posted this on the wrong thread, I'll post a similar one in the CPU section.
 
I'm pretty sure the GPU is fine, since the error is CPU-bound (according to the MSI debug lights). And yes, a friend indeed bought the FE 3080 for himself too, so he can test my card, just to make sure

I think something's wrong with either MOBO, CPU, or PSU or poor case ventilation.

I'm considering testing my old 600W PSU (it's decent) with my current setup (2600x and rtx 3080) and if I encounter any similar issues.
Do you think that's a good idea? or the total power will trip my PSU?

I'm getting a Noctua thermal paste tomorrow, so I will also try to reseat CPU/reapply thermal paste and try to improve airlow a bit.

I guess I posted this on the wrong thread, I'll post a similar one in the CPU section.
Honestly, your CPU temperatures won't improve by changing the paste and they are well within spec. It seems more likely your AIO is not mounted properly in the first place. If you still have the stock cooler that came with the 2600X, you could install that and see what kind of temperatures you get. It should reach up to about 85c under load.

The only other part that could be at fault or likely be at fault would be the motherboard. I still think the PSU is the likely failure point. Even though it's new, that doesn't mean it's not defective.

Your old 600watt PSU won't be able to properly power the RTX 3080, so don't bother attempting that.
 
Dec 23, 2020
18
0
10
Honestly, your CPU temperatures won't improve by changing the paste and they are well within spec. It seems more likely your AIO is not mounted properly in the first place. If you still have the stock cooler that came with the 2600X, you could install that and see what kind of temperatures you get. It should reach up to about 85c under load.

The only other part that could be at fault or likely be at fault would be the motherboard. I still think the PSU is the likely failure point. Even though it's new, that doesn't mean it's not defective.

Your old 600watt PSU won't be able to properly power the RTX 3080, so don't bother attempting that.

Yes possibly the AIO doesn't sit very properly, will check after the new paste arrrives. I also still have the stock cooler, and will try it if nothing solves this.

I hope the motherboard is fine, since I was planning to use it with a 5600x once the bios update comes, but I wasn't planning to buy a new mobo, dam they are pricey.

In the meantime I'll probably replace back the 2070 and install the old psu just to get thru this holiday. the temps were fine, framerate not so much but it'll do for now until I check with a new psu

thanks a lot!
 

g-unit1111

Titan
Moderator
Honestly, your CPU temperatures won't improve by changing the paste and they are well within spec. It seems more likely your AIO is not mounted properly in the first place. If you still have the stock cooler that came with the 2600X, you could install that and see what kind of temperatures you get. It should reach up to about 85c under load.

The only other part that could be at fault or likely be at fault would be the motherboard. I still think the PSU is the likely failure point. Even though it's new, that doesn't mean it's not defective.

Your old 600watt PSU won't be able to properly power the RTX 3080, so don't bother attempting that.

I agree, that PSU is insufficient for a 3080, you need a very strong and solid PSU for that GPU. I wasn't too familiar with that Sharkoon unit (we don't have that brand here in the US) but after Googling it, it looks like it's a Seasonic based unit. I would say that the PSU isn't a problem here. I would also agree to check the AIO mounts in case that is indeed the issue. You could also try an air cooler and see if the CPU boots up that way.
 
Dec 23, 2020
18
0
10
I agree, that PSU is insufficient for a 3080, you need a very strong and solid PSU for that GPU. I wasn't too familiar with that Sharkoon unit (we don't have that brand here in the US) but after Googling it, it looks like it's a Seasonic based unit. I would say that the PSU isn't a problem here. I would also agree to check the AIO mounts in case that is indeed the issue. You could also try an air cooler and see if the CPU boots up that way.

There is no issue with PC booting up. It has a healthy POST. The CPU debug light only turns red and display gets cut when there is immense load, e.g. during 4K rtx on Cyberpunk. During idle it's fine.

Alright, I won't try the old 600W PSU. It was a 80 Bronze. On HWMon, the max my CPU drew was 105W and my 3080 drew 320W, so I still believe a 600W PSU can handle it, just barely. But not worth the risk of damaging components, IMO.

However, today I just used DDU and replaced my 3080 with my old 2070 instead. Now I don't get any CPU debug light issues - like I had before. The stock 2600x and 2070 won't probably load the 850W Sharkoon to fail anyway. Now it works fine without any debug lights. The CPU isn't taxed at the settings for 2070 anyway so no issues there and CPU temp 58-60°C, and GPU 65-70 °XCwhile gaming, too. I might keep it this way until Christmas is over and I can order a new PSU.

Sharkoon might be a decent/OKish PSU brand, but I'm leaning towards the fact that my sample might be an unlucky one.

I will test tomorrow my 3080 FE with a friend's PC who also got the 3080 FE on the same day as me (we tried really hard for a month to get it at MSRP). At least then I can make sure the GPU is OK.

The Noctua thermal paste arrived today, so I'll remount cpu wand rad/fans for better airflow.

Thanks for all the input, I will update here in case someone has the same issues.
 

g-unit1111

Titan
Moderator
There is no issue with PC booting up. It has a healthy POST. The CPU debug light only turns red and display gets cut when there is immense load, e.g. during 4K rtx on Cyberpunk. During idle it's fine.

Alright, I won't try the old 600W PSU. It was a 80 Bronze. On HWMon, the max my CPU drew was 105W and my 3080 drew 320W, so I still believe a 600W PSU can handle it, just barely. But not worth the risk of damaging components, IMO.

However, today I just used DDU and replaced my 3080 with my old 2070 instead. Now I don't get any CPU debug light issues - like I had before. The stock 2600x and 2070 won't probably load the 850W Sharkoon to fail anyway. Now it works fine without any debug lights. The CPU isn't taxed at the settings for 2070 anyway so no issues there and CPU temp 58-60°C, and GPU 65-70 °XCwhile gaming, too. I might keep it this way until Christmas is over and I can order a new PSU.

If your old GPU works fine then that would indicate that it's a PSU issue. The 3080s require a lot of power to run, more than what previous generations of cards (9xx/10xx/20xx) have in the past. I would check and see what the actual OEM is on that unit because if it's not a Seasonic or Super Flower, that could be what the cause of the problems are. A lot of power supplies are not what they are advertised.
 
Dec 23, 2020
18
0
10
Thanks!

I should mention I'm running HWMON in the background, and I saw my clockspeeds on all cores go up to 4.9-5 Ghz - with low loads (installing software, watching 4K YT etc.). The PBO is on.
View: https://imgur.com/MkZRzAr

Is this normal?
(But when the 3080 was in the case, I tried with and without PBO, had the same CPU debug light issue)

OK, I'll get a new PSU and try that way. If the problem persists, that means it's time to switch for a b550+5600x kit :)
What would be a good PSU? I checked the Tier list but not everything's available right now.

My options (considering availability and price) are:

Corsair RMx 1000 = 165€ - that's what I'm considering actually. 850 isn't available for 2 weeks here. Gives me headroom for potential 5600x overclock in the future and will run at lower utilisation and quieter, I suppose.
Seasonic Focus GX850 = 135€
Seasonic Focus PX850 = 165€

Which would you suggest? Corsair is a bit long (180mm), but my case allows up to 200mm. It's gonna be a pain plugging those cables but anyway.
 
Thanks!

I should mention I'm running HWMON in the background, and I saw my clockspeeds on all cores go up to 4.9-5 Ghz - with low loads (installing software, watching 4K YT etc.).

Is this normal?
Use a different monitoring program like Core Temp 1.16 or HWInfo. Running a 2600X at 4.9+GHz is far from normal. What is your CPU voltage at idle and with max reading? Check your bios for CPU multiplier being higher than 42.

Unless you tried to manually overclock your CPU, but didn't change it back, you may have a problem with your bios and need to reset to defaults to see if it fixes the higher than normal frequency. If resetting to defaults does not fix it, you can try updating to the newest bios for your motherboard, which was released on the 22nd and now supports Ryzen 5000. 7C02v1H1(Beta version) is the one you want.


 
Dec 23, 2020
18
0
10
The 5 Ghz on HWMon was only a spontaneous reading. I didn't try any manual overclock for my CPU, ever. The multiplier on bios is 36x.

My Vcore at idle is around 1.4V, min 1.36V, max 1.45V, but I thought that was normal for Ryzen.

On AIDA64 stress test, I can only get to 3.9 Ghz on all cores. Max reading is 4.1 Ghz.

I actually did a bios update yesterday, but not to the beta version, the previous one. That had not fixed the issue with CPU Ez debug light with my 3080.
I was not sure if the latest one will not support the older CPUs like my 2600x.
 
The 5 Ghz on HWMon was only a spontaneous reading. I didn't try any manual overclock for my CPU, ever. The multiplier on bios is 36x.

My Vcore at idle is around 1.4V, min 1.36V, max 1.45V, but I thought that was normal for Ryzen.

On AIDA64 stress test, I can only get to 3.9 Ghz on all cores. Max reading is 4.1 Ghz.

I actually did a bios update yesterday, but not to the beta version, the previous one. That had not fixed the issue with CPU Ez debug light with my 3080.
I was not sure if the latest one will not support the older CPUs like my 2600x.
Are your readings from before or after updating the bios? Your voltages are "normal" with PBO enabled, so 1.45v makes sense, but may not be needed for 4.1GHz.

Turn off PBO and set CPU voltage offset to negative and see what happens when you play a game. There is a chance your motherboard or the CPU have a defect that is affecting the automatic boosting of the CPU and may be the cause of your issues.

If nothing I've suggested works, you may need to enable spread spectrum if it's not already enabled.
 
Dec 23, 2020
18
0
10
Are your readings from before or after updating the bios? Your voltages are "normal" with PBO enabled, so 1.45v makes sense, but may not be needed for 4.1GHz.

Turn off PBO and set CPU voltage offset to negative and see what happens when you play a game. There is a chance your motherboard or the CPU have a defect that is affecting the automatic boosting of the CPU and may be the cause of your issues.

If nothing I've suggested works, you may need to enable spread spectrum if it's not already enabled.

I got 4.9-5 Ghz (they are really just a momentary spike) readings BEFORE as well as AFTER flashing the latest non-beta bios yesterday. They happened also not with heavy loads, just regular non-heavy use. However the previous crashes I was talking about happened not while the CPU is on high clockspeeds, the clock was around 4000 when they happened. Power settings are "balanced" not "Ryzen balanced".

However I just cleared my CMOS with the jumper. Now I don't see any spikes reaching to that range. But up to 4300 in games.

Since yesterday I changed my 3080 with 2070 (still with the new Sharkoon PSU), I didn't get any issues with CPU debug light as well.

Right now the old 2070 is installed, so I will check my temps during gaming.
Then I will test the 3080 once again, if this error was due to some weird OC presets in the BIOS and maybe resolved with CMOS clearing.
I'll keep you guys updated.

Thank you Mr. Third-Eye and merry christmas :)
 
Dec 23, 2020
18
0
10
Ok, just tested my 3080 FE at my friend's PC. It works flawless so that variable is out.

However on my PC, even after clearing CMOS/resetting bios, I even saw 5200 MHz on HWMonitor during idle-ish tasks, I think that can't be true, and if it's Hwinfo64 was not open, unfortunately.

Since the GPU works fine, either my PSU is bad, or my CPU or mobo is bad. The latter is more likely imo, since I'm not comfortable with these high clocks.
 
Dec 23, 2020
18
0
10
Ok with Hwinfo64 all the clock speeds are what they supposed to be. Around ~4.0 GHz allcore and 4.2 single core.

I checked all my PSU cables and sockets once again, installed the 3080 once again. First, timespy benchmark went without issues, after that, I fired up cyberpunk and again, black screen and CPU debug light.

I'll also check, reseat the CPU and the aio soon. As soon as I am able to get a new PSU, that will hopefully solve it. After that, if the problem persists, my plan is go and get the 5600x which I can run on my mobo and was planning to get anyway, see if the problem is CPU caused, and if not, at last get a new mobo.
 
Ok with Hwinfo64 all the clock speeds are what they supposed to be. Around ~4.0 GHz allcore and 4.2 single core.

I checked all my PSU cables and sockets once again, installed the 3080 once again. First, timespy benchmark went without issues, after that, I fired up cyberpunk and again, black screen and CPU debug light.

I'll also check, reseat the CPU and the aio soon. As soon as I am able to get a new PSU, that will hopefully solve it. After that, if the problem persists, my plan is go and get the 5600x which I can run on my mobo and was planning to get anyway, see if the problem is CPU caused, and if not, at last get a new mobo.
At this point, I kind of feel like the PSU is probably fine since it's new and I suspect the CPU is at fault, but I might just be the motherboard. I would take the CPU out and check the pins, there may be a broken, bent or missing pin. Also, assuming there are no problems with the CPU pins, you should RMA your motherboard for a replacement before buying a new PSU or CPU. If a different motherboard changes nothing, you likely have a faulty CPU.

Edit - I forgot to mention, I have seen some similar situations like this with Ryzen CPUs. In two of the three cases I saw, the CPU was the problem. One CPU had a missing pin and the other was just defective with no bent or missing pins and was RMA replaced. The third case, the motherboard had faulty components and was RMA replaced. Getting a new CPU might just fix the problem, but then you will have to decide if you want to spend extra money before getting your Ryzen 5000. Getting the motherboard replaced will cost you shipping to the manufacturer.

Honestly, if you plan to play at Cyberpunk 2077 1440p to 4k resolution, you won't lose much fps playing on a Ryzen 5 3600X or 7 3700X compared to a Ryzen 5 5600X.
 
Last edited:
Dec 23, 2020
18
0
10
At this point, I kind of feel like the PSU is probably fine since it's new and I suspect the CPU is at fault, but I might just be the motherboard. I would take the CPU out and check the pins, there may be a broken, bent or missing pin. Also, assuming there are no problems with the CPU pins, you should RMA your motherboard for a replacement before buying a new PSU or CPU. If a different motherboard changes nothing, you likely have a faulty CPU.

Edit - I forgot to mention, I have seen some similar situations like this with Ryzen CPUs. In two of the three cases I saw, the CPU was the problem. One CPU had a missing pin and the other was just defective with no bent or missing pins and was RMA replaced. The third case, the motherboard had faulty components and was RMA replaced. Getting a new CPU might just fix the problem, but then you will have to decide if you want to spend extra money before getting your Ryzen 5000. Getting the motherboard replaced will cost you shipping to the manufacturer.

Honestly, if you plan to play at Cyberpunk 2077 1440p to 4k resolution, you won't lose much fps playing on a Ryzen 5 3600X or 7 3700X compared to a Ryzen 5 5600X.

Thank you! It seems I no longer get the debug light problem.

I just took my cpu out of its socket, reseated my cpu, used my noctua paste and mounted the aio cooler.
While I'm there, I also adjusted my rad fans for intake position from the front, the rest of the top/rear fans are in exhaust mode as before.

Temps are still higher than I would like on idle, 45-50°C but no higher than 66°C on load now (4K MSFS2020).

I think reseating the CPU somehow fixed the CPU debug light problem.

I'll inform if the error still occurs.
 
Dec 23, 2020
18
0
10
Nope, the error is still present. In fact, this time, I managed to log the hwinfo64 information up until to the moment it crashed witch CPU debug light.
Nothing out of ordinary in terms of CPU temp (51°C), or power usage.

It is really weird that for the past hour or so the PC run fine (including intensive gaming), then all of a sudden it goes cpu red light, when GPU and CPU both have to work hard (gaming).

Still, the PSU can be faulty to deliver enough power, or the mobo/cpu is somehow defect.
 
Dec 23, 2020
18
0
10
Here is my hwinfo64 log during the crash.

By the way, on idle, my temps are much better now, at 39-40°C but every few sec (5) there is a momentary spike of 10°C (temp goes 50°C) which is reduced in 1-2 sec again to 40°C. This behaviour was present before even I got my 3080 and new PSU, but the temps were around 5°C more before reseating/reapplying thermal paste.

So temps should not be an issue at all.

New PSU is coming tonight (Seasonic Focus GX 850), so I'll try that if that fixes the problem (somehow I'm not optimistic).

Other than that, I'm really baffled. The fact is, my mobo and cpu is over 2 years old now (just out of warranty), which I had run with my 600W PSu without issues. If one of them had any issue, I would've discovered them by now, they didn't have any issues running with my 2070. Even now, when I replace the 2070 with the 3080, the PC runs fine without any debug lights. Probably, the reason is that my 2070 draws much less power, and not pushing my CPU or MoBo (I mean power delivery of its PCIE) hard enough to fail.

I tried my 3080 at a friend's PC, it ran without issues (but to be fair, we might have not tried it long enough to crash, but 3080 FE's are pretty stable and good units as far as I know).

The fact that I can not diagnose the problem is really frustrating.. worst.. christmas.. ever..
 
Dec 23, 2020
18
0
10
Latest update: New psu seems to have solved it. Seasonic Focus GX850.

Gaming for at least two hours with different games, not once have I experienced the cpu debug light error.

In case I encounter it again, I'll post it here.