Question [SOLVED] RTX 3080 Underperforming (Low Speed), Strangely Power Limits Itself, Never Gets Above 60C

[glilopi]

###Troubleshooting Help:

**What is your parts list? [Consider formatting your parts list.](http://www.reddit.com/r/buildapc/wiki/pcpp)**

[PCPartPicker Part List](https://pcpartpicker.com/list/H8Z8Kp)

Type|Item|Price
:----|:----|:----
CPU | [Intel Core i9-12900KF 3.2 GHz 16-Core Processor](https://pcpartpicker.com/product/2sstt6/intel-core-i9-12900kf-32-ghz-8-core-processor-bx8071512900kf) | $579.50 @ B&H
Motherboard | [Asus PRIME Z690-P WIFI D4 ATX LGA1700 Motherboard](https://pcpartpicker.com/product/vx...-atx-lga1700-motherboard-prime-z690-p-wifi-d4) | $233.96 @ B&H
Memory | [Team T-Force Delta RGB 32 GB (2 x 16 GB) DDR4-3600 CL18 Memory](https://pcpartpicker.com/product/ZR...b-ddr4-3600-cl18-memory-tf4d432g3600hc18jdc01) | $136.99 @ Amazon
Storage | [ADATA Swordfish 1 TB M.2-2280 NVME Solid State Drive](https://pcpartpicker.com/product/y4...2-2280-nvme-solid-state-drive-aswordfish-1t-c) | $89.99 @ Amazon
Video Card | [Gigabyte GeForce RTX 3080 10GB 10 GB GAMING OC Video Card](https://pcpartpicker.com/product/9z...0-gb-gaming-oc-video-card-gvn3080gamingoc10gd) | $1899.00 @ Amazon
Operating System | [Microsoft Windows 11 Home OEM 64-bit](https://pcpartpicker.com/product/dKkWGX/microsoft-windows-11-home-oem-64-bit-kw9-00633) | $119.99 @ Newegg
Monitor | [Asus ROG Strix XG27AQ 27.0" 2560x1440 170 Hz Monitor](https://pcpartpicker.com/product/wfH7YJ/asus-rog-strix-xg27aq-270-2560x1440-170-hz-monitor-xg27aq) | $449.00 @ Amazon
| Prices include shipping, taxes, rebates, and discounts |
| Total | $3508.43
| Generated by [PCPartPicker](https://pcpartpicker.com) 2022-02-18 13:43 EST-0500 |

Cooling: iBUYPOWER 360mm Addressable RGB Liquid Cooling System
Case: iBUYPOWER Slate HAKO MR Tempered Glass ARGB
PSU: Not sure unless I remove panels/shrouds to get to it

Describe your problem. List any error messages and symptoms. Be descriptive.

Recently purchased an iBUYPOWER pre-built due to the current GPU supply chain issues (https://www.ibuypower.com/Store/Gaming-RDY-SLHBG205). Decided to run some benchmarks on 3DMark Timespy (Steam demo version) and found that my GPU appears to be underperforming (see links below).

https://imgur.com/a/iFw86sH

View: https://imgur.com/a/iFw86sH


This has been confirmed with UNIGINE Heaven and Userbenchmark (yes, I know this is a bit outdated for Directx 12 applications), which flagged my GPU as performing in the 6th percentile compared to other 3080s. For reference, the CPU appears to be about where it should be from the 3DMark and CPUz benchmarks. Notably, the CPU is a bit lower than the average TimeSpy score, but this could be because I'm running factory clocks on everything. The CPU score is within the bell curve whereas the GPU is way below it. The best TimeSpy GPU score I could get was 14600 vs. factory scores of 17000-18000 that I have seen on other threads. I have gotten scores that were much lower (11000 - 12000), seemingly inconsistently even when the GPU is commanded to idle higher (via Nvidia power management settings). This is strange to me.

What I have noticed is that the GPU does not go much beyond base speeds (1440 MHz) during any of these benchmarks and nowhere near the factory rated boost speed of 1800 MHz. I have also read on other threads that stock cards should be going up into the 1900s and approaching 2000. Using MSI Afterburner (no OC beyond factory) and GPUz, I can see that these low speeds are maintained even while the voltage (appears to be on the curve) and temperatures (never exceeds 60C) are well below their limits. I see that the power limit goes from zero to 1, presumably indicating that the card is limiting its power draw, and stays flat there. The voltage or temperature limit booleans are never flipped to 1. The interesting thing is that the percent power draw (in afterburner) never goes beyond 60% (shows in Afterburner but not GPUz for some reason). Fans stay at 50-60% or below. Something else that is weird... there is no indication (reads as zero) of absolute power draw (i.e., Watts) in MSI, GPUz, or even HWMonitor. I'm not sure if this is just a vendor choice, weird bug, or related issue. Multiple inputs read as zero on GPUz (video engine load, bus interface load, all of the power draws, all of the PCIe and 8-pin slot power/voltage). My feeling is that this may be a related issue (GPU erroneously limits itself on power while not running at high speed, temperature, or voltage... combined with weird/no values being reported for power).

It should be noted that this behavior is consistent across benchmarks and gaming. For reference, with graphics maxed (RTX off) on Cyberpunk 2077 (post 1.5 patch), I get 120 fps and about 60 fps with RTX on Psycho at 1440p. I understand that this performance is great, compared to most machines, but it still doesn't make sense that the 3080 is not clocking itself up to full potential.

List anything you've done in attempt to diagnose or fix the problem.

1. Should be noted that the machine initially booted but failed to display on the first try. Resolved the issue by reseating the RAM. Not sure that this is related but figured that I should include.
2. Updated Windows 11
3. Installed latest Nvidia drivers (clean install)
4. Tried removing drivers using DDU then updating (clean install)
5. Changed Nvidia power settings to prefer maximum performance, Windows power settings to high performance, BIOS EZ Tuning to extreme performance, all with minor impact besides making my GPU idle at a higher speed when I launched the Timespy interface. Weirdly, I also saw some lower scores (11,000) when I started messing with these settings.
6. Removed and reseated GPU (PCIe slot and power connections). I did notice that there was some small amount of residue (white mushy material... thermal paste or material from the GPU card perhaps?) in the holes (partially clogging at the front face not fully packed) above the slot. This wasn't a huge amount that I felt too concerned about it. I cleaned out what I could. It appears only to be in a few of the holes (which I don't believe impact the slot performance... could be wrong). This material is on the key (shorter side) of the PCIe slot. I will go back with a duster can and alcohol to just cross this off the list but couldn't do that at the time.
7. Clean install of Windows 11 and updates, ensured VBS is off
8. Clean update of Nvidia drivers
9. Ensured XMP and Resizable BAR are on
10. Tested PCIe interface in GPUz (confirmed it runs at 16 4.0 under load)
11. Attempted to move power limit from 100% to 99% in Afterburner (thinking it might change performance if there was a bug)
12. Flashed motherboard BIOS to latest version
13. Looked for any obvious, loose connections inside the case

Post relevant photos of build/parts here.


Provide any additional details you wish below.

I've scoured the internet for any other similar issues, and it seems like most folk's underperformance issues (Timespy scores) are much closer to the average or the result of an obvious bottleneck (thermal or CPU). I've seen some similar issues where the PCIe isn't running at full capacity, but as noted above I believe that mine is functioning correctly (at least for data transmission). To me, the fact that the power limit is being hit at low speeds/temperatures while supposedly not drawing much power, coupled with the fact that a lot of the power sensors (possibly not included or reported from this card?) aren't appearing in my monitoring software may not be a coincidence. The only other things I could think to check would be the PSU (not really sure what I can do myself here, would think that I would see instability/crashing if the PSU supply was inadequate) or trying to clean the PCIe slot (don't think this is the problem).

At this point, I'm thinking that the GPU may be faulty. I contacted iBUYPOWER's technical support and walked through some of these issues. The support agent said he thinks it's likely a GPU hardware issue since I had already tried all of the above, including flashing the motherboard BIOS (which he recommended). They have offered to RMA the GPU which seems like the logical next step, but I'm still hopeful that I'm missing something else possibly. I'm hoping to not go down a rabbit hole if the replacement GPU doesn't work. At this point, I'd probably just ask to send the entire machine back. I understand that the issue could be the GPU, PSU (I would maybe expect crashes or instability if this was the case), motherboard (not really sure how else to diagnose... CPU seems to be operating fine), or some software bug (at this point, I don't know what else it could be). I'm at a bit of a loss because, as mentioned above, the machine still performs very well compared to the average machine (Cyberpunk 2077 at 120 fps on ultra 1440p w/o RTX), but I don't feel that the GPU is performing up to its potential/acting funny. It's strange to me that the GPU is not running itself up on temperature or speed when playing these demanding titles! I definitely would like to fix this now and avoid issues down the road, including not maximizing the performance/FPS that I should be getting today.

I would appreciate any suggestions/opinions/moral support that the GPU may be faulty and an RMA is warranted. Thank you!


SOLVED - UPDATE 02/19/2022

Tried flashing to a previous stock GPU BIOS without any luck. I was going to try a new PSU when I fixed the issue. Turns out it was the residue/thermal paste that I thought wasn't a big deal. I would have cleaned it before but didn't have an air duster can/brush. See before picture below:

https://imgur.com/a/r1jQzM0

View: https://imgur.com/a/r1jQzM0


I cleaned it out with a fine interdental brush and careful application of compressed air. Took me quite a while to meticulously get all the bits out, and I had no expectation of success. All my previously missing GPUz sensor inputs are now reading, Timespy/Userbenchmark scores are where they should be, GPU runs up into boost clock territory, temperatures come up as expected, and the card only limits itself when it truly hits the power limit. Just as expected. I'm honestly shocked that this was the cause, as I didn't think those holes were really all that important (compared to the slot). Clearly I was wrong. Thank you all for the support. I feel like I've really been on a journey (have spent hours working/stressing) with this and glad to have reached a resolution. Still plan on reseating the CPU cooler heatsink, but I think I'll save that for tomorrow

Updated Timespy scores:

https://imgur.com/a/UyLJK3j

View: https://imgur.com/a/UyLJK3j


Updated UserBenchmark: https://www.userbenchmark.com/UserRun/50646482

 

[drivinfast247]

Of all the stuff you listed, you left out the most important part, the PSU. I am assuming it's a iBuypower brand and that pretty much means it's trash and needs to be replaced.

 

[glilopi]

Of all the stuff you listed, you left out the most important part, the PSU. I am assuming it's a iBuypower brand and that pretty much means it's trash and needs to be replaced.

That's correct. It is iBuypower, not sure on wattage bc it's under the shroud. Is there any way to test whether this could be the issue? As mentioned, I figured that the PSU failing to supply adequate power would result in voltage sags/instability. I wouldn't think that the GPU would indicate power limit switched on (refusing to pull more power from the PSU) if this was the cause?

 

[glilopi]

Of all the stuff you listed, you left out the most important part, the PSU. I am assuming it's a iBuypower brand and that pretty much means it's trash and needs to be replaced.

Touché. I pulled some stuff off. It's a High Power 800W (HP1-J800GD-F12S) which admittedly appears to be sketchy, but I had kinda heard that about their brand. Is there any way to diagnose if this could be the issue? I still think it's odd that the GPU Power Limit boolean is at 1. I would think that value would be reported by the GPU if it hits its power limit, but perhaps it could report that based on the value it's trying to pull regardless of it's actually able to pull it? I also think it's odd that the power number isn't reporting in the monitoring software, but maybe this is a red herring?

 

[drivinfast247]

Touché. I pulled some stuff off. It's a High Power 800W (HP1-J800GD-F12S) which admittedly appears to be sketchy, but I had kinda heard that about their brand. Is there any way to diagnose if this could be the issue? I still think it's odd that the GPU Power Limit boolean is at 1. I would think that value would be reported by the GPU if it hits its power limit, but perhaps it could report that based on the value it's trying to pull regardless of it's actually able to pull it? I also think it's odd that the power number isn't reporting in the monitoring software, but maybe this is a red herring?

Go to userbenchmark.com and run the test. Link results here after.

 

[glilopi]

Go to userbenchmark.com and run the test. Link results here after.

Here ya go! Asus PRIME Z690-P WIFI D4 Performance Results - UserBenchmark

 

[glilopi]

Of all the stuff you listed, you left out the most important part, the PSU. I am assuming it's a iBuypower brand and that pretty much means it's trash and needs to be replaced.

I went ahead and bought a Corsair RMX 1000 (yes, I know this is more than I need atm). It will get here tomorrow. I'll swap it and see if that makes a difference. If it does, then great. If it doesn't, I'll wait on the new card and probably want to keep the upgraded supply anyways.

 

[drivinfast247]

Here ya go! Asus PRIME Z690-P WIFI D4 Performance Results - UserBenchmark

Your 3080 is throttling. Whether it's heat or power I'm not sure. Install HWINFO64 in let it run in the background as you play a game. What's the highest GPU temp recorded.

EDIT: GPU actually looks okay.
Check your Windows power settings and try setting to performance.

 

[glilopi]

Your 3080 is throttling. Whether it's heat or power I'm not sure. Install HWINFO64 in let it run in the background as you play a game. What's the highest GPU temp recorded.

EDIT: GPU actually looks okay.
Check your Windows power settings and try setting to performance.

Yeah, it indicates that the Power Limit (goes from 0 to 1) is being hit, but the actual power draw is like 60% maximum for any activity. Highest GPU temps I ever see are like 60C... Have tried the Windows power settings, no change.

 

[Styphoryte]

Yeah, it indicates that the Power Limit (goes from 0 to 1) is being hit, but the actual power draw is like 60% maximum for any activity. Highest GPU temps I ever see are like 60C... Have tried the Windows power settings, no change.

You've checked your Nvidia control panel power settings as well? Edit: I'm an idiot, haha! I'm glad you got the dust out of there.

 

[drivinfast247]

Yeah, it indicates that the Power Limit (goes from 0 to 1) is being hit, but the actual power draw is like 60% maximum for any activity. Highest GPU temps I ever see are like 60C... Have tried the Windows power settings, no change.

It looks like you solved the issue. That's an odd one so be sure make it obvious in the first post. This way if someone has a similar issue they can see your solution.

The holes are of no concern. It's the residue that's visible in the pic in the pin slots.