Question Help wanted with configuring GPU

mauporte

Honorable
Sep 22, 2017
16
0
10,510
So I just recently built a new PC (my first one in 6 years and the very first time that I have some of the newest components (ish).
CPU: i5-13600KF with an AIO (DeepCool LS520)
GPU: PNY XLRB 3060 Ti
RAM: Kingston Fury Beast DDR5 5200Mhz (2x16GB)
ROM: 2TB NVMe
Mobo: Asus Prime Z790-P wifi
PSU: Gigabyte P850GM

I updated BIOS, installed a legit copy of windows 11, updated all my drivers, downloaded all my software and games and then enabled XMP.

I then booted up games like RDR2, Cyberpunk 2077 and Kingdom Come Deliverance, and even in the lowest settings for those 3 games, I barely get 24fps.

After running a series of tests with Unigine, afterburner, etc. I have narrowed down the problem to the GPU clock being stuck in 270Mhz. Temperatures never reach more than 43°C and the card is receiving 335W; I have no idea

Can somebody help me?

PS: I'm not overclocking anything yet (except ram i guess) supposedly this setup shoudk run those games without overclocking
 
Last edited:

Eximo

Titan
Ambassador
From where did you purchase the GPU?

Hardware Info 64 may give you a reason the GPU is performance limited. If it really thinks it is pulling 335W, then it is likely power capped, but it should not be getting anywhere near that power draw being a 200W nominal card.

A vBIOS update may be a solution, but it might be safer to return the GPU for a replacement.


----

2TB NVMe is not a ROM, just an SSD.

EEPROM is what the BIOS is stored on. A true ROM is static data that can't be changed. Been a while, but you might find that on a very old PC with Basic or something loaded onto it. Updates required replacing a socketed chip (Though these were also typically EEPROM).
 
PSU: Gigabyte 850GM
I corrected it: P850GM

Not fully sure, but according to this PSU tier list, you PSU model comes under the Tier E "dangerous" category. Assuming you have the exact same model.

https://forums.tomshardware.com/thr...er-list-rev-14-8-final-update-jul-21.3624094/

Gigabyte | P-GM [non-Aorus] <=850W [1]. But again like I said, I'm not sure if your PSU model also falls under this exact same category/tier or not.

Tier E • Potentially dangerous in multiple scenarios

• Armageddon | All units
• BR ONE | BR-X
• Corsair | VS 2012 [orange-label]
• Cooler Master | Elite - Realpower - B2
• Diablotek | All units
• EVGA | N1 - N2
Gigabyte | P-GM [non-Aorus] <=850W [1]
• LEPA | MaxBron
• NOX | All units [2]
• Raidmax | Non-rated units
• Rasurbo | All units
• Solid Gear | Neutron
• Tacens | All units
• Trust | All units
• Ultron | All units
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
From where did you purchase the GPU?

Hardware Info 64 may give you a reason the GPU is performance limited. If it really thinks it is pulling 335W, then it is likely power capped, but it should not be getting anywhere near that power draw being a 200W nominal card.

A vBIOS update may be a solution, but it might be safer to return the GPU for a replacement.


----

2TB NVMe is not a ROM, just an SSD.

EEPROM is what the BIOS is stored on. A true ROM is static data that can't be changed. Been a while, but you might find that on a very old PC with Basic or something loaded onto it. Updates required replacing a socketed chip (Though these were also typically EEPROM).
The thing is I bought in Mexico and PNY doesn't respond to any of my attempts to contact them and doesn't seem to give support here (but I am still trying)

I'll try Hardware Info 64, didn't know of its existence

I updated BIOS to the latest version before I even installed windows 11 and this was last thursday, in asus's support site they only have 2 firmware related things: BIOS and ME firmware update tool; I am not sure how to manually update just vBIOS, but this sounds about right to me as a solution.

old habit, back in the 90's rom was used a lot in the general public to just refer to storage
I booted Cyberpunk right now, Nvidia Performance Overlay gives me this:

Render Latency 32.3 ms
CPU Utilization 12%
GPU Utilization 99%
GPU Clock 285 Mhz
Memory Clock 7201 Mhz
GPU Temperature 42°C
Fan Speed 1299 RPM
GPU Power 339 Watts
GPU Voltage 0.7 Volt
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
You can check your serial number against this:


thanks for this, my serial number is 20193G001615, so it's safe according to that list, but truthfully, it would surprise me the least that they would send defective units over here without care
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
They have a BIOS update tool on their website, just select your card:


I got a "no need to update" after I ran the tool provided for my card. I also ran HWINFO64 and the only warning that is shown to me is that secure boot is not enabled in the os, but I this is my first time seeing these menus
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
I'm also noticing that even though XMP is enabled my 5200 ram sticks are apparently running at 2600, but maybe this is idle speed or something I've never really paid much attention to ram in the past because I have never really gone over the standard before
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
PerfCapReason is what should tell you what the GPU thinks is the problem.

Power, just like what MSI Afterburner tells me, that the limiter is power, but at the same time, according to nvidia performance overlay the card is sucking up more than 330W. In the meantime, GPU-Z tells me that the power draw is 168.2% TDP (200x1.682=336.4W). So how can power be the issue?
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
What display resolution are you using? 4k?

Can you run userbencmark test and post link to results here?

Nope, this is FullHD
I have some results from Unigine Heaven Benchmark 4.0, I'll run userbenchmark shortly

FPS: 26.4
Score: 666
Min FPS: 14.6
Max FPS: 62.1
Renderer: Direct3D11
Mode: 1600x900 8xAA windowed
Preset: Extreme
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
Can you run userbencmark test and post link to results here?


UserBenchmarks: Game 29%, Desk 113%, Work 31%
CPU: Intel Core i5-13600KF - 117.9%
GPU: Nvidia RTX 3060-Ti - 15.8%
SSD: Adata LEGEND 800 2TB - 146.2%
SSD: Kingston SNV2S1000G 1TB - 264.3%
SSD: Adata SU630 960GB - 31.3%
RAM: Kingston KF552C40-16 2x16GB - 160.2%
MBD: Asus PRIME Z790-P WIFI
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
Faulty GPU power sensor is certainly possible. It won't throttle up because it already thinks it is running maximum power.

that makes sense, but puts me at the mercy of PNY (not only do they not reply to my submitted ticket, but they also don't allow me to login to my account that I created when I submitted the ticket, it tells me "You are not authorised to access this portal").
 

mauporte

Honorable
Sep 22, 2017
16
0
10,510
You have high background cpu utilization.
Might be some malware or cryptominer running.
Scan your pc with malwarebytes anti malware.

Can you show screenshot from GPU-Z ?
(upload to imgur.com and post link)
Might be fake rtx 3060 ti.

no, I had some things running I restarted and redid the benchmarking, still the same problem with the card but none with the cpu
besides it is a fresh install