RTX 4000 performance problems on clock speed
RTX 4000 performance problems on clock speed
I recently assembled my initial computer. I own a RTX Quadro 4000, which I purchased used but the seller claimed it had been rarely used. I plan to use it for CAD and 3D modeling, though I noticed on YouTube that it handles several games well, so I started playing LEGO Pirates of the Caribbean (it doesn’t demand much graphics power). Suddenly, I heard a buzzing noise coming from the GPU. I assumed it was the fan, but it wasn’t coming from there—it spun freely. After searching online, I realized it might be coil whining, which is annoying even if not critical. I ran a FurMark stress test; despite constant 99% usage and an 80°C temperature, the GPU produced no sound. Later, thanks to RTSS, I found it was running at 1100mhz at maximum, compared to 1920mhz in the game. This discrepancy suggests the issue lies with voltage/frequency settings. I attempted undervolting, but MSI Afterburner refused to cooperate—it instantly returned to default after changes. I suspected a BIOS blockage and considered modifying it to undervolt. It’s not very straightforward. My question is: why does the card operate at 1100mhz (or higher) during stress tests (99% load, 80°C) but remain silent, whereas in-game it reaches 1920mhz (71% load, 60°C)? Isn’t a stress test meant to push the GPU? Does the game seem to be more demanding than expected? The card should ideally run at 1100mhz (up to 1620mhz with boost) as per its specs (https://www.techpowerup.com/gpu-specs/qu...4000.c3336). The ticking or rattle sound appears after around 1600mhz. Was the card overclocked by the previous owner? If so, how can I revert it to stock settings? I have limited experience with this hardware. Do you have any insights about Quadro series cards? Thank you in advance for any advice.
Furmark and its related versions are referred to as "power viruses." They place significant strain on the graphics card, which often leads modern drivers from AMD and NVIDIA to limit performance. What you're seeing likely reflects the card functioning normally in a game scenario. You're demanding high frame rates within POTG, which is why coil whine appears. Lowering your frame rate should alleviate the issue (consider using V-Sync or a frame rate limiter). For stress testing, avoid Furmark and its derivatives like MSI Kombustor. Instead, try programs such as Unigine Heaven or Superposition benchmark. GravityMark is also recommended.
Additionally, boost clocks typically rise when the card reaches its 3D power profile. Higher boosting is normal, indicating the card is ready to increase performance. Any overclocking occurs within the software and isn't inherent to the hardware itself.
According to the TPU link, the standard boost frequency is 1545MHz, but the card can exceed this if it supports it under certain conditions.
Last updated: January 21, 2024 by Godlygamer23 (Added info)
Thanks for your input! I haven't used V-Sync before, but I've checked it in other games and benchmarks: in Forza Horizon 4 the card still causes issues (coil whine, same settings as LEGO), yet it maintains 50/60 FPS. Since this card is mainly for CAD work, I ran a Blender benchmark. The photo attached shows the frequency remains high, matching the game, but the card doesn't produce sound. Watts stayed consistent across tests. I plan to try V-Sync and other stress tests as you suggested. Your concern about FPS is valid—being noisy only in-game is frustrating. Thanks again, EDIT: I've also tested Unigine Heaven with normal settings and with V-Sync enabled. In the first case, the coil whine made a weak noise (see stats below), while in the second it was quiet. I'll try V-Sync with Lego and FH4 and will update you. The fact that FH4 ran at 50 FPS is worrying; if it's really an FPS issue, why does it still sound odd? Are there other factors at play? Apologies for the repeated questions—I'm just learning and hoping to find a solution. Best, EDIT: January 22, 2024 by WorldSolemnya
I limited my FPS and applied Fast VSync, which made everything run smoothly without any whining. Lego handles even 300FPS without issues, but FH4 had trouble, so I set it to 90FPS using the Nvidia Control Panel. I changed the cap each time to discover the optimal setting for every game. FH4 was stable at 50 or 60FPS before and after capping, which is confusing—I think it relates to power delivery. I don’t have all the answers, but it helped! Thanks a lot for your advice!
I have no restrictions on temperature, not that I've consciously defined any. I'm unsure how to achieve this. In every test or benchmark, the GPU never exceeded 80°C. Now, with capped FPS, performance stays around 60-70%, and temperatures remain at 50-60°C.