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

Memory Clock At Slowest Speed

Joined
Dec 31, 2018
Messages
2 (0.00/day)
Location
Chicago
Hey all.

So I managed to corrupt the VBIOS on my Gigabyte 750 Ti low-profile card (GV-N75TOC-2GL) when I shorted out the fan header. Long story short, the header no longer works, but the card still does. I was able to install a second GPU and flash the BIOS on the 750 Ti using TechPowerUp's "Gigabyte.GTX750Ti.2048.140801.rom". The card now works, but the memory clock is locked at 405 MHz. I checked the BIOS using the Maxwell BIOS Tweaker and it's set to 2700 MHz, which is correct. However, it won't go past 405 MHz no matter what I do. I haven't messed with any Afterburner settings and the ROM is unedited. I tried resetting the GPU using "nvidia-smi -rgc" via an elevated command prompt, but that didn't work. I'm thinking there may be an incorrect setting somewhere in the BIOS, but I don't want to mess with it too much and cause further damage. The only thing I tried was to edit the clock speed to 1350 MHz and that actually worked. But anything over 1500 MHz and it defaults back to 405 MHz. Also, setting it to 1350 MHz gives me the same framerate in FurMark as leaving it at 405 MHz. So even though Afterburner says it's running at 1350 MHz, it's behaving like 405 MHz.

What could be causing the limitation on memory clock speed? I figured if the short is what did it, it would have killed the card outright and not done something wacky like limit the memory clock.

The photo shows usage while running FurMark at 720p with no MSAA. I'm getting 11 FPS whereas I used to get over 40.
 

Attachments

  • memory_clock.gif
    memory_clock.gif
    18 KB · Views: 60
  • gpuz.gif
    gpuz.gif
    23.8 KB · Views: 73
Joined
Feb 22, 2022
Messages
548 (0.66/day)
Processor AMD Ryzen 7 5800X3D
Motherboard Asus Crosshair VIII Dark Hero
Cooling Custom Watercooling
Memory G.Skill Trident Z Royal 2x16GB
Video Card(s) MSi RTX 3080ti Suprim X
Storage 2TB Corsair MP600 PRO Hydro X
Display(s) Samsung G7 27" x2
Audio Device(s) Sound Blaster ZxR
Power Supply Be Quiet! Dark Power Pro 12 1500W
Mouse Logitech G903
Keyboard Steelseries Apex Pro
Electronics do not have a singular failure mode. You have probably damaged your card by shorting the fan header. In this case the result is that the memory is running in "safe mode".

Test it in a different computer if you have that option. Since there is a possibility the problem is software related.

But it smells like a hardware issue to me.
 
Joined
Aug 30, 2006
Messages
7,199 (1.11/day)
System Name ICE-QUAD // ICE-CRUNCH
Processor Q6600 // 2x Xeon 5472
Memory 2GB DDR // 8GB FB-DIMM
Video Card(s) HD3850-AGP // FireGL 3400
Display(s) 2 x Samsung 204Ts = 3200x1200
Audio Device(s) Audigy 2
Software Windows Server 2003 R2 as a Workstation now migrated to W10 with regrets.
The card probably detects "no fan" and therefore is in ambulance mode, like the engine management light in a car: you can still start and drive the car, but it is limited to a low maximum RPM at very reduced power. It allows you to limp home. The "no fan" or "damaged fan controller" detected is allowing the GPU to operate at its absolute minimum allowing you to "limp to your diagnostic".

The card now works
Does the fan work? Properly? Can you monitor fan speeds?

I think you've probably damaged the fan controller, and there is little hope without replacing parts on the GPU board. And I have no idea what the circuit diagram for that fan controller is, and if it is on- or off- the actual GPU die. It could be all over.
 
Last edited:
Joined
May 20, 2020
Messages
177 (0.12/day)
Location
WorldWideWeb
System Name MPG
Processor i3-10105F
Motherboard Z490 Gaming Plus (MS-7C75)
Cooling Be Quiet Dark Rock Advanced
Memory GSkill RipJawsV 32Gb 3200
Video Card(s) GTX1060 6Gb AERO ITX OC
Storage NVMW WD Black SN750 Samsung EVO 840 SSD and Mechanical TBs...
Display(s) Asus VS278Q
Case CM HAF-X
Audio Device(s) ALC1200 Yamaha AV Creative DTT2200
Power Supply Corsair RM-1000
Mouse SteelSeries Rival
Keyboard MS Sidewinder X6
Software Windows 10/11
"when I shorted out the fan header"

And by that im sure that the gpu didnt suffer any other damage...lol
Oh its just the fan header...
 
Joined
Dec 31, 2018
Messages
2 (0.00/day)
Location
Chicago
The card probably detects "no fan" and therefore is in ambulance mode, like the engine management light in a car: you can still start and drive the car, but it is limited to a low maximum RPM at very reduced power. It allows you to limp home. The "no fan" or "damaged fan controller" detected is allowing the GPU to operate at its absolute minimum allowing you to "limp to your diagnostic".


Does the fan work? Properly? Can you monitor fan speeds?

I think you've probably damaged the fan controller, and there is little hope without replacing parts on the GPU board. And I have no idea what the circuit diagram for that fan controller is, and if it is on- or off- the actual GPU die. It could be all over.

I've run the card without a fan plugged into the header for years. I only just shorted it out a few days ago when I accidentally touched the contacts on the back of the PCB. But the reason I do not believe it is running in full "ambulance mode" is the settings in the BIOS for P08 show that the core clock should be locked at 810 MHz and the voltage locked between 893.8mV-956.3mV along with the memory clock being locked at 405 MHz. However, since the core clock is still running at 1032 MHz and the voltage is well over 1.1V and the perf cap reason is VOp, I sincerely do not believe it's running in full P08 mode. I feel like something about the updated VBIOS is keeping it from running at 2700 MHz because it also only sees 1024 MB of RAM and not the full 2048.

If anyone out there has a GV-N75TOC-2GL with the OEM BIOS and not the updated one from TechPowerUp's database, would they mind sharing it? I'd like to try rolling it back to see if that helps before scrapping it.

Update: Case closed.

I found the original VBIOS file by doing a Google search for the card which brought me to the "unverified" version 82.07.55.00.71. Having looked at the GPU-Z for that card many times, I immediately recognized the 55 and the 71, so I gave it a shot. Lo and behold, it was the correct VBIOS for my card. It is now reading all 2GB of RAM and is running at 2700 MHz on the memory clock.

Let it never be said that persistence doesn't pay off. When several people tell you a board is fried but your tenacity kicks in, you can accomplish almost anything.
 

Attachments

  • gpuz.gif
    gpuz.gif
    24 KB · Views: 40
  • memory_clock.gif
    memory_clock.gif
    18.3 KB · Views: 34
Top