PNY RTX 4070 Ti XLR8 OC - New worse BIOS/version?

[Einkoro]

While I doubt it was the vbios from MSI the 4070ti super started showing as 8x instead of 16x for the pcie width plus WHEA logging correctable errors across the PCI bus and then outright failed to be detected by windows 11 anymore. Old card works fine in the same system so I suspect it’s a dud and sent it back. Hopefully no one notices the vbios flash.

Honestly after poking around at it sounds similar to the issue der8auer had after a week with a 4090 strix sample that ended up needing reballing with bad pads under the die. Same approximate time to start flaking out and same symptom of the width being wrong only for that card.

Kinda a shame it was super efficient in comparison and even 100rpm more still quieter than this asus tuf. Plus minimal coil whine unlike this one. Not sure what I’ll consider now don’t really want to touch another PNY after the fan curve changes. Suprim X? Maybe just wait for Blackwell at this point.

 

[Evoportalz]

Hi everyone, first thank you zetorson the override option corrected my MSI After burner issue.

I am announcing today that I am developing a small (and cheap) add-on board that will reduce the fan speed by a few hundred rpm. I have ordered parts for the prototype. Basically the board will clip into where the fan headers are and you'll plug the fans into the board. Fan speed will be reduced by either 5 watt resistors or few diodes in series. I'll be pulling out my oscilloscope tonight to test exactly what PWM duty cycle PNY is using to see if indeed they set it to 30% duty cycle at 30%. Of course I suspect it to be more. Once the prototype is complete I will post some pics and results.

Evo

 

[JRR]

Hi everyone, first thank you zetorson the override option corrected my MSI After burner issue.

I am announcing today that I am developing a small (and cheap) add-on board that will reduce the fan speed by a few hundred rpm. I have ordered parts for the prototype. Basically the board will clip into where the fan headers are and you'll plug the fans into the board. Fan speed will be reduced by either 5 watt resistors or few diodes in series. I'll be pulling out my oscilloscope tonight to test exactly what PWM duty cycle PNY is using to see if indeed they set it to 30% duty cycle at 30%. Of course I suspect it to be more. Once the prototype is complete I will post some pics and results.

Evo

I dont know how itl work out or if ul void ur warranty, but good luck bro. Post ur results

 

[Evoportalz]

I dont know how itl work out or if ul void ur warranty, but good luck bro. Post ur results

It's OK, most people do not have an education in technology or electric engineering. Unplugging the fan cables will not void your warranty also.

 

[Evoportalz]

Boyz..... check this out.....

I measured the fan PWM signal at 30% 1450rpm and it is in fact 32% fan speed being sent to the fans (not 30%) and the duty cycle correctly scales all the way to 100% fan speed. See my oscilloscope screen shot for proof. It is indeed the slightly higher than normal duty cycle of 32% and the fans that are not spinning at the correct RPM relative to the 32% duty cycle PWM signal they receive. I was out partying all weekend so I had no time to work on my add-on board but I'll have it completed soon.

PNY needs to lower the duty cycle to match 30% of the fan's max RPM and not just blindly throw a 32% duty cycle signal at it. Either that or they need to use fans that correctly scale their RPM relative to the PWM signal. It is the fans that are not scaling correctly to the 30% (32% real) PWM signal.

 

[Evoportalz]

Hello,

I finally had some time to do more experimenting. I added some resistance to the +12 side of the fans and the RPM is STILL 1450rpm exactly. The duty cycle auto adjusts so that no matter how much resistance I add the fans will spin at 1450rpm. Adding a simple resistor to lower the speed will not work.

My solution now is to make a simple pulse doubler circuit to feed the card double what the actual RPM is. Either that or I'm just going to flash the card with the non-XLR8 version of the card since it's the board.

 

[cheeseman]

I'll just leave it here.

Nvidia 30% and 0 RPM

This is the release repository for Fan Control, a highly customizable fan controlling software for Windows. - Rem0o/FanControl.Releases

github.com

 

[Evoportalz]

I'll just leave it here.

Nvidia 30% and 0 RPM

This is the release repository for Fan Control, a highly customizable fan controlling software for Windows. - Rem0o/FanControl.Releases

github.com

View attachment 346243

This has already been posted in this thread. Further more it does not solve the issue where the fans spin at 1450rpm at it's lowest operational speed (30%).

 

[Evoportalz]

Update: I flashed my PNY 4070 Ti Super 16GB XLR8 to the non-XLR8 version that's hosted on Techpowerup since it's the same PCB board. It was way easier than I thought. Fan speed is now 1300rpm at 30%. It is noticeably quieter now and everything works perfectly including the RGB.

If you want to do this you need nvflashk (not nvflash) link below.

GitHub - notfromstatefarm/nvflashk: Flash (almost) any vBIOS to (almost) any nVIDIA GPU

Flash (almost) any vBIOS to (almost) any nVIDIA GPU - notfromstatefarm/nvflashk

github.com

My goal is to find a solution for 1000rpm at 30% and I will NOT give up!!! So far I'm happy with the results, lowering it just 150rpm made a real difference!

 

[Evoportalz]

I will not give up. I promise...

 

[Evoportalz]

I did boyz I did it!! RAHH HA HA HA HA!@!! Flashed my card with the Gainward BIOS from techpowerup here: Gainward BIOS from Techpowerup

Fan speed is now 1150rpm at 30%. MASSIVE difference from 1450!!!! Card is VERY quiet now and still super cool due to the massive heatsink.

 

[Ynead]

Here is a vbios update sent by PNY itself. 30% = ~1130RPM. There is still an issue with it refusing to go down to 0 RPM when not in use though. Would be really nice if someone found a solution for that.