Gtx 1050ti

[tomasartigue]

It would likely work fine since the flash tools are programmed to support certain chips. Looks like you just got unlucky by not verifying it first. Though it isnt unheard of for AMD or anyone else to use a chip that is supported by the flash tool of another, for cost reasons they tend to stick to certain models and only keep those in there tools internal DB to keep them light.

As long as that version of nvflash (since it sometimes matters) works with the GTX760 you shouldnt have any issues.

Though you could spare yourself the heartache if you already have the hardware programmer, you can simply flash a 1050 vbios onto it manually if you really wanted too.

But because it's an AMD chip and the Nvidia GPU isn't compatible, could it not read it and everything remain the same because it doesn't detect a compatible chip?

 

[Solaris17]

But because it's an AMD chip and the Nvidia GPU isn't compatible, could it not read it and everything remain the same because it doesn't detect a compatible chip?

I have no idea what this sentence means or how I was not clear enough.

 

[StViolenceDay]

1050ti powers VBIOS chips with 1.8Volts (Pascal is the first NVIDIA generarion swithed to this voltage)
GTX760 and RX570 uses 3.3Volts
So their flash chips are not compatible

Also, this may be a reason why the hardware flasher can't read it.

Read yours VBIOS chip marking, google its datasheet and note the size (always is tge power of 2, typically specified in bits for flash ICs, like 4MBit=512KByte). You have to find flash IC of the same size or 2-4 times bigger, those may be compatible.

Also, if they are 1.8V only accordingbto the datasheet - you want be able access them via 3.3V programmer without an adapter (but should be able to access via nvflash after soldering on)

Also, this really looks like a flash problem. The "expected 0x4E" from your last screenshot of initial post corresponds to 'N'. And normal nvidia VBIOS indeed starts with NVGI text 'signature'.

So your problem indeed looks like VBIOS IC problem. To inspect it original contents you can solder it back and save content via
nvflash --save bad.rom or nvflash --fullbackup

Maybe the saved result would be different from time to time. In such case, I'd suspect not the flash IC itself, but maybe some resistor near the flash IC physically hit/damaged, that prevents stable communication. Look for visually suspicious damaged resistor near flash IC/between flash IC and GPU

 

[tomasartigue]

1050ti powers VBIOS chips with 1.8Volts (Pascal is the first NVIDIA generarion swithed to this voltage)
GTX760 and RX570 uses 3.3Volts
So their flash chips are not compatible

Also, this may be a reason why the hardware flasher can't read it.

Read yours VBIOS chip marking, google its datasheet and note the size (always is tge power of 2, typically specified in bits for flash ICs, like 4MBit=512KByte). You have to find flash IC of the same size or 2-4 times bigger, those may be compatible.

Also, if they are 1.8V only accordingbto the datasheet - you want be able access them via 3.3V programmer without an adapter (but should be able to access via nvflash after soldering on)

Also, this really looks like a flash problem. The "expected 0x4E" from your last screenshot of initial post corresponds to 'N'. And normal nvidia VBIOS indeed starts with NVGI text 'signature'.

So your problem indeed looks like VBIOS IC problem. To inspect it original contents you can solder it back and save content via
nvflash --save bad.rom or nvflash --fullbackup

Maybe the saved result would be different from time to time. In such case, I'd suspect not the flash IC itself, but maybe some resistor near the flash IC physically hit/damaged, that prevents stable communication. Look for visually suspicious damaged resistor near flash IC/between flash IC and GPU

So, to clarify... I installed a chip that runs at 3.3V, and the GTX 1050Ti Pascal board only has 1.8V, which wouldn't be enough to power the chip. And if I solder the chip's positive terminal with a cable to a 3.3V power supply on the board, will it work?

So, to clarify... I installed a chip that runs at 3.3V, and the GTX 1050Ti Pascal board only has 1.8V, which wouldn't be enough to power the chip. And if I solder the chip's positive terminal with a cable to a 3.3V power supply on the board, will it work?

nvflash says that this chip is compatible, but as you told me, since it doesn't have 3.3v, it will never work because it never loads its BIOS.

So, to clarify... I installed a chip that runs at 3.3V, and the GTX 1050Ti Pascal board only has 1.8V, which wouldn't be enough to power the chip. And if I solder the chip's positive terminal with a cable to a 3.3V power supply on the board, will it work?


nvflash says that this chip is compatible, but as you told me, since it doesn't have 3.3v, it will never work because it never loads its BIOS.

If you tell me that giving 3.3 V directly to the chip would work, I have no problem, obviously I don't mind losing the board. I'm just testing things out of experience, so it could become bricked without any problem.

 

[StViolenceDay]

So, to clarify... I installed a chip that runs at 3.3V, and the GTX 1050Ti Pascal board only has 1.8V, which wouldn't be enough to power the chip. And if I solder the chip's positive terminal with a cable to a 3.3V power supply on the board, will it work?

No. 1.8V is not only power level, but signals level too, and many things are connected to 1.8V line except VBIOS chip. Finding some flash chip for 1.8V standard is the only way (nowdays they are not extremely rare/new/expensive)

 

[tomasartigue]

No. 1.8V is not only power level, but signals level too, and many things are connected to 1.8V line except VBIOS chip. Finding some flash chip for 1.8V standard is the only way (nowdays they are not extremely rare/new/expensive)

Thanks. Your answers are clearing up a lot of my questions, and I have one that's keeping me thinking. I got a chip (25q64fwsig) and (W25q80bvssig) from an online store (Mercado Libre).
And the one that comes with the 1050ti is (w25q40....)

I don't know which one to buy, and I think the Q64 is 1.8v, but I don't know what voltage the Q80 works at.

 

[StViolenceDay]

According to Winbond the voltage is coded in the letter after 2nd digit group - pciture from here https://www.winbond.com/productResource-files/Winbond_Serial_Flash_Product Brief.pdf

Clock MHz table can be ignored as far as I undetstand (it denotes max clocks, lower is ok).
So, I suppose your original chip is W25Q40E.

Others are not listed in this table, but are listed on the website - https://www.winbond.com/hq/product/...l-nor-flash/?__locale=en&selected=8Mb#Density
So for W25Q80 the 1.8V variants are W25Q80E and W25Q80P

W25Q64FW is indeed 1.8V flash. It 16 times huger capcity then needed, so I'm not sure that it would work, but interesting experiment to try!
My guess/feel is that 1050ti would be able to read it but fail to write it with nvflash (read commands are more compatible then write commands). But who knows!
So if you would be able writing it with your programmer - try to writing it before soldering on board

 

[tomasartigue]

Gentlemen, I can finally say we did it. It took me about $5 to fix it myself, and it's all thanks to you who stayed to lend a hand on the forum. My brother will be very happy, and so will I for him and the experience of getting everything working again.

Gentlemen, I can finally say we did it. It took me about $5 to fix it myself, and it's all thanks to you who stayed to lend a hand on the forum. My brother will be very happy, and so will I for him and the experience of getting everything working again.

I finally installed the W25Q64FW chip, so we can say that the experiment was a success and, although it has an excessive amount of memory, it works correctly.

 

[Gyiyg]

There's also the RTX 5050 that might be just around the corner soon™
Probably anywhere 3-4x the performance of a 1050Ti.
I remember fighting these things with a friend trying to get VR support.

He also could get something like a used RTX 4070 (non super), depending on if his PSU can handle it.

 

[Dr. Dro]

Interesting, so it was just the BIOS chip that went bad? I wonder how that happened. Pretty unusual to me, were you toying with it earlier, flashing random BIOSes?

 

[tomasartigue]

The PC was shut down correctly from the operating system, but the next day it wouldn't boot. It didn't even give any signal to the keyboard or mouse. Everything was very strange. It seemed shorted out until I replaced it with a 3.3V chip. Everything started normally, but there was no video. A guy on the forum told me that Pascal only works at 1.8V. A genius.

 

[StViolenceDay]

I finally installed the W25Q64FW chip, so we can say that the experiment was a success and, although it has an excessive amount of memory, it works correctly.

Interesting. How did you flash it? External programmer or nvflash?

 

[tomasartigue]

Interesting. How did you flash it? External programmer or nvflash?

I didn't use nvflash because I already had the external programmer with the 3.3 to 1.8V converter. But if you want it as data, I'll try using nvflash to see if it flashes such a large memory and we'll clear up any doubts.

He also could get something like a used RTX 4070 (non super), depending on if his PSU can handle it.

This is Argentina. A 4070 can cost between 850 dollars or 1000 dollars, always speaking in dollars.

Interesting, so it was just the BIOS chip that went bad? I wonder how that happened. Pretty unusual to me, were you toying with it earlier, flashing random BIOSes?

This graphic arrived with a number 2 engraved on top of the chip. From what it seems at some point it came back to warranty and it would have already happened but I don't really know

 

[iameatingjam]

Good job. Sorry about what I said before. I thought it wasn't worth fixing but thats not for me to decide.

 

[Dr. Dro]

Good job. Sorry about what I said before. I thought it wasn't worth fixing but thats not for me to decide.

In most cases it isn't, but this one turned out to be an unusual and perhaps comparably easy fix - I had suspected the problem was the BIOS chip when I read the OP and saw that the nvflash was reporting 00.00 invalid in the first post (cards with a damaged GPU core don't do this when they Code 43 - those are unsalvageable), but by then OP had already fixed it. I was gonna bring it up but, was too late for the thread. BIOS chips very rarely ever go bad, even if flashed incorrectly - even those situations are usually reversible if you have a HW programmer.

Nice happy ending to this case