Graphics Card upgrade for M6800 ?

Very interesting, ha. Good job getting the card working again!

I thought I saw someone else recently get code 43 when flashing a different model vBIOS onto a card. It seems that these cards have a security feature to prevent cross-flashing like this. Maybe you got around it by flashing while the card was "online" and it only kicks in after it is properly power-cycled or reset somehow?

These cards also verify the signature on the vBIOS at startup (new to the Pascal generation). I rather doubt that any vBIOS that you modify will load.

@RMSMajestic is trying different 1070 vBIOS's in the M6700... Maybe he'll find a good one... See this thread.
http://forum.notebookreview.com/thr...-quadro-p5200m-and-gtx-1070m-on-m6700.827595/

 

I had to remove the card to flash it with the programmer. There's quite a bit of discussion of flashing modded mobile Pascal bioses here: http://forum.notebookreview.com/threads/mobile-pascal-tdp-tweaker-update-and-feedback-thread.806161/

I haven't looked into the source of that tool much (it's in github as well), but assume it's doing something similar to rom-fixer to preserve the checksum.

 

There is(or at least there was) no source for the tool, the source zip is just the exe.

 

Pls elaborate "TPU". I'm a novice with most of this

 

TPU = TechPowerUp
They have a vBIOS database.
https://www.techpowerup.com/vgabios/

 

980M upgrade, good news & questions. Using info from this, and other threads (THX!) completed 6800M GPU upgrade to Clevo 980M today. I know it pales in comparison to the 1070 and such upgrades some of you are doing, but I'm pretty stoked about it.

I encountered and, thx to some posters here, was ready for some upgrade issues. The Clevo 980M X bracket takes a smaller screw than the Quadro version, so I swapped X brackets. Both used a thin double sided type tape rather than glue, so it was a fairly easy swap. Used Artic MX-4 and some Gelid thermal pads. Also, as noted here, the Quadro heat sink arrangement doesn't line up with the 980M at the lower portion of the card. I used my benchmill to carve out a slot in the copper sink to better mate with the 980M.

Reassembled the 6800 and it started normally, win7 detected the 980 and upgraded the drivers. Ran a userbenchmark to compare with one I had previously run with original Quadro GPU. It reported the 980M was "performing below expectations" at the 33% percentile. GPU Z snapshot included. Any ideas about the below expectations?

 

Here is a run from my Quadro M5000M from before I upgraded to the Pascal card. My DX9 results are quite a bit higher than yours. The cards are very similar so I don't think that the difference should be this much.
https://www.userbenchmark.com/UserRun/12460917

Here is a trick I have learned with NVIDIA. Go to the NVIDIA control panel, "Manage 3D Settings", and find the power management option. Set it to "prefer maximum performance". You can do this system-wide on the "Global settings" tab, which I do not recommend because it will use up battery power and run the fans more, or you can do it only for programs you select on the "Program settings" tab. I found that if I do not do this, for applications/games where the load can come in "bursts" rather than all of the time, the NVIDIA card will often clock down when it doesn't seem like it should and performance can suffer.

 

If you just let windows install a driver, you should download and install the updated Nvidia drivers.

 

As suggested, uploaded the drivers from Nvidia and set the GPU for max performance. Also disabled vsync to run the benchmark. Similar reading as before. Looked around at some other 980M tests and saw some similar, average results on benchmark. It's still 3-4 times better than my previous Quadro gpu, but odd that it doesn't score better in benchmark. Temps are good, fan activity isn't much different than with Quadro. Only thing I see on gpuz is the pcie bus running at 8x3 vs card capability of 16x3.

 

Also strange, I get x16 for PCIe in my M6700 (for both Quadro M5000M and P5000).