Mobile Pascal TDP Tweaker Update and Feedback Thread

HI again

I have looked in techpowerup vbios database and found that 3 bioses with 88W TDP

All are 10DE 1C20. I didn't look for 1c60 because i think they are not compatibile
Alienware and MSI bioses looks identical in values in TDP tweaker 1.12
HP vbios have two values higher in advanced section.

But for me the MSI vbios looks most promising but will this be good for GE72VR 7RF when it was coming from GP73VR 8RE

Also I need to ask if the amount of ram matters?? I own one msi lap with gtx1060 3GB version and the bios from GP73VR 8re had 6GB?? Same rops/tmus/cuda but different amount of ram?

 

I'd also like to try this out, but I have 6gb gtx 1060, let me know how it goes.

 

For now I am waiting. Maybe some guru will response about comaptibility.
Also I want to know if the nvflash goes bad I will be able to use it again to flash back my backup Rom with nvflash again.

 

10DE 1C20 - msi gtx1060 vortex
10DE 1C60 - msi gtx1060 laptop

 

vortex and laptop i would say.
I think it should be like this:
1c60 mxm
1c20 bga

My lap has 1c20

 

So anyone??

Is it possible to flash (using NVFLASH) stock vbios from GP73VR 8RE gtx1060 6GB to GE72VR 7RF GTX1060 3GB??

MY stock backup rom and the one from GP73VR 8re
https://www23.zippyshare.com/v/JaI8vf65/file.html

 

Mine gtx 1060('mxm) in clevo p170em came from vortex with devID 1C20.

 

did you flash custom vbios on your clevo laptop? I also have gtx 1060 1c20 made by clevo/kapok.

 

Nope, i didnt flash any vbios. It is running on stock, but but colleagues from the forum had to flash modified vbios to make the card work in p170 smseries.

 

Does anybody flash with the programmer tools a 1080 with higher tdp?

 

Fascinating! Is this a custom VBIOS that I can flash to my Zephyrus M too?

 

I just saved my original vbios with the hardware programmer and modded it with a tdp of 125 to 150 adjustable. After that just flash it back und ddu your drivers for a cöean reinstall. Then check gpu-z if the watts monitor will show up.

Gesendet von meinem SM-G965F mit Tapatalk

 

What is a generally considered maximum safe temperature limit for a GTX 1080?

Also, I ran into an interesting problem . . . flashed my 1080's in circuit with no issues. After some testing I decided that I wanted to up the power limits a little more so I tweaked the vbios and tried flashing again in circuit using Mini Pro. The flash would not take no matter what. It actually ended up corrupting the data on the vbios chips.

I had to unsolder the chips and flash them out of circuit to the new vbios. After resoldering the chips, everything is copasetic.

What would cause both chips to take a flash in circuit once, but then never again . . . but then take a flash out of circuit with no problems? I'm using the tl866cs with a 1.8V adapter and SOIC8 clip.

 

87C is where it will throttle harder and harder.

 

Maybe I'm a bit confused. With a tweaked VBios, with all temp limits set to 91C, and the TDP limit set to 215/258 Watts, both my cards run flat out all the way up to 91C before they start throttling. I think that is how it is supposed to work with these tweaks.

What are people's experiences with pushing that temperature envelope higher? For example, had anyone run their cards at 93C for years with no issues? 95C?

Back when I had 6990m's The hard thermal limit was 110C. I'd ran them at 102C+ for years with no issues.

I guess my question is how conservative is Nvidia being with the stock 91C limit?

 

I would say that the higher the temps the lower the lifespan of the card. Some live longer, some are dying earlier. It is as it is with human. Some are more resistant to stress and some aren't.

Gesendet von meinem SM-G965F mit Tapatalk

 

They run flat out because power limit is lifted off. Before power limit was limiting maximum voltage those cards could reach. Now without power limit those cards can reach over 1V and more voltage = more heat. You need to use something like MSI afterburner and make custom voltage/frequency curve. Default curve just use way more voltage than needed. My 1070 couldn't hold 1683 MHz boost at fluctuating 0.85-0.87V and was heating like crazy. After unlocking card 1696 MHz flat line at 0.8V with 6-7 C dropped temperature.

 

Smaller process and more thermally dense, plus pushing that far maybe gets you 5% extra to run the cards into the ground?

 

If he try run the chips +-95c. What with the graphics power delivery which is cooled by same heatsink? Will follow same direction/path. A short lifespan.

 

I wouldn't refer to the 6990M as an example of longevity... ;p


I edited the temp limit on my 1070s for a while because on the P370EM the fans wouldn't spin to full until something reached >93C so the slave would sit on 90C throttling hard by ~30% with the fans on about 2/3rds, I can't say much other than the temp limit in the tweaker tool edited vbios works and 94C didn't kill them on the spot (as once the slave was able to hit 94, the fans would spin to full then settle in mid 80s without throttle). Rarely saw heavy enough usage to see those temps in games though, cpu was more the limiter

Unlike the 1070, I've not found power limits on 1080 especially in SLI to be that great a performance benefit tbh.

 

VRMs are designed to fet toasty and they can tolerate higher temps than the core. It's the silicon itself which I think would degrade.

 

Gots a question about this. I backed up my stock vbios using ch341a aand modded it with the preset of this software. My mxm 1060 has a dev Id of 1c20 and I have the sliders tto adjustable. After flashing and going into afterburner, the card is still behaving like before, even tho I set the target w to be 90 from the orig 78 The sliders are still greyed out and the w still doesnt go past 50 with the voltage not going past 1.050. Is this normal?

 

Have you ddu the driver and reinstalled it?

Gesendet von meinem SM-G965F mit Tapatalk

 

Did you read back (verify) the chip contents after writing?

 

Voltage shouldn't go past 1.05 V. That's more than needed. Try to play with voltage/frequency curve in MSI afterburner. Try locking frequency at specific voltage. Also is it mxm gpu or bga?

 

Nah. I havent tried that yet. Let me try and I'll let you know

Of course. I dont put it back in my system before doing this needed step

It's an mxm GPU on a 980m board with a bGA chip(Chinese 1060 from ebay)

 

I had the same problem as you cause i didn't do a clean install after flashing. After i cleaned the driver with DDU and did a clean install everything wprked flawlessly.

Gesendet von meinem SM-G965F mit Tapatalk

 

May I ask what driver and section you are using?

 

That doesn't matter. The installed driver didn't read the new vbios settings or didn't apply the changes. Thats why you need to do a fresh install of the graphics driver.

Gesendet von meinem SM-G965F mit Tapatalk

 

I'm just asking for newer drivers sake. I have problems understanding sections as they change them with every release. Reinstalling j95 417.71 now. Just finished ddu

 

Oh okay. Hope that will fix it for you!

Gesendet von meinem SM-G965F mit Tapatalk

 

So a ddu and reinstall of the drivers worked for the power adjust. Does the temp slider always stay greyed out even when you select adjustable?

 

What i know is that the temp slider won't work either. Good to hear that the the power adjust slider appeared now.

Gesendet von meinem SM-G965F mit Tapatalk

 

Yeah but what's weird is using furmark to stress I get power throttling even though it's at only 60w(I set the target to 90)

 

Try not to use furmark. Furmark stresses the vrms way to much. Use synthetic bemchmarks like heaven or valley. For recording the power consumption hwinfo worked really well for me

Gesendet von meinem SM-G965F mit Tapatalk

 

Trying heaven now. Am using GPU z to measure power

 

With hwinfo you can set up the osd showing the power consumption on screen while using heaven or valley . Furthermore you should always keep in mind that your card just uses the mxm slot as power delivery - correct me if i'm wrong.

Gesendet von meinem SM-G965F mit Tapatalk

 

I know. I have GPU z on top of my 1080p heaven bench. With the auto oC from after burner, my clocks are at 1974 under toasty conditions at 1.05v and the perfcap being vrel(voltage) the card still isnt pulling more than 60w

 

here is a log from gpu-z during a firestrike run with a curve on the core. it never goes above 65w and barely goes above 60 at times
i am using an alienware 17 r1 on the a14+ bios and a 4930mx proc. people have used 230w tdp msi 1080s in this laptop with success.

 

Hm the card should consume much more power at this voltage. Just to be sure, you have pulled the slider in afterburner all the way up to it's max?

Gesendet von meinem SM-G965F mit Tapatalk

 

Maybe you can pm one of those guys already did this mod in your machine? Have you set "prefer maximum power" in your nvidia control panel?

Gesendet von meinem SM-G965F mit Tapatalk

 

I did at one point, but didnt see any change. Didnt while running fs

 

No. Good point. Doing that now.

 

What % card utilization you get?

 

So it went from 1228x to 12466 in firestrike making that change in the control panel, with just very slight increase in power draw, spiked at 70 once, then hovered around the 60 mark

98 to 100

 

Seems fine. Have you tested Firestrike before and after? Graphics score should have increased quite a lot.

 

it increased from 11518 to that, but thats after an oc with afterburner, even before this tdp mod

 

As I understand you got alienware. Try as this guy did: http://forum.notebookreview.com/thr...other-socket-victory-against-bga-crap.803637/

There's section The GPU power throttling issue

 

i saw that, there isnt a setting in my bios for disable adapter warnings unfortunately, but i havent seen anyone else with my laptop sying that this needed to be done anyway. doesnt the 980m use 100w of power?

EDIT: there is a setting for this in my bios, i went back just to make sure i wasnt being foolish. it is now disabled. trying the testing again

 

Still not going past 60. i wonder if its just reporting wrong. the card does get to 80c and to just shy of 2ghz on core while benching and stressing with oc