Origin EON-15S/Clevo P150EM GPU Upgrade - 675MX to 970M

Alright, so after a very long wait for shipping, my "new" 970M arrived last week... and it's dead, kind of.

Hope there's no policy around 'necroing' threads, didn't think it appropriate to start new one to document progress started here.

So few things I've learned. The default P150EM BIOS from Origin (and I'm guessing Clevo too, from what I can tell there's no real difference) doesn't recognize a 870M out of the box in BIOS. Not really a big deal, it still boots up with an "empty" VGA slot and detects the device in Windows which is great. Getting older Drivers from NVIDIA which may work with it is basically impossible from what I can tell, however Windows will set it up as a VGA Adapter straight away, and then hunt for a driver via Windows Update if your running Windows 10.

Here's where things started to get interesting. When Windows 10 finally installed it prompted me to restart, but when it came back up, the card was gone. I don't just mean "Windows no longer recognizes this device" gone. Completely gone, nothing could detect the card, not even NVIDIA bloat software, a dozen reboots, driver identifier or GPU-Z. It just stopped working, stopped putting off heat, and 'died'. So card came out, half dead 675MX went back in to test I didn't kill my Motherboard, and it's all good. DDU everything again and start from zero.

Second time around Windows suddenly recognizes there's a 970M in it again, so I let it install it's software since still can't get a driver to work and I haven't learned the tricks of Inf modding yet. Card magically works as a 970M (for the first time) after it's driver install reboot, but things aren't exactly normal, still.

Certain games, applications, etc (including NVIDIA Control Panel) crash when Anti-Aliasing is set too low or off. (Pretty sure that's the first time anyone has ever typed that sentence out, given having AA on is usually a causes for crashes.). NVIDIA Control Panel won't open, at all, and I can't work out how to reset it. GPU-Z also detects three different GDRR5 memory manufacturers for the card, although this may be normal?

I'm a little stuck on what to do going forward. Yes, I technically have a completely stock P150EM with a 970M and it 'works', but it's pretty temperamental. I haven't done a full clean W10 install yet, but it's on my list when I get an SSD since the Clevo donated its to a desktop. Would flashing a new VBIOS help here? Is it worth going the Prema BIOS route (although that concerns me given I'm not using an exact Clevo) and is that likely to improve my stability issues? Or is driver modding to something newer a better solution (and how to I learn)?

 

I recommend Prema Bios for only one reason: you have the Fn + 9 option that puts your fans to the maximum speed. It's as noisy as a freaking jet engine, but you have a guarantee that is going to run cool. My temperatures won't go above 74°C when I activate this thing. It's still not going to detect your 970M on bios (it shows "empty" here for me, also), but I didn't have any issues of disconnecting randomly and stuff. The only issue I had is when I used less thermal pads that I should did and the card was turning off by itself after minutes on intensive games, forcing a restart to be detected again. Applying thermal pads correctly fixed that issue (obviously).

Also, I know there's an EUROCOM v3b BIOS that is also made by Prema, but it wasn't available to public. This bios will detect your card properly (it will show Geforce 970M inside the bios setup and stuff), but besides that, I don't think it has more differences between that version and 1.02.17PM v2. You can send an e-mail to Eurocom asking if they can help you with that, but I'd say it's your last resort if you still have issues on v2.

About drivers, does your card has this same Hardware ID?


If so, let me know. I know exactly what do you have to change in drivers to install correctly.

 

Hey @PhOeNiX_H, thanks for the reply!

I appreciate the info about Prema BIOS, I'd heard about max fan mode however wasn't sure if there were any other advantages such as card compatibility. I've reached out to Eurocom before making my purchase trying to learn and see what support they could offer via BIOS/Drivers/etc and they weren't super helpful because (understandably) they've got a business to run and only offer support if your a customer. I'd rather not change too much whilst I'm getting this card working correctly, however I think Prema V2 will probably be my next step.

I've double checked the thermal pads a few times, and although I'm reusing the original and they're the placement seems to be fine, getting some replacements is on my to-do list as well but they're hard to find in Australia right now. Plus most of the crashing occurs straight away, when it tries to load things rather than actually gaming. I've had Haven running on loop for an hour without any issues, so I don't think it's a power delivery overheating issue.

Now the thing that has me most excited is that apparently I do have the exact same card. Small world I guess, or a common card? I'd done a bit of a search on it and hadn't had much luck on laptopvideo2go so assumed I'd gotten something weird. When you've got the time, I'd be extremely grateful for a walk-through on the driver setup and hopefully that resolves my crashing with AA off issues.

(Copy of my hardware IDs below, just in case I got too excited and misread, but pretty sure we're both running 5105.1558.)

 

Yeah, what I'm worried it's about the randomness of the issue. Let's just hope the card it's not defective.

About what you have to do to install the drivers correctly:

-Go to https://www.nvidia.com/en-us/geforce/drivers/ and download the newest driver for 970M as you are supposed to do.
-After downloading, extract the .exe (RMB > Extract files. You'll have this option if you have WinRAR installed, but I'm almost sure that 7zip also has something alike). It's irrelevant where you extract them.
-After that, go to the path you extracted the files > Display.Driver and look for the file "nvcvi.ini". Open it. Notepad will be fine for this job.
-You'll have to look now for "6555" (without quotes, of course) and check if they appear 3 times and 3 times only. If so, replace them with 5105.
-Save the file and exit.
-Now you'll have to start windows without driver signature enforcement activated. You can see here 3 ways to do it. The first one is easier and it will work, but if you want to keep updating drivers after that without having to do this all the time, the second and fourth way will probably be better options.
-After disabling driver signature enforcement, go to the path you have extracted the files and run setup.exe. If everything worked correctly, you will have the option to install Drivers and Geforce Experience. It will only show a warning in the middle of the process (saying the driver is not signed yadda yadda yadda) and asking if you want to continue.
-Now the annoying part: Windows will keep reverting drivers to the version that comes with Windows Update, which is a 5 years old driver, lol. For stopping that, go to Control Panel, look for Device installation settings and after you click it, change to "No" and confirm. This could give some issues for installing things like external HDDs in the future, so if this happens, just change to "Yes", reconnect whatever you want to automatically install, wait for finishing the installation and change to "No" again. It's annoying and I know there's an option to add Hardware IDs manually to stop Windows Update for reverting back your drivers, but it didn't work for me. There's also a way to change this configuration changing the registry, more details here.

Those are for Windows 10. I have no idea how you should do on older versions (like 8.1 or 7). I think the driver manipulating part would be the same, but I can't guarantee it.

I have no idea what exactly the Prema Bios does besides the manual max speed fan, but seems like there are a lot of stability patches. Let's hope that is enough to solving your problem.

If you want some details about how I did my thermal pad installation, good people helped me on this topic.

I hope that helps you. Please inform us the results

EDIT to adding some stuff that you've asked before:
-Flashing a new vBios could help. In my case, I've used the Prema OC v2. If you need it, PM me.
-Installing a fresh Windows could actually fix the issue and it's a option to try if none of those steps work. Since you had a nvidia card before, it's not that bad, but I still recommend using something like DDU. I can't help you on that, though, because I just went directly to a fresh install knowing those issues could happen.

 

That's probably the most comprehensive and easy to follow Driver modding guide I've seen, and it worked perfectly first go around¹. As a bonus, it looks to have solved my random "AA-Off" crashing issue in at least two programs that I've tested, so that's a good sign. Thanks for the amazing write up.

Unfortunately however after my last post I started noticing some checkboard artifacts in certain games when moving at high speed (like this example, not my screenshot or game though). With the working 461.09 drivers I can lower AA enough that it goes away, so I'm guessing there's more to my situation than simply the drivers. Possibly memory overheating.

New issue aside, I'm a little more confident in the card now, and am considering that perhaps my reused pads were deformed enough I'm not getting good contact on the memory modules since the stock 1mm blue pads are a little firmer and have less give than the 1.5mms. The seller was supposed to supply fresh pads and a few other things, but didn't so I might peruse this option next week and see if I can find anything local, even if they're cheap and nasty. Did you end up going with .5mm pads and stacking, or ordering a mix of 0.5mm and 1.5mm?

Also one thing that's bugging me from your thread, I noticed Khenglish suggested that you'd been supplied a Dell card. Is it safe to assume based on the device IDs that my card is also a Dell (it came with a different style XBracket too, with no standoffs like the Clevo one has)? The reason I ask is, GPU-Z Reports it as a Clevo Card, but I also noticed it's memory seems to be a mix-match of different brands based on it's self reporting, which seems unusual.





1. For anyone reading this in the future looking for guidance, I DDU out of habit whenever installing new drivers, so your success may vary if you don't.

 

Tbh, I couldn't confirm before flashing vBios if it is a Dell/Alienware card, so I can't say anything about that. Yeah, I know it was risky af, but I read about it on many places and didn't see nothing about incompatible bios and stuff, so I just Leeroy Jenkins'd, lol.

Yeah, I got some artifacts on mine using old thermal pads, also. This is probably the real issue in your case. Used thermal pads are like used thermal paste: if they are too old, they'll probably won't work as they should. The good news is: using new pads will probably be enough to solve the issue. The bad news... if this won't solve the artifacts, the problem is a lot more serious. Like I said: let's just hope the card is not defective. Working better with newer drivers is usually a good sign.

And yes, I bought a couple of 100x100x0.5mm kits and I just stacked them up. Meaker said it was safe, so I didn't even bother buying different heights. The only thing that you have to pay attention is about thermal conductivity, because it needs to be less than 11W/m.k. Not following this will result in a hard thermal pad and you will have thermal issues as you have now. Like I said on that topic, I got mine here. I know you are on Australia, so it'll probably not be that easy to get one of those, but you can use as an example.

Information on GPU-Z is relatively the same for me. BIOS version is different (of course) and memory modules are also different, but other than that, it's kinda safe to assume that our cards are almost the same.

Here is all the information of mine:



Ah, I also did a schematic of how I applied my thermal pads. I hope this helps:

 

I'm really tempted to flash that VBIOS now, but equally concerned I'll break something more. The seller is being a pain about the whole thing and insisting it's my fault, but can't and won't prove that the card was tested more than "does it boot".

The Thermal Pads I ordered (Also from Amazon, but a local seller. Pretty sure it's the same product anyway.) came and I fitted them, but graphics checkerboard artifacting still continues. Guessing the card's somewhat defective, but weirdly it only does it in one game that I've found, and runs fine with Haven/Superposition benchmarking. This has me wondering if it's a driver/software issue not a hardware one, as I'd have figured maxing out settings on Unigine benchmarks would have caused it to pop up.

Really appreciate that visual guide for the pads though. It was a little fiddly stacking them but all seems to have gone fine. Out of interest, do you know/remember what your 970M idles at? Mine seems to bounce between 40-43° with an ambient of about 22°C, I haven't seen it hit above 70°C under max synthetic load. It's funny listening to the fan kick on and off every 30 seconds idling though.

Guess I get to decide whether I do a fresh install, flash the VBIOS (and maybe BIOS), or just live with a weird bug in a single game...

Edit:
Decided to flash the updated VBIOS, issue still remains but the card survived. So, guess that's something.

 

For the 900 series vbios is interchangeable between brands as long as the memory size matches.

Gpu-z reports all the memory chips your vbios has entries for. Only one entry is running.

Your artifacts are strange. They look memory related, but this is a kind of artifact you usually only get when a card crashes, and my understanding is yours is not crashing. Memory overheating or not having good die contact around the outer rim of the core die (this is where the memory controllers are) could be the culprit, but I also would not rule out a defective card. Drivers or vbios will not cause what you are seeing.

 

My idle temperatures are about the same. I played Spelunky 2 for a lot of hours and this is what I got:


But reaching only 70°C is really good. Probably because you have the right screws for X-bracket. In my case, I needed to use some replacement since my last card was a 7970M.

And, yeah. Like Khenglish said, artifacts are not normal, but if you get them only in one game it's kinda fine... question mark? It could be bad, because this could be a beginning of a major issue.

Depending where you bought from, you could just ask for a refund, even after seller being an asshole. I mean, you tried to solve the issue "peacefully", it's not your fault.