Upgrading my Ranger

It is not the liquid metal. The iGPU cores are under the same die as the CPU and the die is impervious to it. I would say that based on your testing in the system that uses Optimus, the 4930MX CPU is defective. It should not be using the iGPU in your Ranger, but it apparently is for some reason. The fact that the problems all go away with the 4710MQ installed and follow the 4930MX to the MSI is ample evidence that the CPU is no good.

 

Thats the first time I have seen an issue like that, so even with the iGPU disabled its still misreading the sensor ?

You've done the requisite trial and error, it looks to be the CPU is on its way into forced retirement. Shame the 4980hq doesnt work as I still havent gotten it into a platform yet :/

 

Alright then, time to put this sucker out of its misery. I'm not too surprised it died though, its been running at high voltages and with high heat its whole life lol, it held up quite well. I'll probably just frame it on my wall as "the first CPU I have managed to break" or something funny.

4980hq in the Ranger is interesting though, something I'd like to put on it's bucket list alongside getting a 1080 to work and forcing the MXM slot to give more power. I don't think much would come of it though, but this thing could kick ass again if one could find a way to get the 4980hq to work. L4 cache seems to be really, really underrated for some odd reason, dunno if its price or just die space that's stopping intel/AMD from using it. My mother has an iMac with a 5775R and only 8 gigs of RAM, and its still holding up damn well after 6 years.

 

Price from the manufacturing and also to the consumer were way high. Also doesnt seem to do squat for synthetic workloads so for all intents and purposes just seems a more expensive CPU with a good IGP. Never could get it working even with updated microcode. Someday might get a P157SM-A for science but not for a while yet

 

The strangest part of this ordeal is that the iGPU should be disabled when running directly off the dGPU, so the iGPU sensors shouldn't even pop up. Do you see these with the 4710MQ installed?

I personally don't see any iGPU sensors, although this is within Intel XTU. I see you're using hwinfo to read the sensor data, so I'll go ahead and look at that myself, then update this post with my findings.

Edit: Yeah I see the temperature, power, and voltage for the graphics cores in hwinfo. That's weird. The iGPU isn't even being used (power and voltage are both 0), so it shouldn't even show up. They don't show up in Intel XTU.

 

Just for sanities sake, see if you can manipulate IGP in Throttlestop when in eDP, if no change then its a bad sensor. If it does change then thats truly an odd scenario.

 

Well, right now, I'm running my 4710mq's iGPU voltage at literally -1.00mv, and the CPU still works fine, the graphics core temps always sit around the fifties, even if the cores of the CPU are really high.

I'll send a screenshot of my Ranger with the 4710mq once I get back home.

 

Here's what HWInfo shows me power wise for the 4710mq:

 

Didn't read all 76 pages but to me sounds like that 4930MX is bad.

You can always try for a 4800/4810/4900/4910MQ CPU as well. They can be turbo bin OC by +4 multiplier.

I run my 3920XM @ 4.2Ghz for daily on my M18xR2 and it works well with the Quadro P4000 (between 1060/1070 in performance). Haswell @ 4Ghz should be enough for 1070.

 

Hahahaha, I don't blame ya.

The main issue is that I really want to set some records specifically with the 4930mx as the multiplier is fully unlocked, hence why I didn't go with a 4900/4910mq. Not really a big deal, just a bit sad and frustrating that it died on me quite early.

 

Boys, I think I figured out a genius idea for cooling the 1070 VRM's without having to use a cooling pad or raising the laptop off of the ground.

Sketch is quite rough, and I've been up for quite too long, so I'm not in the best "thinking" state, but I though of this:



I would take out the useless DVD drive and then add a heatpipe (orange) to the VRM area, and then getting it to the side of the DVD slot where I would get it on a fin stack of some kind (gray). I would then put a fan (green) at the fin stack, and then get a splitter for the fan header (blue). I would then cut some holes in the side and then cut the bottom panel a bit so the fan could breathe. This would allow the laptop to take up its intended footprint without the need for a cooling pad.

 

Alright, for those who didn't see in the Overclockers lounge, here's my CBR20 record with my 4710mq:



If HWBOT allows the microcode hack, I may purchase a BIOS programmer and go nuts, but we'll see

 

Alright boys, after tearing apart the entire laptop twice, switching through my displays, and re-seating the quadro, it did not work. In fact, it was so broken that I couldn’t even get NVFlash to acknowledge the cards existence lol.

I’m not really angry, but just impressed that a card this physically simple can refuse to work at all. The only thing it was capable of doing was getting really hot, and that was it.

So I guess i’ve learned my lesson, to not buy any MXM card from Ebay ever again, as the past 3 times that I have tried, they have not worked in some form or another.

If anyone has any lower end Kepler MXM cards to potentially sell to for cheap, I would be interested. If not, I always know @Reciever has the “check if my MXM slot is working” 860m, but I would like to refrain from borrowing it as it wouldn’t be used for its intended purpose of testing hardware.

 

Im not going to do anything with the m6800, at this point its clear I just dont have the level of interest to pull away from other projects so I can send the cards back over to you if you want them. Just cover the shipping and you can have it.

 

If you're sure you wont be using the cards any time soon, I wouldn't mind paying shipping to get 'em back for a bit, just PM me whenever you have time and we'll sort something out

 

Nah, havent even crossed my mind been too focused on other things.

 

I've noticed that my 4th core is usually a few degrees cooler than the rest for quite a while, and I really hope its not the heatsink. I may need to do some stuff to the backplate as I have a feeling it may be that.

 

I might put a bid on this 4930mx on Ebay, but is there any useful information that can be found on the etchings on the sides? i.e. Batch number or date produced?

 

That damaged die doesn't look too good

 

Oh damn, nice catch. I had assumed it was something the die was reflecting, but the more I look at it, the more I see that that might be damage.

 

Alright, I'm going to acquire some pressure paper shortly as the large difference across each core temp keeps bugging me

If the cores are super different to do uneven mounting pressure, (which is what I assume is the problem as I have noticed the bracket seems to be shorter on some sides,) then could adding washer to the bracket be a possibility. I remember @Reciever mentioning this earlier in the thread, but would adding washers be a good idea?

 

No that's a bad idea, warped my heatsink lol

 

Do you know of any other things I could do? I would like to buy another bracket, but I don't think I can find one, and I don't even know if I can take the bracket off lol.

 

Maybe if you could source some of those o rings people use for mechanical keyboards that are more gummy in nature that would allow for a bit of play for you to fine tune the pressure? Or just grab a wide rubber band and use something akin to a hole punch and put maybe two layers per screw so you have some resistance but not enough to warp the heatsink.


I used the washer trick for the y510p but it didn't work for me on the ranger