Dell Precision M6700 Owners Thread

So I've been wondering something about the thermals. I have great thermals on my K5000M with a 135MHz core overclock, like sub-70 C at max bore and the CPU has never gone above 82 C, even in Cinebench at max load. But I am curious about two central thermal mechanisms.

The first is chipset cooling. In my research I have never seen a thermal pad or paste on the chipset despite the aluminum CPU heatsink contacting it. Would there be any tangible benefit from putting paste or a pad on it or is the thermal transfer already decent enough?

The second is VRAM cooling. I have a +500 MHz overclock (+250 MHz effective) on the video memory and it sits very stable in games and GPU stress tests, but I am somewhat concerned about half of the VRAM chips lacking active cooling. Do the two large thermal pads attached to the back of the K5000M transfer enough heat to not cause problems further down the line?

(Edit with notes on overclocking: If you are a numbers person and, preferably, a tech nerd, I highly recommend using a simple vbios mod to overclock your GPU. If you have a Kepler GPU in the Dell Precision M6700, you can follow this guide I followed on the LTT forum.

Link:
https://linustechtips.com/topic/105...-kepler-and-probably-maxwell-ii-bios-tweaker/

For those that are more tech savvy, here are my own tips. Before jumping to the 1006MHz clock speed used in the guide, I would recommend setting something like a 900MHz clock, flashing the vbios and then using something like Nvidia inspector or MSI afterburner to fine tune the clocks on both the GPU core and memory. This will allow you to maximize your performance to match the binning of your graphics chip. I was able to get to 1020 MHz with this method, granted I settled on the suggested 1006 MHz due to stability issues. Memory clocks could go as high as 2500 MHz, a 1 GHz overclock from stock, but a 500 MHz OC up to 2000 was stable and all I needed without the risk of instability. I would also recommend plugging in your memory clocks in the vbios after fine tuning the GPU, despite what the guide may say. It will seriously improve performance in many applications; however, on Dells with Optimus, I am under the impression you need to have Optimus enabled to flash the vbios. Once finished with vbios modding, you are fully free to disable Optimus (or leave it on if you're into that) and enjoy your much stronger GPU to its fullest.)

 

The Kepler Quadros have a lot of room for overclocking. There are modified vBIOS images for the K5000M that can help out with this. You'll probably hit the 100W power limit well before you have to worry about it getting too hot.

Regarding the vRAM, memory has less need for active cooling. My system came with thermal pads covering the bottom vRAM chips but I think that they are basically useless without a heatsink to transfer the heat to. Still, it's never been a problem. I don't know about overclocking... It should be fine if you are getting a stable system, and if it causes problems, just bring the speed down some. (Though I imagine that you would quickly see diminishing returns when overclocking the memory, compared to overclocking the actual GPU core, so do some measurements and don't set it higher than you really need...)

My system did come with a thermal pad covering the PCH/"chipset" and I've kept it there every time that I have torn down the system for an upgrade. I don't think it's that big a deal but I think that you should add one. Heat won't transfer to the heatsink very well without a pad or paste... You can't guarantee "direct contact" without something like that.

On the other hand, if your CPU is not going above 82C under sustained load then I question if it is running at full turbo speed? Seems like it should definitely hit upper 90's running in the mid/upper 3 GHz range for an extended period of time.

 

[Redacted] (I found out what I was doing wrong, don't need help anymore.)

 

I had some concerns on my 6700, enough so that I switched to the Firepro 6100m, with finer nm cpu, in order to keep things in the cool. This was back some time ago, still using Artic paste. Since then, I've switched over on my 6700/6800, and now 8770w, to Gelid pads (ultimate), with a lot of good things to say. I normally now build out the GPU/Heatsink to flush (with conductive below non-conductive, case you are stuck with conductive paste)(since I've switched gpu's so many times, habit), and these bad boys are both easy to use, and have granted a strong 10 degree drop in operating/load temps (even started dropping a bit between the CPU/chipset). I used to pull high 70's even with the Firepro, and 80+ on the CPU, now it's more like 65-72 under load, and rarely have I seen the CPU dip into the 75+ (Although the dual/4thread CPU's seem to be pretty terrible about spiking)

Just a drive-by, but the Supermoth got the Tesla M6 working on his 6800, and I've tried it on both my 6700 and 6800 with success (In case anyone is hunting news on graphic card options on a budget). Hoping to test on both this weekend with benchmarks, but easily outdoing my older GTX980m and quite a bit more budget friendly.

Quick guide here :
http://forum.notebookreview.com/thr...m-working-updated-switchable-graphics.835018/

 

Is 82c a bit hot for a GTX 970m with stock clock speeds? I was expecting that it would be lower than this. I am using Corsair Tm30 thermal paste. I kept the stock cooling pads, maybe I should have swapped those out too and go for all new fresh ones.

 

This is about what I see on the M5000M under high load ...
This isn't an issue for a mobile GPU, the throttle point is 100C (and you'll likely hit the power limit well before you hit that temperature).

 

Lol let me guess, the max wattage is 100W? I've seen that tossed around on here, it seems like my graphics card doesn't draw more than 102W. (With light overclock)

 

Yes, with the Maxwell cards it is 100W. The power limit is set in the vBIOS. I'm not sure if you can override it on GeForce cards (you cannot on Quadro cards, to my knowledge). But higher power is possible in the system... Pascal cards can draw around 110W.

 

Speaking of GPU's, has anyone recently used Optimus in Windows 10 and, if so, how is the experience? I have been tempted to switch over to Optimus for the better battery life and (somewhat) quieter operation, but I am also worried I will run into a variety of application-specific bugs and issues. I used Optimus for a short period of time when modifying my K5000M vbios but saw that a couple benchmarks had problems switching the graphics card (which was a little off-putting); however, I saw no major issues when playing games and video playback seemed to be fine. I didn't spend enough time with it to really get acquainted, so I am wondering if someone else has been using it more extensively and, if so, if they would recommend it.

 

No problems on Win10 enterprise and Tesla M6 here, with switchable graphics mode enabled and compute/graphics mode option available. It wasn't the most straightforward of processes though.

I wrestled with Win10 pro and the 6100m quite a bit beforehand. Mainly, understanding & installing the pertinent drivers and BIOS, before ever attempting any DGPU upgrades. Win10 pro had a inherit issue with signature blocking a lot of NVIDIA driver packages as well, which in the end led me to disabling win10 driver reinforcement service. I, similarly had a setup for a bit, running the single DGPU, and having both bottlenecking and weird frame issues and CPU temp spikes.