M6800 Owners' Thread

See: http://forum.notebookreview.com/threads/precision-m4800-cpu-upgrade-tdp-concerns.826646/
@unnoticed's posts specifically. The information basically applies to all models from this era.

 

Dear M6800 owners,
I have this strange problem with my M6800 running i7 4810mq / 16GB Corsair Vengeance 1600cl9 / Quadro M5000M / Windows 10 Pro / 240W power supply. The issue is when the gpu is under 100% load (AAA gaming or heaven 4.0 at extreme preset for example), after 10 minutes or so, the cpu starts to throttle. I was initially thinking that probably the motherboard isn't able to deliver all the juice to the gpu so the cpu is trottling(not immediately but after 10-20 minutes approx.) to as low as 1.2ghz or even less. I have done the repasting with AS 5 and the temps are totally under control for both GPU and CPU. If I stop the load / game, within the next few minutes the core frequency is regained back to its turbo levels.
I have attached a (file).csv of HWinfo64 logging that captures the above behaviour.

My doubt - there are people that installed 1070n in this workstation in this forum and I remember that the 1070n is power capped at 120W TDP while this M5000M is TDP capped at 100W. This 100w limit is hard locked as adjusting this via vbios editor in the power tuning page doesn't alter it. I have confirmed it via gpu z info. However, if a 120W is okay to use under 100% load why not this 100W card? Why is it pulling down the cpu frequency? Is dell motherboard intelligently thinking that 'since the card is pegged at 100%, let me reduce cpu freq. to (do what exactly)?'

Can someone please help me with this?

 

I can't comment on the 1070n or M5000M but when I first installed my 980m, I would get a lot of throttling after a few minutes. My solution was to use a vbios editor to remove the 980m turbo boost completely, and now the cpu and gpu don't throttle anymore. So, maybe the motherboard is power-limit throttling?

 

Thank you for quick reply. My GPU is throttling based on TDP limit which is fine. I'll try removing the 3d base and boost (will disable) and keep TDP Freq as entry #1 and entry #0 in trial one and trial 2. Will see if that makes any difference. I also noticed that while the cpu throttles, in HWinfo it says "on-demand frequency %" will reduce from 100 to as low as 10ish which leads to .45ghz cpu frequency(checked it in task manager and HWinfo)

 

use throttlestop to avoid any throttle, need to force clock modulation and something else to 100%, multiplier at max, also you could undervolt it

 

I checked as said above and it's still the same. Also tried disabling speed step and c States in bios. This time clocks are reported as 3600mhz in HWinfo but once the on demand clk changes, the throttling happens and the true clocks are displayed in task manager. So no change here.

 

Hi! Thanks for the suggestion. I used throttle stop a year ago or something. I'll check with that clock and chipset modulation checkboxes and set it to 100. I will reply once I try that. Fingers crossed..

Edit: tried it out. It's still fails to maintain frequency.

 

Did you turn on throttlestop ? Also try to force speed shift at "0"

 

Yes what do I say about the blunder that I forgot to turn-on TS. Sorry for the confusion caused. I made the GPU stress out at 100% (while temps less than 72c) and the on demand clock reduced resulting in cpu throttling. At around 1.4ghz I opened TS and check that clk modulation to 100% and turned it on. It works.

 

One temp to check for CPU Throttling is PCH Temperature, in my case if it shoots above 89 - cpu would turbo less, sometimes even throttle.

Location of PCH is really dumb, it sits behind GPU (opposite side of MBO) but not cooled in any way with black tape-ish material over it.

I have my cooling pad with 2x 140mm fans connected via USB and PCH temp is cca 20c lower and I've never seen throttling since, even if I return stock vBios of 980M that goes well over 120W while card turbo boosts.