P955ER throttling

It stutters because of MSI Afterburner's hooks into the driver when monitoring power draw.

Disabling afterburner eliminates the stutter.

 

I spoke too soon.

The cause of the stutter is setting anti-aliasing to high and above.

Dropping it to medium fixes the stutter.

Source: https://steamcommunity.com/app/582160/discussions/0/1480982338948636354/

I can replicate the same stutter and fix on the older P751DM2-G. Will upload a new video soon.

 

Funny that my GTX 980M doesnt stutter even tho the FPS are low on maxed settings.

 

And what are your anti-aliasing settings, again?

Below high.

 

everything maxed

 

In that case, make a recording of the same and post it on Youtube. If possible, use MSI afterburner for ingame stats.

 

need to find that specific moment, can you provide save file?

 

Start the tour (in Alexandria). Main menu. The landing area is where the benchmark was started.

 

See this for more details on the tour: https://support.ubi.com/en-GB/Faqs/000031846/Discovery-Tour-Mode-of-Assassin-s-Creed-Origins-ACO

Expect no combat. Just exploration.

 

Related, here is what I get when rendering some footage in Power Director 15 on the Q6.

Using the CPU only. Stats included below.

 

Don't be too sure about that.

Let me show you what I get on the Q6 with AIDA64 loaded up, and with a combined CPU,FPU+GPU stress test.

 

Just check the FPU+GPU stress test boxes. More heat and higher Power consumption

 

Here comes the tears.



Deliver me from this quagmire, O Lord.
For the lights are dim and the temps are high.
(And this is with LM).

Take a look at the maximum value under Total System Power.

 

Lo and behold, unstoppable throttling.



Abominable. LOL

 

Oh lord

 

Which I guess explains this horrendous result:

 

And it gets better

Hey, Clevo: Time your ODMs considered an actual recall.

Even with liquid metal, this is beyond salvation.

 

This gives @Papusan PTSD

 

Gramps (played by B. Cribbins)!! @hmscott, an image from the past of Doctor Donna.

 

Jodie Whittaker Gets to SDCC | Doctor Who
http://forum.notebookreview.com/threads/jodie-whittaker-gets-to-sdcc-doctor-who.821395/

Of course, my favorite Doctor's are from many generations ago... you never forget your first Doctor.

Spoiler

Season 12 Trailer | The Collection | Doctor Who
Doctor Who


Doctor Who: Season 10 - TV Launch Trailer (1972-1973)
VERITASERUMUK


Doctor Who: 13 Doctors Ranked From Worst To Best
WhatCulture

 

would Liz like this choice, or would Harry, Sara Jane and Romana?

 

And with Battlefield 1: Pretty decent improvement so far!

 

I doubt so.

You scored 15k in FS.

I can;t get past ~12k on the Q6, even after LM + 200W PSU: https://www.3dmark.com/3dm/27895918?

 

Kill background tasks, fresh driver install? Your CPU score looks fine, it's the GPU that's holding you back.

 

At this point I'd suggest getting the old PSU back.

The EC kills the GPU more off than before. It's so low right now it's scary.

 

Going through the same **** right now.
Liquid metal, and a 200W PSU.

BIOS settings: PL1 and PL2 are at stock.

Let me show you a horror story.



I may have botched my application, so I'll re-open it today and inspect.

 

Right before tj max.
Now that's some premium quality notebook right there. It's fine tho, this forum will spread how fantastic clevo garbage is and mislead more people into keeping a company alive that should have taken its head out of its bum a long time ago.

 

The PSU isn't the limiting factor, but have you actually seen different performance with the 200W one vs the 150W one?

Are you using CCC and if so which version (new or old) and which setting do you have it on?

 

And for these curious:

Yes, the heat sink is shared, ie the CPU + GPU share the same pipe layout(s).

Why this is a bad idea: Take a look at a relatively light CPU load, say, 7-zip.

See what it does to the GPU temperatures?



That GPU on IDLE does not exceed 40 degrees.

Yet when the CPU alone is loaded up, the GPU's temp goes up, steadily. What devilry is this?

And yes, this is after a repaste with LM (following all the recommendations here).

 

*Slaps the Eurocom Q6*

"This thing can reheat your coffee faster than your microwave!"

 

The P955/P950 isn't meant for liquid metal. The CPU heat sink has a tripod design and cannot deliver enough pressure to the CPU die in order to stop the liquid metal from degrading. Repaste the laptop with GC-Extreme or IC Diamond and see what temps you get.

 

Uggh... I'm sorry to see that your particular model has the tri-pod, shared heat design.

But from most posts we've seen here, @RampantGorilla is 100% correct, the tri-pod designed heat sink is better served by a thick TIM.

 

It's a nice feature

 

Hmm. Will do.

Which of the two thermal pastes above does a better job, in your opinion?

It's the first time I'm coming across these brands.

 

Really?
https://imgur.com/a/2GjuqVs

 

Yes, really.

The heat sink is one piece, joined by copper piping.

 

I've edited my post with an link to a picture of the heat sink taken from this video.

 

The MSIBOOK Gt75 ain’t much better. Even worse because it’s contain an so called unlocked i9 (+$600) intended for overclocking. But barely able to run 3.7GHz. Yoo see... Boiling at 99C.
You beat it with only 1C I will say it’s close race

One of the pipes above graphics occupy 1/2 of the rear heatsink grills for Cpu.

 

Yep, its' a one piece heatsink assembly.

@Papusan and @RampantGorilla would a copper shim help with a tripod-based heatsink like this one?

I intend to use a copper shim and liquid metal to compensate for the flaw mentioned above. Would that be feasible?

 

If the fits is very bad and with low pressure maybe it could help. But you should soldered the copper shim to the copper cold plate. Otherwise it will be needed with several layers of Liquid metal. On die, shim bottom, shim top and heatsink. Not a good solution.

Or you could try Alienware’s fix for the TRIPOD mess. Bending the legs for better pressure and fits.

Maybe sanding the copper heatsink cold plate can help.


Edit. Not much will help if the heatsink grill and fans is at its border and can’t remove (suck up) more heat.

 

You have a point. When I opened it up, the LM had a "bubbly" appearance, similar to what it'd look like if you touched it "lightly".

Taking this into consideration.

 

If the fits is bad or all too low pressure from heatsink on die, a thick thermal paste will work better. + mentioned above. Not many other options. Maybe you can find a better heatsink. Not all is in equal bad quality.

 

I'll look into the recommendations given above for the thicc thermal pastes.

 

And don’t forget the pads. Often a problem they are too thick. Check imprint from components.

 

The imprint seems minimal.

 

Update: Hey guys, I think I found out where the issue is: Clevo Control Center.

Hear me out.

I uninstalled it earlier today so as to set up the FanControl in Obsidian's control station, rebooted, and then ran the usual AIDA 64 stress test. I had it running for 10 minutes and stepped off to make coffee, and on returning, I observed that there were no throttle flags being triggered. I did not make much note of it, so I proceeded to install and activate Fancontrol and extras, and then jumped into BF4.

BF4's results seem quite promising. At most, I got a peak of 83 degrees Celsius for the CPU and about 77 degrees Celsius for the GPU, and that's after over an hour of gaming.

Then I proceeded to make a short AIDA64 run, using @Papusan 's recommended testing methodology, shown below.


I'll leave it running for a few hours and report back. Had to stop it as I'm working on a video editor at the moment.

Another observation: Obsidian's fan control, with the automatic fan profile, subjectively sounds "louder" and "flatter" than the Automatic profile from Clevo.
The fan curve is smooth, and does not "hiccup" unlike the one implemented by CCC.

 

Battlefield 4 retested on the same machine.

 

Yes, but the CPU and GPU heat pipes aren't shared meaning that CPU won't heat up the GPU and vice versa. For example look at the XPS 15:


The CPU and GPU share the same two heat pipes meaning they will heat each other up. This is not the case with the P955.

 

And a short BF1 test

 

Try encoding an video using x265 in Handbrake. It will give you a more accurate idea of CPU temperatures when the CPU subjected to a real world workload.