Would you be willing to run a Cinebench R23 run?
70+w just feels “off” - considering the max TDP is “supposed” to be 54?
Did you change any settings to increase the max?
I agree, I’m running those now. I’ll add them to this post when done
** So I have a theory about the previous results having run R23 runs.
I’ve a feeling the AMD 780M usage is tied to the CPU package power in HWInfo’s sensor, thus when Heaven was running, for some reason the iGPU was contributing to the graphics bench. If this power usage reported as additional CPU power, despite the CPU drawing less for the CPU benchmark, it would show something really high.
Unfortunately, that doesn’t leave me with much options to really stress my 65 & 75C limits, since I can’t force the fan to run without another heat source haha.
Regardless, these are the R23 scores at 65C. CPU only used 23W the whole run with no fan, a terrible score as a result - 11441
Then I tried running R23 with Heaven in the background, this scored even worse, though the reported W was much higher. Probably a combination of an entire other benchmark in the background, plus my suspicion of the iGPU draw, score was 9866
Funny, increasing the temp limit to 75 with Heaven, gave a similar score to only R23 at 65 (11632)
Finally, I ran R23 alone, but with the UTU built in extreme preset (100C limit, full power limits)
I observed the power draw leveling at ~52W during this run, with a final score of 15556.
This is right in line with other results here, so I’d say the 0.5mm shim works just fine.
Probably no need to get a 4th heatsink though PSierra haha!
3 Likes
Awesome! Not questioning you in the slightest - if you have an 80w CPU I think you truly struck gold… dont ever sell it lol
Well that looks reasonable, Okay when you get 80w Package Power its including the Power for the GPU Part. The Problem on that Die is, that the Transfer Area for the CPU alone is not that big and if the CPU alone pulls 80w its hard on the Limit of Physics, but the Big GPU Part + the CPU Part pulling 80w together is doable for the Heatsink. I assume you CPU alone is doing about 25w in your Full Load Scenario. Thats the Reason for the Good Temps.
1 Like
Ahh that explains the wattage discrepancy.
FWIW I did an r23 run with mine set to 65 and pulled up to 33w
Well my replaced motherboard from my other laptop was getting decent thermals, but the replacement one doesn’t. Before it was consistently getting 16k in Cinebench R23 and now I’m only getting around 15k.
The max core temp is 100c and the lowest core temp is 92c. I don’t recall the other laptop’s diff, but I don’t recall that big a difference.
I do also recall it hitting 4.3 GHz all cores consistently, but the replacement is at 4.16 GHz all cores consistently.
So clearly the other cooler or liquid metal was better applied than this replacement laptop.
It’s always something…
I’m going to cut my losses. My OG laptop has my dbrand skin too, which I’m going to lose out. At least the display is slightly better (I also give up on this), and the touchpad and spacers seem to be fitting better.
I’m just going to wait for the PTM to come in and just replace the one that’s on here. I don’t want to have to deal with swapping motherboards only to have to remove it again anyways to replace the TIM.
I may try swapping the display and the touchpad/spacers to see if I can keep my previous chassis with the dbrand skin on? I dunno… so much to try now that I have two FW16s to swap parts to see what fits better.
1 Like
With the 7840HS CPU on the FW16.
We normally see cores 0-3 cooler than cores 4-7 in the amdgpu output.
Where are the cores on the die?
Looking at the FW16 laptop, are 0-3 on the left and 4-7 on the right or what?
My thought here is, when one removes the heat sink, which part of it are we expecting to be badly conducting heat, and thus which part are we expecting to lack paste/LM etc.
Is this affected by the orientation of the laptop when in normal use or in a backpack etc.
So I discovered with support that I have a way better score on CB23 without my dGPU installed (14500). But is that what’s to expect? I had 15k with my new mainboard and dGPU.
My problem is not the score or the performance itself. I don’t really care as for right now. I just can’t stand the raging fan noise all day long if not on saving mode (windows). I could go without dGPU to make is a bit less noisy, but my docking station has the cable management for the USB-C display output on the dGPU.
Max TDP of my CPU without dGPU is 48W with an average of 40W (That I find acceptable).
With dGPU it drops to 30W average (same 48W peak, but it’s in the first 10 seconds then it thermal throttles).
Support is as helping as possible but I’m at the “just clean your fans” step again.
2 Likes
I closed of Parts of the dGPU Vent Holes with black vinyl (roughly 5/8 on each Side (begin closing from the inside near the USB C Port as the highest frequency Noise is coming from the inner side vents)) That reduced the Fan Noise drasticly and it changes the pressure and Airflow to push alot more Air through the CPU Vents at the same Fan RPM.
2 Likes
I raised a similar question a while back and got crickets:
Swapping back to my old laptop, performance got worse! What used to be 16k is now at 14.5k and the delta between coolest and hottest core is larger!
Coolest core is 88.6c and hottest is 10c. Clearly the TIM was pumped out.
I guess it doesn’t matter…we just need to wait for the PTM TIM that Framework will send us to arrive.
Kudos to Framework to not give up on their customers after the final sale for sure.
1 Like
If you don’t want to wait - the thermal grizzly has been working great for me. Its only like $10-15 for a pad.
1 Like
@knipp30 If you don’t want to wait - the thermal grizzly has been working great for me. Its only like $10-15 for a pad.
And the Thermal Grizzly Phasesheet lasts for 4-5 “repastes” of a Laptop 16 
2 Likes
Yessir!
I used one sheet to do:
My CPU
My GPU
My steam deck
I have enough still to likely do:
My wife’s FW CPU and GPU
I’d be more than happy to. I’ll test again in about 6-8 weeks
Just did that, don’t know if that’s a placebo but it’s been an hour and I find it already better. Fans don’t speed as much and even when going crazy it sounds less annoying.
1 Like
can you pleas elaborate on this a little bit more? I tried to run Cinebench R23 alone but got horrible results Perhaps this preset would help?
Its definitly not a placebo. The Airways for the dGPU are a little bit too big so the fan is loosing alot of its airmass through there. This airmass is more needed for the CPU. I didn’t test GPU Temps but they should be that bad. The GPU was never above 77C originally when using 100w.
I don’t think it’ll help you in this case, from your own report, the CPU was at 100*, so it was thermal throttling at that point.
Universal Tuning Utility is a third party software to adjust the AMD CPU’s power and thermal limits. It cannot really set limits higher than the max thermal throttle designed to safeguard the CPU however, so if you’re hitting 100*, no matter what the max power you set, it will throttle until it caps at 100*. The extreme power profile in UTU basically just unlocks all power limits aside from the thermal throttle, so it should force whatever the maximum performance is possible out of the heatsink design/efficiency.
You’ll probably need to repaste your default LM with a PTM pad to get better results. If you have some experience with this, it’s not too difficult, but if not, I’d recommend trying to get a RMA.
This is a good point. Did you apply the vinyl on the outside like a wrap, or block off the interior?
The GPU never gets anywhere near hot, so if it’s quieter and better on the CPU, I’ll give this a shot