I have raised Support ticket already - waiting for resolution. But looks like there similar case.
If the clocks are roughly the same here it certainly looks a bit fishy
Something is really wrong there. It looks like you are only scoring just above 9k with a 7940. With my 7840 I am getting like 15,300-15,500.
Today i replaced my Mainboard and now the Temps are much more even and i get the expected 15300 to 15600 on CB23 MC (I thought it should be above 16000, but thats the 7940)
before it was consistently under 15000 as low as 14300-14500
1 Like
So, the new one also have one core sticked to 100C ?
I got ~15200 in the first ~20 seconds, but then it lowered to ~10k (I believe due to overheating and lowering CPU power from 45-52W to 35-39W).
Example with new run:
Yes but it stays at 48-52W during the 10 min Loop.
The old Mainboard was falling to under 40W
Allcore Clocks stay between 4,3 and 4,1 GHz during CB23MC
1 Like
Here i did 2 Benchmarks in CB2024 with the new Board.
There you can even see my old Score and the new Score. The Difference is massive. But i must admit, i modified the new Mainboards Cooling Solution further and am now Running PTM7950 instead of the Framework Liquid Metal. And i get better Scores, than with the Factory new Mainboard. I Think the LM + Heatsink is not perfectly thought after or there is a Quality Problem with the Liquid Metal Sheets.
I did some Shots from under the Heatsink and the whole assembly and what i did for further Modification in the Batch 20 Thread
How was the process to remove the liquid metal?
I actually thought about “downgrading” to a PTM7950 pad - but keep talking myself out of it.
Not that easy. The Heatsink was pretty glued down with the Border Sponge and the hardened PTM LM. When it was of, i scraped the Surface with the Flat Plastic Side of the Framework Screwdriver until it broke free as a whole plate. Same on the Heatsink side. All the Glue and residues got wiped with Cotton Swaps and Isoprop.
3 Likes
I got replacement motherboard and after change it I got much better results, however CPU still raised to 100C on one core. I have also ordered new heatsink - will try to replace it with PTM7950 or Thermal Grizzly LM (I have some left from previous Razer Blade 15 application).
What’s great now - on stress test I got 50-54W continuous power now instead of 37-39W on previous motherboard.
CPU-Z OLD: AMD Ryzen 9 7940HS @ 4467.47 MHz - CPU-Z VALIDATOR
CPU-Z NEW: AMD Ryzen 9 7940HS @ 4638.9 MHz - CPU-Z VALIDATOR
Cinebench for new after ~10 minutes (note all cores 80-90C and only one 100C with 53W - I got 15244-15697 score now, instead of ~10k before):
UPDATE:
After ~hour the Core having 100C temps go down - maybe LM have to settle like PTM?
CPU-Z: AMD Ryzen 9 7940HS @ 4689.93 MHz - CPU-Z VALIDATOR
Is it still that low? Would be crazy if so. The PTM LM needs Burn in to align and squeeze out the not needed Material. But more than an Hour Full throttle is to long for an average User.
The burn in should be done with normal use and most users who now see that there is not the right performance have their Devices for more than 2 Months now.
Yes, after several cycles of stress tests (5 min stress, 5 min idle) within ~hour I have now much better temps - no 100C anymore (max ~95-97C and other cores on 90-92C) and power goes to 53-57W on stress for 5 mins.
Note: I have Ryzen 9 (7940HS). Best score on Cinebench 23 is 15729 though…
Okay thats roughly the same score i get with my 7840hs and i hit 100c well currently 99,9 
my best 10min Loop was around 15950 and my best single run 16363
1 Like
So, after few days of testing, results are even better (in terms of temps), however Cinebench 23 score about the same (~15600-15700).
Look at the max power and max temps for cores:
That’s my results on Cinebench 24 (also on new/replacement board), but with Ryzen 9 (7940HS):
And temps (note: 10+ mins for multicore it was continuous 54W with ~96C max!):
I have not been using my laptop that much past months (wfh on desktop) but recently had to go out more and use my FW16 more frequently and for longer periods of time.
I started to notice that it wasn’t performing as expected.
I started monitoring temps and seeing one sensor much higher than the rest.
I just did a reboot and close all apps and did the r23 benchmark.
If I am reading the other posts correctly, I have an issue?
During the benchmark the temp4 goes easily to the high 80/90’s while the rest stays at 50/60s.
On Fedora40 while doing the benchmark in performance profile and wall powered.
Well yes thats definitly RMA worthy. Its a Minimum of 3000 Points off of where the 7940hs should sit. Thats nearly 20% missing Performance.
1 Like
Thx, I figured. temp4/core4 is always 10-15c above all the others. I also found the fan is ramping on the most basic things.
Update, I will get a replacement motherboard.
2 Likes