Part of the problem is not many are even aware they have this is issue. Only awareness will generate enough noise for Framework to do something about it. It’s going to be a costly fix if there’s even to be one.
2 Likes
It feels like having a Mercedes-Benz AMG but fastest you can drive it is 65mph!
3 Likes
I don’t want them to go bankrupt, I don’t want to “press for refund/free replacement part” but they should do something about it.
A well written guide on how to re-paste the CPU, for instance. Goes a long way.
As I have demonstrated even with really cheap (and not super high-end) paste, you can get okay results. Certainly better than 35W, and probably a lot more stable. It doesn’t exactly flow easy.
3 Likes
I wish Framework would actually say somewhere what minimum Cinebench R23 score should be expected (under a long list of expectations, eg the 7940HS, a particular Windows ISO with no additional software whatsoever installed which might run background tasks, certain ambient temperature, etc.).
I paid an extra $200 for what I knew was marginal performance increase, but what’s $1800 vs $2000 anyway so I yolo’d it, and I support Framework’s mission.
But then I look at NotebookCheck and see they got 16519, and a minimum of 16352, and meanwhile on my 2nd board I’m getting max 15737pts (albeit with some software installed, no longer clean Windows). And I’m just so tired of this, since they took my computer to a service center for 12 freaking weeks to resolve my macropad not working.
4 Likes
that’s insane. Do they ever told you why/how?
I have been enjoying my Framework 16 (batch 1) and not bothering to benchmark it…until now. The fans are always ramping up even when doing small tasks so I got to wondering, then found this thread.
I can only manage a 12-13k score in cinebench and the power never goes above 33W. This appears to be a problem and is very frustrating considering I paid the $200 more for the 7940HS.
I haven’t used these tools before. How do I tell where the problem is?
Nope, their communication was abysmal.
They wouldn’t say anything for weeks, and then I’d sent a frustrated email and magically they’d have an update the next day. Roughly a month for them to finally admit they’d received the laptop (even though shipper tracking info shows they received it in 2 days), another month for them to decide the laptop was unrepairable and they’d send me a different one — and then yet another month for them to actually send me a laptop. Apparently they would only send me someone else’s broken laptop that they’d been able to repair, and didn’t have any repaired ones on hand. I wish I’d recorded serial numbers so I’d know for sure which parts are new-to-me versus replaced. And after deciding my laptop was unrepairable, I can’t believe it still took them so long to send me something.
Every step of the process was like pulling teeth.
1 Like
Cinebench R23 doesn’t run natively on Linux, does it? I have no idea whether the translation would artificially reduces scores for this particular test or not. Note that OS power mode can make a big difference; Windows Performance versus Balanced profiles aren’t significantly different for me, but Best Efficiency dramatically drops the score.
In any case, I don’t know how to interpret your data either from your screenshots.
On Windows, I use HWiNFO to see temps and power. It will also log data to CSV, so I can use Excel to make a graph of it. I typically look at CPU Core Temperatures and CPU Package Power. If your system power profile was set to prioritize efficiency, I am guessing you’d see lower power draw but also lower temperatures. Since you have only one core at 100C, that seems to me you’re not actually using all your CPU cores for some reason, or your CPU cooler is having major issues and not able to cool your CPU evenly.
Here’s results when I got 15737pts on Windows 11 24H2 after a cold boot of (giving computer several hours to cool down to ambient as much as possible), where I had as many startup programs disabled as I could get (I know e.g. Adobe and Chrome keep running stuff in the background, but it’s not easy for me to disable them and they always re-enable themselves after a few days/weeks anyway), and did a few other things to try and boost score as high as possible without freshly reinstalling Windows.
And to re-iterate, NotebookCheck got a score of 16519 with the system they received from Framework, without having to do any of the PTM7950 and lapping stuff tried by @PSierra117 and others in this community.
3 Likes
Bro how to get a graph like this?
Your Core 3 is overheating which drag your whole CPU down, you can see it with the s-tui graph. With your turbostat output you can see that your cpu power is very low : 33 Watt which is less than it should be.
How do you know they’re actually loading all the cores? Maybe they’re just using core 3 and not actually loading the whole CPU. That would also explain the uneven loading and reduced power draw.
As I said, HWiNFO supports logging to CSV. Then just open the data in Excel, select the columns with data you care about, and make a graph.
I personally use type “Scatter With Lines”, and I also make an additional column with formula to have the X Axis times relative to start instead of absolute computer time that HWiNFO reports.
2 Likes
I have setup s-tui to only show cpu core information:
s-tui has 2 zone : a summaries zone and a graph zone
in the graph and summaries zone there is 3 types of data :
- per core temperature (the Tctl 0 is the total reported by the cpu and it is used for fan control and cpu throttling)
- per thread frequency
- per thread load
As you can see all core/thread are used 100% and the core 3 is overheating. In normal condition the core temps sould be even (or within a few degres, 10°C max difference)
2 Likes
Thank you @Clement_Fouche , this explanation helps a lot. So is your core 3 also overheating? Same core as mine?
All cores are showing 100% usage right there in the graphs. It was in the middle of a cinebench multicore test at the time the screenshots were taken.
yep core 3
OH well, I just submitted a support request. We shall see what they can do for me. This is very frustrating.
1 Like
I missed that thanks. I was confused by “Acpitz” and thought it was a different language, and didn’t notice headings in center of graphs when I personally was expecting them at left of graphs.
I have had my new mainboard installed for only a week, and it has already degraded to the point where I have lost 10-20% performance of my already middling scores when it was new.
As I said in my referenced comment, I got a score of 15,234 and average 51W package power with just my 7940HS when the board was new, now I am getting about 13,596 with average package power around 36W. Core4 is capped at 100C while Core1 is at 69C, nice…
I am still in contact with support as I was already unsatisfied with the replacement board and now that it’s already gotten significantly worse in just a week, we shall see where this goes…
7 Likes
Apparantly Kryonaut extreme is 14.2W/mK.
It is not cheap, though. 2 grams will run you $21.69.
Igor’s Lab report that these are apparantly trash (?)