Displayed core frequencies wrong on M1 Ultra #2329
Comments
|
looks like it does not calculate correctly for ultra. Does it change at all or always stays at 600? |
|
hmm, please you that build and share the logs: #2285 (comment) |
Have never seen a different value than 600 MHz.
The distribution of workload across cores… I haven’t checked it against say Apple’s utility (tricky because it constantly changes over time). But I think it is ok. This workload is one single application exporting (converting) many files by distributing the work over many cores. Probably the worker threads go to Performance Cores, while the coordination is done by a thread running on Efficiency Cores. Somehow I see (on both MacOS, but also Windows) that a single threaded load is spread out across multiple Efficiency cores. This is probably (if we forget the other Performance cores) to spread the thermal load better across the surface are of the chip: better to have 4 cores (2 per physical M1 Max chip) dissipate 25% of the energy on average than to have 1 of the force dissipate 100% of the energy. Again: not important for this use case where most of the power goes to Performance Cores. And I still think the shown graph and bars in your app are ok. If really needed to figure out what’s going on, I can do testing on the M1 Ultra if you spec what to test. |
Where do I find the logs? Path or link to instructions? |
|
you can find logs in the Documents folder |
Ok. Hmmm. File date is May 19th 2024. So probably not what you are looking for. Do you expect me to run a special build? |
|
Looks like the M1 Ultra identifies itself like you expected in your code: 
|
|
Yes, please read the comment I share and run the build from that. |
I first tried 2.11.24 release. No change: still 600 MHz. |
|
Special build of 2.11.23.
Content (light load on machine): |
vdhamer
changed the title
|
hmm, it looks similar to m4. Try to run the latest release. Update, oh no. Because the latest release fix it only for the m4 platform. I will make a separate build for your case |
|
please try this build: |
600 MHz now became 600000 MHz = 600 GHz. Hmmm. 
|
|
ok, so please run command: |
|
it will looks something like that: |
|
|
thanks, I will try to adjust code to these values |
|
Out of curiosity, how can I trust that your software is not writing to such registers (they look like settings relevant for overclockers)? No direct reason to be suspicious. But somebody should ask these questions... |
|
what do you mean by |
|
@vdhamer The "reg" in "ioreg" stands for registry not register. See I don't think any user level code could directly change CPU frequencies or voltages, etc. If it could, that would constitute a bug in macOS. Overclocking is not something that is done on Macs, especially not on Apple Silicon. And given the tight integration of hardware and macOS as well as the security of macOS it would be exceedingly difficult to achieve. |
|
oh, there is no writings to any registry, I'm not even sure macOS allows it. Also I would like to know apple silicon could be overclocked) |
|
I don't overclock anything, and recognize that overclocking is usually for serious gaming. Which is so far hardly popular on MacOS. But I suspect that the values you are looking at are a) written by the CPU into hardware registers, b) because it makes the design more tunable for Apple in late stages of development. The fact that Kapersky writes "On a Mac, keep up to date with new releases. Mac updates don't just add new features – they also remove malware, including rootkits. Apple has built-in security features to protect from malware. However, there are no known rootkit detectors on macOS, so if you suspect a rootkit on your device, you should reinstall macOS. Doing so removes most apps and rootkits on your machine. As above, if the rootkit has infected the BIOS, it will require a repair to fix – and if the rootkit remains, you may need to buy a new device." My point is that if you adjust anything that isn't really meant to be adjusted by end users, you are at least taking a risk. Arguably even a fan speed reduction, unless a too low fan speed gets compensated by CPU frequency or voltage reduction. My point is that all this starts to be scary stuff as soon as you reconfigure anything. So the question is how safe it is for end users. Are you only using public and thus approved APIs? Any (even temporary) changes to settings other than fan speed? |
|
Again as @fiwswe it's not about writing, but about reading. Im just curious why you want to overclock Mac, and where you hear that if popular on macOS (cause I never heard about that at all)? |
|
Basically if you aware about the system I can only recommend to not use any app outside the AppStore at all. PS: Stats is not published in the AppStore because:
|
|
For what it is worth, I agree with your remarks: I would stress to users that it is (with the exception of the fan) that is a read-only tool. But it would be good for users and for visibility of the app to have an App Store version. While keeping the source code on GitHub if you like: good for getting help, good for trustworthiness. The two disadvantages you mention are actually both opportunities:
Distribution strategy: What you could aim at: charge a small one-time price via the AppStore (it costs you USD 100 yearly just to be able to distribute via the App Store). So charge USD 1-2? Anyone who wants to avoid that, can still use your .dmg image files. Most will see this as a risk and an inconvenience. Benefits of App Store: you reach more people, it looks (and is) safer for non-experts, it looks safer for IT people, you get some income beyond Patreon, you might get featured by the AppStore if you do a great job, you reduce the update frequency to say 10 releases per year instead of 50. Drawback: you need UI changes to make it Mac like, and you may be asked to remove the fan control feature from the App Store version (if it violates a rule). Alternatively: create a basic (low cost, simple) version which is enough for 95% of the users. And a Pro version with all the settings you provide for nerds. This avoids trying to be 2 different solutions for 2 audiences in the same app. Just trying to help: I have no personal benefit if you make these changes. I don't use fan control. I don't use the advanced features. My own (unrelated, mainly iOS) app is both App Store and Github but free and has no Patreon option. |
|
there is one problem: I need to spend like 3-6 months to make app fit AppStore regulations. And still a lot of features will be not available, so it's not worth for me at all. Also I'm not interested in AppStore, I have literally one app installed from that: Xcode. |
|
update: I have take a look and I have even accepted build in App Store (build from 2020). But for some reason not published it and drop idea with App Store. Also, every build is signed with app dev account. So it's not like I need to buy a developer account, I already have it and use it to sign every build. I was thinking today about that, and maybe I will make another approach with App Store. |
|
I checked the naming “Stats” for 5 minutes. There are more apps that use “stats” to mean “we give you easy access to info”. So you are not alone there. I tend to stick to the classical meaning: “when there are many instances of X (e.g. patients, cars, user sessions) any aggregation of data across all these instances (e.g. life expectancy of patients, duration of user sessions) can be called statistics: you describe a population by summarizing data (average, sum, std dev, distributions…). Your 1 line summary in GitHub (“macOS system monitor in your menu bar“) explains well what the app does. I would edit it to “ macOS system monitor on the menu bar”. But that is just a detail. Important to have the app name, slogan and icon well aligned. So the name should echo what is in the slogan. If you change the icon, users may think it is a different app, so best to keep the icon, or to change the icon a year after a name change. I like that you are thinking with an open mind about putting your app on the App Store. Say you sell 1000 @ 1 $. It might be worth more, just an example. Then Apple early 1000 $ - 30% for you. I expect you don’t get that kind of number via Patreon. And if the app ever starts showing up in reviews or being featured, the numbers might be 10x higher (there are not too many macOS users, and as you mentioned, the App Store there must be very quiet compared to iOS). |
vdhamer
changed the title
|
I've seeing 600mhz on my M1 Ultra too, was there a new build that fixed this? 
|
|
Not that I know of. But I provided some requested logging data and system settings to Serhiy, so I guess it will be fixed in a release or two: the issue has a label “Needs to be fixed fixed” ;-)
…Sent from my iPad
On 17 Jan 2025, at 21:40, Carl Moebis ***@***.***> wrote:
I've seeing 600mhz on my M1 Ultra too, was there a new build that fixed this?
image.png (view on web)<https://github.com/user-attachments/assets/1d7884ef-e920-4ea7-a8c2-ca9addace69d>
—
Reply to this email directly, view it on GitHub<#2329 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ADGGUWA7AF5X4SUQQL6HIS32LFTB5AVCNFSM6AAAAABVAGSQNSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKOJZGE2TENRYGQ>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
|
yeap, it still needs to be fixed. Unfortunately I have no time for that this week. Tomorrow I will make a debug build that allows to find the problem. Previously it was looking like wrong read of cpu frequency map. But it must be ok, so the problem somewhere else... |
|
@vdhamer I'm just curious right now, why you shared the log from m4 pro that was shared by another user in another issue? PS: the log you shared as your was shared here: #2285 (comment) |
|
Certainly not my intention. Don’t know what happened. Maybe compared both and sent you the wrong one. So must be an error. Want me to generate it again, and right this time?
I don’t have an M4 Mac. I only own an M1 Ultra Mac Studio (don’t have a notebook).
…Sent from my iPad
On 17 Jan 2025, at 22:14, Serhiy Mytrovtsiy ***@***.***> wrote:
@vdhamer<https://github.com/vdhamer> I'm just curious right now, why you shared the log from m4 pro that was shared by another user in another issue?
—
Reply to this email directly, view it on GitHub<#2329 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ADGGUWFUC2YAFICKQABBF3T2LFXBVAVCNFSM6AAAAABVAGSQNSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKOJZGIZDMOJUGE>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
Found the file you were expecting. I had sent you log.txt from Downloads instead of log.txt from Documents folder. If needed, I can reload the special patched version and send a fresher file. |
|
@exelban yep this is still happening with the last 2 releases |
|
@exelban still happening in 2.11.28 .... iStats Menu doesn't have this issue, it properly reports the current CPU freq. are there any logs or diagnostics I can provide? |
|
Hi. To start I need a logs from the build of this post #2285 (comment) |
|
Log for my M1 Ultra which keeps reporting 600 MHz on new versions of Stats. |
This is on an M1 Ultra during heavy compute load (export of 500 photos in Lightroom Classic).
The number of 600 MHz stays the same during idling and during heavy workload.
I don't expect it to be that low. And expect (unsure) that the frequency can auto adjust depending on work load.
Stats 2.11.23 (currently latest version).
The text was updated successfully, but these errors were encountered: