Israeli-Eagle
Active Member
Yep, but that issue started already of W10 2015 and never became better, parts even worse. Because of that most professionals & semis use alternative tools instead.
I think cpu usage is broken again
- Thread starter Pinkie Pie
- Start date
I mostly used the classic time-based CPU usage metric in the past to understand how my resources were being utilized during a game session. I would leave Hwinfo or Hwmonitor running in the background while gaming and then I would come back at the end to check the maximum values for CPU usage. With this kind of information, I could determine if I was hitting any sort of bottleneck or comprehend better the reasons behind my CPU temperatures, voltage and workload in a given scenario.
With this new Task Manager metric standard, how can I read it in order to obtain the same information I always looked for? How do I know if my CPU is being used to its full potential in a certain game (acting as a bottleneck, for example) or if it still has some performance headroom to spare? Sorry if it is a stupid question, I genuinely feel lost most of the time now.
With this new Task Manager metric standard, how can I read it in order to obtain the same information I always looked for? How do I know if my CPU is being used to its full potential in a certain game (acting as a bottleneck, for example) or if it still has some performance headroom to spare? Sorry if it is a stupid question, I genuinely feel lost most of the time now.
Israeli-Eagle
Active Member
Exactly, I felt the same way. But there is simply no other way imho as the others better ones do.
And no reason ever to disable HWiNFO or some useful tools, they actually 'eat' almost nothing on strong hardware.
And no reason ever to disable HWiNFO or some useful tools, they actually 'eat' almost nothing on strong hardware.
That's not a stupid question at all, it would be an interesting indicator. But the CPU Utility won't tell you that. The maximum value (100%) would have to be normalized to the highest performance limit for a given CPU. But since this limit depends on many factors (e.g. Multi-Core Turbo Limits, various Power/TDP limits, power delivery/current limits, temperatures, etc.). Such performance-wise limit would be very hard to determine, if not impossible. That's why the CPU Utility in fact has no upper limit, it can go over 1000% when Turbo clocks are efficiently engaged.. 100% in this case means the CPU is utilized at the base frequency.
Vimes
Active Member
I noted the problem after I updated from 7.42 to this latest 7.44 build.
Not noting this thread I reported it here....
Problem with 7.44 build
Now I have changed to this......
it seems to work as it did previously.......
Hopefully I used the right values to change, and show OSD.
Combining the core clocks with the core utility values.
Thanks
Not noting this thread I reported it here....
Problem with 7.44 build
Now I have changed to this......
it seems to work as it did previously.......
Hopefully I used the right values to change, and show OSD.
Combining the core clocks with the core utility values.
Thanks
Israeli-Eagle
Active Member
Hmmm... Would you see different clocks for example in CoreTemp?
Vimes
Active Member
When I used 7.44 and noted the almost zero CPU usage, linked to the thread where I posted a screenshot, I had the resource monitor open in Windows 11 at the same time. It reported what I would have assumed to be more typical CPU usage, somewhat confirmed by vcore, temp and fan usage.
When I made the noted changes to 7.44 they seem to be again more aligned with the noted resource monitor and what I would have seen previously.
Whether there is doubt or uncertainty on the validity of that, both that type of monitoring and that of resource monitor, I do not know. What I do know is that I seem to be back where I was.
This was when I used 7.44 initially without changing those OSD reading sources, as such.............
more by chance at the time of taking that screeshot as it showed some CPU usage. Otherwise the game in question was running at an indicated 0.0% usage.
Took me some time to figure out what to change, hopefully correct for what I wanted. But it is done now.
When I made the noted changes to 7.44 they seem to be again more aligned with the noted resource monitor and what I would have seen previously.
Whether there is doubt or uncertainty on the validity of that, both that type of monitoring and that of resource monitor, I do not know. What I do know is that I seem to be back where I was.
This was when I used 7.44 initially without changing those OSD reading sources, as such.............
more by chance at the time of taking that screeshot as it showed some CPU usage. Otherwise the game in question was running at an indicated 0.0% usage.
Took me some time to figure out what to change, hopefully correct for what I wanted. But it is done now.
Israeli-Eagle
Active Member
Simply first compare with real good other tools, but not trust MS Task Manager!!
And the previous/first pic of Cities Skylines was for sure wrong, it cannot have CPU usage like Cyberpunk 2077. The second one looks very correct, because it's known that it uses mostly only one core and C# based. And generally games usually need much more GPU, but much less CPU.
Would any other tools (for example Process Explorer, CoreTemp or System Informer) display something very different?
And the previous/first pic of Cities Skylines was for sure wrong, it cannot have CPU usage like Cyberpunk 2077. The second one looks very correct, because it's known that it uses mostly only one core and C# based. And generally games usually need much more GPU, but much less CPU.
Would any other tools (for example Process Explorer, CoreTemp or System Informer) display something very different?
Vimes
Active Member
Ok I'll need to look at some of those other tools, thanks for the idea.
Going back to the "native" version of 7.44...................
there can be some single core usage of the CPU. It just seems ver light usage of the CPU, by the way the game is Transport Fever 2, not that it matters.
Do you know if those other utilities you mentioned are more accurate in their displayed output than what I have been using, in other words what HWiNFO provided before this change and then combined with resource manager through Windows..?
BTW Both TF2 and CS can be pretty intensive in terms of their CPU loading. Especially as the map fills and the total number of agents tracked increases.
Going back to the "native" version of 7.44...................
there can be some single core usage of the CPU. It just seems ver light usage of the CPU, by the way the game is Transport Fever 2, not that it matters.
Do you know if those other utilities you mentioned are more accurate in their displayed output than what I have been using, in other words what HWiNFO provided before this change and then combined with resource manager through Windows..?
BTW Both TF2 and CS can be pretty intensive in terms of their CPU loading. Especially as the map fills and the total number of agents tracked increases.
Last edited:
Israeli-Eagle
Active Member
Ups, sorry... Transport Fever 2 is good stuff and I think real compiled code. But when really zero (last pic) then there is a problem for sure! Hmmm.....
And all well known tools are based like HWiNFO (again since 7.44) of legacy time-based, even the Process Explorer from MS SysInternals. But Windows itself with Task Manager since W10 2015 is known as very buggy and often totally wrong. For example on my own home system always only displayed 100% and never anything else, happens often already when any VM installed with Hyper-V or sometimes already when different Power Plans (come per default already with AMD chipset drivers). While the bad MS Task Manager is already in a shell (.NET) and cannot always get out for correct data.
And all well known tools are based like HWiNFO (again since 7.44) of legacy time-based, even the Process Explorer from MS SysInternals. But Windows itself with Task Manager since W10 2015 is known as very buggy and often totally wrong. For example on my own home system always only displayed 100% and never anything else, happens often already when any VM installed with Hyper-V or sometimes already when different Power Plans (come per default already with AMD chipset drivers). While the bad MS Task Manager is already in a shell (.NET) and cannot always get out for correct data.
Vimes
Active Member
This could get a little confusing to follow. This is a screenshot of 7.44 BUT with modified settings, as I noted previously, and the start of a NEW map in TF2.......................
.............as expected the P and E cores are not clocked up to their max and the load per core and overall load is low. I would just not expect the game to show zero loading of all of the cores during gameplay. Whilst it is not Cyberpunk, it does utilise the CPU. Another game I tried showed similar readings.
Anyways.........Back to enjoying playing the games now.
I could be testing rather than playing all day
Thanks for your thoughts and help
Point taken on what you mean. I'm just not sure I had an issue on what I was previously seeing nor a solution with what I have now got.
Maybe some others could test who are far more experienced.
Oh and now, as indicated already, if I use the modified way of trying to represent what I had in 7.42 I get % readings for each core than can go beyond 100%. I did not have that before.
.............as expected the P and E cores are not clocked up to their max and the load per core and overall load is low. I would just not expect the game to show zero loading of all of the cores during gameplay. Whilst it is not Cyberpunk, it does utilise the CPU. Another game I tried showed similar readings.
Anyways.........Back to enjoying playing the games now.
I could be testing rather than playing all day
Thanks for your thoughts and help
Point taken on what you mean. I'm just not sure I had an issue on what I was previously seeing nor a solution with what I have now got.
Maybe some others could test who are far more experienced.
Oh and now, as indicated already, if I use the modified way of trying to represent what I had in 7.42 I get % readings for each core than can go beyond 100%. I did not have that before.
Last edited:
Israeli-Eagle
Active Member
Vimes
Active Member
You do make a very good point, but also I think that you can appreciate that what I'm noting, albeit limited testing, does not seem right as much as what I had previously was absolutely wrong.
If task manager capped the value at 100% for core usage I would have expected to see that as a maximum when using the previously OSD display.
If task manager capped the value at 100% for core usage I would have expected to see that as a maximum when using the previously OSD display.
Israeli-Eagle
Active Member
I just tested Transport Fever 2 and looks very normal for me (tho not used with OSD):
(The 100% max because I used hard benchmarks an hour before and not reset)
(The 100% max because I used hard benchmarks an hour before and not reset)
Pinkie Pie
Well-Known Member
Not first time it broke, honestly its broken within task manager it self as well cos games like world of warcraft are showing 0% or 1% cpu usage in details tab in new task manager while its actually using least 20% sometimes 25% depending if setup game physx via cpu.
Even performance tab more accurate, so its not just hwinfo sometimes miss reporting atleast how i feel about total cpu usage, but also task manager it self, altho performance tab does say Utilization, i suspect everything is utilization in task manager but details tab = cpu usage
I think it may be worth it to bug report it to feedback hub as possibility that its probably Microsoft miss reporting it, or i do not understand how it works, and being to lazy to read thru everything right now as i have to do real life stuff right now.
Even performance tab more accurate, so its not just hwinfo sometimes miss reporting atleast how i feel about total cpu usage, but also task manager it self, altho performance tab does say Utilization, i suspect everything is utilization in task manager but details tab = cpu usage
I think it may be worth it to bug report it to feedback hub as possibility that its probably Microsoft miss reporting it, or i do not understand how it works, and being to lazy to read thru everything right now as i have to do real life stuff right now.
Last edited:
Israeli-Eagle
Active Member
Already your clocks are very fishy and not normal!!
Did you ever install the original AMD Chipset drivers and their Power Plans?
Because a normal desktop CPU never ever has to go down to such extreme low clocks.
Pinkie Pie
Well-Known Member
5950x does not have power plans, powerplans only needed for 3000 series and older.
Last edited:
Israeli-Eagle
Active Member
Yeah, I also just saw that series 5000 not have anymore power plans. Tho for me that is weird and then somehow works like a laptop CPU (low clocks). Hmmm... Well, we cannot understand everything.
Pinkie Pie
Well-Known Member
Obviously want it to have low clocks at idle or it will just idle at really high temps.
Israeli-Eagle
Active Member
Or already MS forced it to AMD and banned power plans, coz otherwise one (of many) reason that the bad Task Manager stops working correctly. So maybe because of that Task Manager for me only displays everything 100% on 3700X, tho all others tools display perfectly. LOL
Anyway... Low clocks means everything based of that core runs dead-slow, which a good desktop CPU would never do. There is also no real reason to clock down, because high temps come only from high loads. Only laptops need it because of lacking enough cooling. And also because of that many people disabled long ago already the bad Cool'n'Quiet...
Last edited: