Task scheduler

@tonyw

Same one as mine.

Best regards,

Niels

It is sad that you beta testers are so far ahead of officially released updates, not blaming you, blaming Hyperion.

I've spent 20 years and money on the OS I love ,μA1 ,Pegasos2 ,A1XE, X5000.
X5000 been my main computer since 2016.

I know it's against NDA but if some of you want to share some new piece of OS4 P.M. me!
I've got my 3 X5000 firmware updates this way because A-EON fail to support their customers. Why? 2000€ fantastic motherboard, abbandoned.. Why?

Put the firmware updates on a-eon.com like every other hardware manufacturer!

Send me what you want, my mouth is shut!

Edit: Sorry Raziel for Hijacking your thread.

@Capehill
Quote:I don't remember how this version of my top works, I implemented different ones.
If it's using the times from the ExecNG task scheduler in struct ETask (which at least in 2006 didn't work for example with 603e CPUs, i.e. BlizzardPPC, yet but only some other CPUs like 750 and 74xy) it needs the old task lists, but if it's using the other, interrupt based method calculating the CPU usage itself instead it doesn't use task lists but only IExec->FindTask(NULL) to create local copies of the Task or Process/CLI names.

@joerg

Kernel 54.20 didn't show task names (top, tequila). Kernel 54.46 seems to display those on both program. Nice.

@Capehill

What about cpu usage, is there a way to poll that too?

Edit:
@khayoz

That's quite alright

@Raziel
Quote:Yes, but apparently only for OS4 developers with access to the internal exec includes.

At least in SDK 53.30 struct ETask isn't in the exec includes.
I can't check newer SDK versions, probably using some AmigaOS-only LhA extension, for example 7-Zip can't extract it (base.lha/Include/include_h/exec isn't recognised as a directory but as an empty file instead).

Getting the CPU usage of a task using PerformanceMonitor.resource should be possible as well.

@Capehill
I checked your tequila sources, seems to be very similar to my 2006 version of top.
There is only one possible problem in it I found: It's using the idle.task usage for "unused" CPU time.
While that works with the original ilde.task, at least 15 years old versions of it, it may not for example if you run my CPUTemp.docky on some of the CPUs it did support (603/604, 750, 74xy, maybe 440) because CPUTemp.docky replaces the ilde.task busy loop by code suspending the CPU (wait state/standby mode).

@joerg

Aww, too bad

@joerg

I wish Exec would provide proper APIs to query CPU/stack/memory usage per task. I suppose most CPU meters are somewhat hacky. My second wish is that OS itself would provide some basic monitoring tools :)

Thanks for checking the sources. If somebody finds your top clone source code, it will interesting to read.

I took a screenshot of both top and tequila running at the same time: http://capehill.kapsi.fi/top/top_tequila.png

It seems that they mostly agree about performance, but top is more sophisticated, it is able to detect more tasks in general use.

@Raziel

Memory usage of each process requires patching lots of system functions.

Memory usage of a specific program is trivial if you can modify the source code. For example in C++ you can overload new/delete operators and you can redefine malloc/free and wrap everything else.

@Joerg: i can confirm the sources i have available make use of idle.task and forbid.

Let me know if it's ok to share it here.

@Ami603
Quote:Please sent the sources to Capehill, maybe it includes something he can use to improve his tequila tool, which does the same as my top.
Having two independent tools which basically do the same thing doesn't make much sense.

@joerg:

Ok, done as told.

@Capehill:

Check your PM.

@joerg

A great little utility- thanks!

@all

It's obvious what most of the displayed numbers represent, but there are some whose meaning is not apparent. Does anyone know what the F:xxxxx, D:xxxxx, and S:xxxxx figures at the top of the display mean? (See Capehill's screenshot for reference.)

@msteed
IIRC it's the number of IExec->Forbid(), Disable() and SuperState() calls per second.
Just like the way too high number of task switches per second on a nearly idle system (usually caused by lots of 0.0x second timers, you can use for example Scout to check the timer.device list) in the screenshot those are calls with some, or even a lot of overhead which shouldn't be used that often.
It's no big problem on NG systems, but one of the reasons for implementing the "top" program was to find out what made AmigaOS 4.x extremely slow on A1200/BlizzardPPC. For example AmiDock caused hundreds of task switches/second even if no Docky was updating anything, resulting in about 20% CPU usage on an "idle" BlizzardPPC system.

Tequila 1.0 is available. It has now a GUI mode too.

https://github.com/capehill/tequila/releases/tag/v1.0 or
http://capehill.kapsi.fi/top/Tequila-1.0.lha

@Ami603, please check your PM.

@Capehill

Ok, sent you an email instead.

@Capehill

Thanks. Looks nice and useful. Sometimes my CPU goes up just after booting.

What I found is, when my CPU clock suddenly listed high CPU after booting Workbench, is that CPUCLock appears to be the culprit. It listed about 97% usage but the priority is -128 or close to. It shouldn't be that high and shouldn't get so much CPU time to register as such. But maybe something is putting it off.

@Hypex

Thanks for testing.

If CPUClock busy loops, then it is probably ready all the time.

I seem to have added CPUClock task to "idlers" list so system is considered to be idling even though CPUClock is spending free CPU. https://github.com/capehill/tequila/blob/master/profiler.c#L276

@Capehill

Okay, yes, can see that. Thanks for explanation.