Project Mercury - An automatick process priority manager / booster

[nikkho]

It seems we have a new 03/20/2016 version (1.2.3.1) which fixed the issue of saving/restoring settings.

[SvenBent]

I updated the donators version previously with an interim fixed version. This weekend i made the normal and donator build the right way. and pushed it to softpedia as well.

Techcenter.dk will be down for approximately a week starting the next couple of days. as i need to diagnose and document the connection so my ISP will do something about my 5-6 disconnect a day.

Also the 0% CPU usage version of the "Don't standby on CPU load" features is scratched. Once windows sends the WM_Powerbroadcast messag,e I can no longer disable suspend, or interfere with the suspend process .
So for the time being that feature will increase CPU load slightly, but still less than alternative software I found. It uses around 0.015% on a Core 2 Quad Q6600

[SvenBent]

I was going through some of my IF settings to see if i could optimize a bit of code here and discovered a typo in a critical place.

The priority adjustment might not have been working on certain (most) system. It is fixed now as it was a simple typo. I'm not pushing the fix right now as next weekend i will hopefully finish next version.

[nikkho]

I was going through some of my IF settings to see if i could optimize a bit of code here and discovered a typo in a critical place.
The priority adjustment might not have been working on certain (most) system. It is fixed now as it was a simple typo. I'm not pushing the fix right now as next weekend i will hopefully finish next version.

This is one of the reasons which make sense on turning it open source. Silly typos, are easy discovered by other users.
Anyway, looking forward next weekend release.

[necros]

+1

[SvenBent]

Ive uploaded my interim version for the donators and fixed my servers virtual filesystem.
You should be able to login again.

The typo is fixed so it should be running perfect again.
The are some small evidence in this version of the changes to come in the next version.

[nikkho]

Ive uploaded my interim version for the donators and fixed my servers virtual filesystem.
You should be able to login again.

The typo is fixed so it should be running perfect again.
The are some small evidence in this version of the changes to come in the next version.

Thanks. I have been able to download it!

[Jaff]

There's any chance for getting working winXP 32bit compiled version?

[necros]

+1 for 32 bit ver

[SvenBent]

Oh go update that system

I will make 32bit versions this Weekend and push thet bug fix to the normal version as well + new features/control UI
32bit versions will be a regular thing now. Seems more and more ppl are asking for it.

[necros]

donator`s ver doesn`t seem to change priorities for ex. to realtime

[SvenBent]

donator`s ver doesn`t seem to change priorities for ex. to realtime

Read you skype
Its an interim version to push the bug fix. so the new UI is wuarterly done at best and not working. It should be working on this weekends real release
The new UI is pretty much just an empty shell and is not connected to anything code wise

[SvenBent]

I haven't had time to sit down and work with PM yet due to mothers day but should be able to tommorow.

I just updated the donators version with another interim/ "beta" where i connected the priority selection from the new GUI to the actual code. so now its possible to select which priority it increases the active process too.

[SvenBent]

V1.2.4.0 is up on www.TechCenter.dk both normal and donators version
Both 64 bit and 32bit versions (Haven't tested it on XP though)

- New Settings GUI
- Increased CPU priority level can now be adjusted and saved
- Fixed typo Ballon to Balloon


Softpedia should be updated within 24 hours

[nikkho]

V1.2.4.0 is up on www.TechCenter.dk both normal and donators version
Both 64 bit and 32bit versions (Haven't tested it on XP though)

- New Settings GUI
- Increased CPU priority level can now be adjusted and saved
- Fixed typo Ballon to Balloon


Softpedia should be updated within 24 hours

Thanks.

[SvenBent]

Plans (In a few weeks i need to finisnh up a article on somethind else first)

- Move CPU slice settings into the new settings GUI
- Clear idle task function. Mostly for preparing machines for benchmarks. May mistakes this and an optimization. it is not
- Re add automatic CPU slice mode. I'm unsure on this is the ressources gained from this might be wasted in the overhead of handling it . bassicalyl it will give you desktop mode ( improvede systme response) when you are kb/mouse active and change to Server ( Improved CPU ressources) when you are not
- Paypal donation link.

[SvenBent]

I received a request that u want to discuss with you guys. Personally i dont like it that much mostly because its a Memory cleaner functions which i consider utter snakeoil unless for very specific reasons.

This would simply "clear memory" on a process when you minimize its window.
Bad idea/good idea ?

[nikkho]

I received a request that u want to discuss with you guys. Personally i dont like it that much mostly because its a Memory cleaner functions which i consider utter snakeoil unless for very specific reasons.

This would simply "clear memory" on a process when you minimize its window.
Bad idea/good idea ?

Sounds good, but only if it can be enabled/disabled.

[necros]

IMHO I think it`s useless as all memory "optimizers"

[SvenBent]

IMHO I think it`s useless as all memory "optimizers"

I agree lol but it goings to take maybe 2-3 lines of code to implant it and another 2-3 to make it selectable on the settings GUI.

ION: I tweaked the Clear memory function for a 5% speed up now. ( tested on a C2Q q6600 system)

[SvenBent]

I updated the Donators version with a new interim version

some changes :
- Moved CPU slice settings into the new settings GUI.
- Improved "Clear Memory" speed by 5%.
- Optimized some IF settings for minuscule reduced CPU usage.
- Implanted "Reduce memory usage on minimize" feature
- Minor change to the Settings GUI layout

[OneDeltaTenTango]

Other than memory footprint and CPU usage does this have any performance advantages over Process Lasso?

[SvenBent]

Other than memory footprint and CPU usage does this have any performance advantages over Process Lasso?

Well less CPU usage + more CPU usage to whatever you do

it depends on how process lasso is set up. if you are runnig process lasso in its standard configuration. that process lasso does not help you in anyway during heavy multitasking.
When you have More processes than cores that uses CPU time. Process lasso cant find a process that uses the entire core and therefore does not adjust anything.
Process lassos philosophy is to find a bad program/process and adjust it down. and that means a SINGLE program/process.

if you start up a bunch of command lines programs ( like brute forcing pngout trials). lets say 8 on a 4core CPU each process takes up around 50% CPU time. which is below Process lasso threshold of 80% (IIRC) and Proceslasso thinks everything is fine even though its all lagging and slowing down.

or if you do start up 4 process and then start a game. you will have 5 CPU heavy processes and linger around 80% each, which will toss process lasso into a epileptic seizure of adjusting some process down due to hitting the 80% mark and leaving some unchanged. But due to this change the remaining process peak ups and now take 80% of CPU time and Process lasso lowers the new offends and release the old once since they are below 80%. but now the newly released ones peaks to 80%and the new restrained one is going below 80% rinse and repeat..
All this constant monitoring and adjusting priority eat up even more CPU time and you are still having heacu cpu process beeing tossed in and out of the same priority as your game (foreground program)so you still experience the lag and slowdown.

The underlined philosophy of how Process lass works in its default configuration is just flawed.


However you do have the opportunity to mimic's Project Mercury's Philolosphy of increasing priority based on the user.
but Process lasso can only increase to "high" which is a rather unfortunate Priory level.
Boosting you foreground program to high means it in the same priority as keyboard/mouse input as well a network drives. non of these really take alot of CPU time but now the need to fight with you foreground application. if you foreground application hit a high CPU usage ( 95+) you wil lexpericne mouse lag and network lag.
Project mercury in it default configuration only increase priorty to AboveNormal which make sure kayboard/mouse nad network still gets it neede CPU usage. but all other programs you are running are still having a lower priorty than you foreground application so you are still getting the benefits.
Since windows ( according to Cbloom) works with a brick wall prioriy (aka it doesn matter how much/lower you priroty s. as logn as its lower you are not gettng cpu time if a higer process needs it) you are still getting the same boosted CPU time from being in above normal as in High as long as there is no other Above normal Priotity Process that takes away CPU time.

if you for some reason wants HIGH Project mercury now has the option to increase to both High and Realtime as well. but i don't recommend it for general use.



if i have to sum it up:
In most situations there will not be a big difference. But in some situations Project Mercury will provide a way more lag free experience.

[OneDeltaTenTango]

Just tested, foreground boost in Process Lasso defaults to above normal.
It also has 2 options for foreground boost.

[SvenBent]

#84

They might have changed it since last time i tested it. or i might confuse it with process tamer. it might be tamer that only allows only High as foreground boosting.
I know process Tamer works to find a bad program on the total CPU usage aka its pretty bad on multicore CPU's . Process Lasso does it per Core usage

I'm not sure Process lasso differ on class and thread boosting. i didn't play around a lot with them. Just mostly interested in making some benchmarks and compared hem against Project Mercury

[nikkho]

Any updates planned?

[necros]

Hi there! How about an option to include https://github.com/tebjan/TimerTool functions in Mercury so not to have 2 utilities instead of 1? Is it makes sense as a feature request?

[nikkho]

Hi there! How about an option to include https://github.com/tebjan/TimerTool functions in Mercury so not to have 2 utilities instead of 1? Is it makes sense as a feature request?

Sounds great!
Thanks.

[SvenBent]

Hi there! How about an option to include https://github.com/tebjan/TimerTool functions in Mercury so not to have 2 utilities instead of 1? Is it makes sense as a feature request?

Sounds like a good idea. I'm very cautious about bloat but this seems to fit the tools purpose nicely.
Some question though:
- Do you just want it as a simpel setting in the settings GUI ?
- Do you need any way of it to change automaticall depending on uer activity / idle ( like increase to 15.6ms when PC is idling )
- both ?
I'm trying to make some benchmarks figure out out the different but cant really seem ti get the same resulst as teha author but it might just be me feeble attemps beeing done wrong


Also in the works:
- movde cpu slice into Settings GUI ( finished)
- added some boxes to groups settings ( finished)
- Currently workin on an AMD boost mode that will intercept .exe files compiled with ICC and fix the bad code dispatch so it should in theory run more optimized code under AMD CPU's

What holding me back:
Due to lack of AMD CPU's i have not been able to verify results yet
also intercepting when windows is about to execute a file. Change it and then feed the changed file to windows for execution seems to be a challenge to me.

[SvenBent]

The timer resolution seem to be easy so far. I've already made a seperate tool for it and just need to import it into project mercury.
however even the timertool and my solutions im only able to hit
15.6ms
10ms
5ms
2.5ms
1.25ms
1ms
0.5ms

Is anybody able to hit other time resolutions ?