Overclock.net banner

1 - 20 of 22 Posts

·
Not a Linux Lobbyist
Joined
·
1,213 Posts
Discussion Starter #1
In the process of hoarding mouse related web files before they can't be found anymore I came across the updated version of Mouse Movement Recorder packaged in the windows 10+8.1+8 MarkC mousefix.

This version of MMR has a new functionality to influence "catchup"

From the readme:
Quote:
If your system is busy, sometimes there might be an occasional delay before Windows
updates the on-screen pointer position and if that happens the POINTER MOVEMENT will
lag behind the MOUSE MOVEMENT.
...
Sometimes there is a small delay between mouse movement and Windows moving the
on-screen pointer.
By default, if MouseMovementRecorder sees mouse movement, but the corresponding
pointer movement is not an exact 1-to-1 match, it may wait for up to 500 µs for
additional pointer movement
As soon as you open up the program you will get prompted with the command list. You can recall it whenever you want by pressing the space bar.
To visualize the discrepancy either: press "s" to show catch up delay.
or set it to "noCatchup", because it's not a real condiment.

Show catchup delay, and loops


No sauce:


Whichever view you prefer. On noCatchup the delayed movement will be shown on the next poll.

Tested on Windows 7 x64, fixed clock frequencies. Timer resolution does not affect this whether its 15.6ms or 0.5ms , nor does hyper-threading, core-parking, dwm, mouse fix (in before...) or anything else.
Windows cursor is just a mess.

How to recreate:

Put your computer under any type of load:
Open some explorer windows, notepad, steam and a web browser if you really need to.
Using two screens and/or playing videos is very good for showing this.
Move the mouse around at different speeds, sometimes it happens more at low speed sometimes at high.
Dedicate at least a couple minutes to testing as this behavior is by definition inconsistent.

This delay happens even without doing such things, but way less often.
If you're on wi-fi, you're basically doomed.
smile.gif


This doesn't not show up on old mouse movement recorder as it behaves like "autoCatchup" without shown delay.
You could observe discrepancies when dragging windows but this is by just moving the cursor.

This brings me to ask: is wm_mousemove the problem or does GetCursorPos get treated with low priority?
If it's the second, then rinput will suffer the same fate.
Could be irrelevant since probably game simulation is set up to "catchup" as well, but that's not granted.

Try it out for yourself. See how bad it is on later windows versions.
 

·
Ex-resident Kinzu shill
Joined
·
2,074 Posts
Curious as to if this is why my cursor feels like it's playing "catch up" sometimes, when I'm moving the cursor left/right, the cursor sometimes slows down and then suddenly speeds up in what feels like an attempt to catch up, of course, this could be entirely based on specific parts of my mousepad being slicker, the mouse feet, or just human error all together.
 

·
Not a Linux Lobbyist
Joined
·
1,213 Posts
Discussion Starter #4
Quote:
Originally Posted by Ino. View Post

Is this the same for windows 7 or is it just for 8,8.1,10?
Quote:
Originally Posted by the1freeMan View Post

Tested on Windows 7 x64, fixed clock frequencies. Timer resolution does not affect this whether its 15.6ms or 0.5ms , nor does hyper-threading, core-parking, dwm, mouse fix (in before...) or anything else.
....

Try it out for yourself. See how bad it is on later windows versions.
Quote:
Originally Posted by Alya View Post

Curious as to if this is why my cursor feels like it's playing "catch up" sometimes, when I'm moving the cursor left/right, the cursor sometimes slows down and then suddenly speeds up in what feels like an attempt to catch up, of course, this could be entirely based on specific parts of my mousepad being slicker, the mouse feet, or just human error all together.
Just download and play around with it. I used to use old MMR to see if I was consistent with 1 count movements. I too felt inconsistency and blamed my clumsy human hands... but actually delay is so short it's probably just that anyway.
 

·
Ex-resident Kinzu shill
Joined
·
2,074 Posts
Quote:
Originally Posted by the1freeMan View Post

Just download and play around with it. I used to use old MMR to see if I was consistent with 1 count movements. I too felt inconsistency and blamed my clumsy human hands... but actually delay is so short it's probably just that anyway.
While I don't see anything from MMR that can relate this "catch up" syndrome to it, it seems that my cursor is jumping because of some weird polling inconsistency/jitter, as every single time the cursor does a sudden "jump" while making small movements, the report rate jumps from something sub 80 to 120+ then suddenly goes back down after 1-2 counts, not sure if that can directly influence the cursor jumping suddenly like that or not. My polling jitter was absolutely horrid last time I looked though.
 

·
Registered
Joined
·
1,557 Posts
I noticed this delay immediately when I went to Windows 10. I thought it was a delay from all of there gathering of information to (sell you out) I mean make your Windows 10 experience better. After going through and turning all of there reporting system OFF. It seemed to help a tiny bit other than when I do searches it seems to always think for a second then proceed. I found this to be frustrating because I was on Windows 7 with a SATA SSD and it was fast. Moved to Windows 10 with a faster system and including a M.2 drive it felt laggy. Looking forward to the conclusion
thumb.gif
 

·
lololol
Joined
·
4,371 Posts
yea of course
it's windows
rolleyes.gif


iirc in win7 if you have compositing off and you drag a window and move the cursor to the top of the screen, the cursor only updates at like 25 fps or something
 

·
Not a Linux Lobbyist
Joined
·
1,213 Posts
Discussion Starter #8
Quote:
Originally Posted by Alya View Post

While I don't see anything from MMR that can relate this "catch up" syndrome to it, it seems that my cursor is jumping because of some weird polling inconsistency/jitter, as every single time the cursor does a sudden "jump" while making small movements, the report rate jumps from something sub 80 to 120+ then suddenly goes back down after 1-2 counts, not sure if that can directly influence the cursor jumping suddenly like that or not. My polling jitter was absolutely horrid last time I looked though.
That's weird, maybe just a kinzu thing
smile.gif
does that happen on all mice? What windows version are you on?
I tried in safe mode, where polling is the most stable but it was way worse...
I usually have a +/- 4 us variance in normal use because I'm forced on wi-fi due to logistics.
Polling is better with wi-fi adapter disabled but windows cursor still has delays.

Quote:
Originally Posted by Lu(ky View Post

I noticed this delay immediately when I went to Windows 10. I thought it was a delay from all of there gathering of information to (sell you out) I mean make your Windows 10 experience better. After going through and turning all of there reporting system OFF. It seemed to help a tiny bit other than when I do searches it seems to always think for a second then proceed. I found this to be frustrating because I was on Windows 7 with a SATA SSD and it was fast. Moved to Windows 10 with a faster system and including a M.2 drive it felt laggy. Looking forward to the conclusion
thumb.gif
Windows 10 has so much background activity even when "optimized" that if I have c-states on, cores will actually stay on full speed most of the time, occasionally down clocking.
Windows 7 (with core parking disabled) stays mostly idle with occasional jumps to full speed.

Quote:
Originally Posted by qsxcv View Post

yea of course
it's windows
rolleyes.gif


iirc in win7 if you have compositing off and you drag a window and move the cursor to the top of the screen, the cursor only updates at like 25 fps or something
Lol, that's why dragging windows shows up on old mmr too. Standard catchup delay is 500us, you can bump it up to... infinite?
 

·
Ex-resident Kinzu shill
Joined
·
2,074 Posts
Quote:
Originally Posted by the1freeMan View Post

That's weird, maybe just a kinzu thing
smile.gif
does that happen on all mice? What windows version are you on?
I tried in safe mode, where polling is the most stable but it was way worse...
I usually have a +/- 4 us variance in normal use because I'm forced on wi-fi due to logistics.
Polling is better with wi-fi adapter disabled but windows curosr still has delays.
It's not just a Kinzu thing, it happens on every mouse I own, and I use ethernet so I never have a wi-fi adapter/controller enabled, I have all USB 3.0 controllers disabled and I'm using the normal Microsoft USB 2.0 drivers, I also have all other onboard crap disabled (USB controllers, unused LAN controllers, SATA controllers, etc) so it's probably something with my OS to be honest. I'm not bothered enough by it to reinstall though.
 

·
Registered
Joined
·
1,699 Posts
Quote:
Originally Posted by qsxcv View Post

yea of course

iirc in win7 if you have compositing off and you drag a window and move the cursor to the top of the screen, the cursor only updates at like 25 fps or something
is windows classic "compositing off"

if so I cannot confirm this, full 144fps cursor movement; the window itself however updates at a slower rate
 

·
lololol
Joined
·
4,371 Posts
on win7 yea

maybe i'm remembering wrongly but there was definitely something repeatable i could do that triggers it
 

·
Registered
Joined
·
296 Posts
If the data on the MMR screenshot is true, this "delay" is menaingless, 11us is waaay beyond what any human being will ever notice
 

·
lololol
Joined
·
4,371 Posts
so mmr is comparing the timing between what and what? (getcursorpos and raw input? directinput?)
also a few us for "1 loop" seems very very slow...
 

·
Not a Linux Lobbyist
Joined
·
1,213 Posts
Discussion Starter #15
Quote:
Originally Posted by Wphantom View Post

If the data on the MMR screenshot is true, this "delay" is menaingless, 11us is waaay beyond what any human being will ever notice
Yes, but if a game does not wait for the delay while checking cursor position it will not pick up the movement.
It would theoretically get seen on the next frame, but with regular input method it just gets dropped with cursor reset.

Input can happen whenever so it's not really that important, but still other imperfections of the windows pointer method..
Quote:
Originally Posted by qsxcv View Post

so mmr is comparing the timing between what and what? (getcursorpos and raw input? directinput?)
also a few us for "1 loop" seems very very slow...
Direct input.
I admit the loop thing isn't exactly clear to me either..
 

·
Mouse lover
Joined
·
2,896 Posts
Quote:
Originally Posted by Alya View Post

Curious as to if this is why my cursor feels like it's playing "catch up" sometimes, when I'm moving the cursor left/right, the cursor sometimes slows down and then suddenly speeds up in what feels like an attempt to catch up, of course, this could be entirely based on specific parts of my mousepad being slicker, the mouse feet, or just human error all together.
Only mouse I ever had do that was the g303

But I run windows 7 with classic theme (aka desktop composition/aero off). Haven't really had any other problems besides that
 

·
Not a Linux Lobbyist
Joined
·
1,213 Posts
Discussion Starter #18
I use a G303, it has no problems of that sort.
Alya's problem is not mouse specific or related to windows pointer micro delays.
He seems to have some issues with polling for some reason.
 

·
Mouse lover
Joined
·
2,896 Posts
Quote:
Originally Posted by Lolcarrots View Post

I want to know more as I'm a current g303 user
thumb.gif
The mouse is great and it never caused me any problems in game, just that on the desktop itself it felt like the cursor was following my movement in a way - in game it didn't appear that way but idk
 

·
Registered
Joined
·
65 Posts
After reading the source code , I think this soft is useless for gamer, it just tells if the mosue motion has problem or not on desktop . Becasue most serious FPS or MOBA games does not use the getcurPOS, instead they use raw_input or Direct_input, so whats the point uisng the direct_input to measure the latency to the getCurPos API ?
 
1 - 20 of 22 Posts
Top