Opened 15 years ago
Closed 7 years ago
#4538 closed bug (invalid)
Moving mouse causes CPU utilization to jump to ~50%
Reported by: | ormandj | Owned by: | mmlr |
---|---|---|---|
Priority: | normal | Milestone: | R1 |
Component: | Drivers/USB | Version: | R1/alpha1 |
Keywords: | Cc: | ||
Blocked By: | Blocking: | ||
Platform: | x86 |
Description
Hi,
On my system, moving my mouse causes CPU utilization to jump to roughly 50%. The screenshot demonstrating this with top was taken while moving the mouse as rapidly as possible in a 2" circle over the desktop background (not a window). This is a fresh R1 Alpha install.
I was also asked take a picture (sorry for the blurriness) in the kernel debugger, this is also attached.
The system is an Athlon XP 1800, with 1.5 gigs of ram, a GeForce4 MX440 using AGP, a small IDE hard drive, and three network cards (the syskonnect driver utilizing NIC is the one connected to a network at the time I was testing this). 'listdev' output is also attached.
Thanks, David
Attachments (6)
Change History (12)
by , 15 years ago
Attachment: | withoutmousemovement.png added |
---|
comment:1 by , 15 years ago
Note: screenshots were taken with an app_server built with acceleration enabled in an attempt to isolate where the problem lies. This did improve performance for things like window dragging (less tearing than with the normal app_server), but made negligible difference to mouse movement.
by , 15 years ago
Attachment: | kdebug2.jpg added |
---|
by , 15 years ago
Attachment: | kdebug3.jpg added |
---|
ints in kdebug with parallel port, game port, serial ports disabled, and ipro100 driver moved out of the way.
comment:2 by , 15 years ago
Component: | System → Drivers/USB |
---|---|
Owner: | changed from | to
Status: | new → assigned |
That's fine basically :-). It likely comes from the fact that UHCI does not actually care about the mouse interrupt interval. It always polls at a 1ms interval which, depending on the mouse still sending them at a higher rate, results in more mouse events triggered than usual. I thought that stippi implemented a certain throttling on the input device side, but it might not be enough. Can you please also attach the output of "listusb -v"?
by , 15 years ago
Attachment: | listusb-v.txt added |
---|
Without mouse movement