Tuning mouse report rate for optimal VSync frame pacing

[fuweichin]

To avoid pointer movement stuttering when coalescing HID input messages as VSync-aligned pointer events, it is necessary to design a mouse report rate (or say sampling rate sometimes) which is an integral multiple of display refresh rate.

So, for common refresh rates 60Hz/120Hz, it's neccesary to use a report rate like 120Hz/240Hz to keep VSync frames even in terms of pointer movement.

Yet I cannot find a mouse which is optimized for VSync frame pacing, that's why I'm interested in DIY/open-source mouse.

Why integral multiple? Here are some testing results with Pointer Movement Timing Analyzer:

Figure1: using 119Hz report rate with 60Hz refresh rate MAY cause 1 jumpy frame per second

Figure2: using 125Hz report rate with 60Hz refresh rate MAY cause 5 jumpy frames per second

Apple continuously optimizes report rates for VSync frame pacing, and now only leaves its Magic Mouse and Magic Trackpad behind.

Table1: Apple Products HID Sampling Rate

Device	Pointer Type	Sampling Rate	Refresh Rate
iPhone 13 Pro	touchscreen	240Hz	120Hz
iPhone 12	touchscreen	120Hz	60Hz
iPad Pro 12.9" 2nd gen	touchscreen	120Hz	120Hz
Apple Pencil	pen	240Hz
MacBook Pro M1	mouse/touchpad	120Hz	120Hz
Magic Mouse	mouse	90Hz
Magic Trackpad	mouse/touchpad	90Hz

[wareya]

Normal speed USB HID report rates unfortunately need to be millisecond-aligned, which means that the only options between 125hz and 1000hz are as follows:

125hz (8ms)
~142.8hz (7ms)
~166.6hz (6ms)
200hz (5ms)
250hz (4ms)
~333.3hz (3ms)
500hz (2ms)
1000hz (1ms)

Any report rates for a USB HID device between these is, in reality, switching back and forth between similar rates to give an average of the measured rate.

The easiest way to avoid input device movement microstutter is to use as high of a report rate as possible. This is one of the reasons that gaming mouse firmware (including this repo) has decided on 1000hz (sometimes 500hz) as the standard. Doing this also minimizes input latency. Modern office mice still use 125hz, but they're just being weird/bad.

Some modern mice are capable of using even higher sampling rates, like 8000hz, over higher-speed versions of USB. For these mice, more sample rates are possible, but for input latency reasons, it's still best to use the highest sample rate that can be stably maintained. The rp2040 isn't capable of these even higher-speed versions of USB.

Also, the 3060 sensor in this mouse has a sampling rate of its own (something in the dozens of khz), so at some point down the signal chain, there's always going to be a small amount of microstutter.

If you're on linux, you can sometimes override USB device polling rate: https://wiki.archlinux.org/title/Mouse_polling_rate#Set_polling_interval

[fuweichin]

Mouse DIY players can take the lead in improving mouse VSync frame pacing even via some just-works solution, as a proof of concept to deliver unique value to the industry, e.g. use 8000Hz (0.125ms interval) sensor then downsample to about 120Hz, e.g. 119.4Hz (8.375ms interval), report rate, which is better than 125Hz in terms of VSync frame pacing, better than 1000Hz in terms of energy efficiency.

That shouldn’t be something only-Apple-can-do. That's why I tend to DIY a mouse.

Indeed, 1000Hz report rate can reduce VSync pointer movement microstutter,

but is at the cost of higher CPU usage, e.g. moving 1000Hz mouse over empty area causes 10% CPU usage(i7-7920HQ, Windows 10), on Windows 10 desktop that causes extra 10% CPU usage if “Show desktop icons” is checked, which means more heat and fan noise.