Does multiple USB devices with 1ms polling slow down the MiSTer?
Does multiple USB devices with 1ms polling slow down the MiSTer?
Is there a limit to how many devices the MiSTer can support even if they are all using USB at 1ms polling?
For example, if you have four players each using a DaemonBite USB encoder, does that cause any issues?
I'm not quite sure how of the MiSTer architecture, but I think the Linux part of the device handles the USB so has anybody tested this?
For example, if you have four players each using a DaemonBite USB encoder, does that cause any issues?
I'm not quite sure how of the MiSTer architecture, but I think the Linux part of the device handles the USB so has anybody tested this?
-
- Top Contributor
- Posts: 381
- Joined: Tue Jun 16, 2020 12:22 pm
- Has thanked: 58 times
- Been thanked: 77 times
Re: Does multiple USB devices with 1ms polling slow down the MiSTer?
I'm not sure but my guess is no.
Not anymore then a slower polling rate or one single usb device at 1ms will slow down the mister.
from what i understand the usb will constantly ask/check the device is there anything new happening and if nothing new nothing will be sent/happen.
Lowering the polling rate to 1ms will tax the mister (probably the ARM cpu) more since it will have to shout anything new more often, but if that is noticeable at all on what the mister is used for is a completely different story.
If more usb devices would mean more lag/tax/etc is probably only if more then one controller presses one or more buttons at the same millisecond meaning maximum 3ms lag added=4ms total for 4 vs 1 controller, and it will be well under the 16ms one screen is shown.
My guess is if so many devices is using the usb bus at once it is the usb bus that will slow down not the rest of the system (as long as nothing in the rest of the system is not wating for any usb device to finish anyway).
The more problematic is probably what chip is in the usb hubs/usb controllers and so on THAT can add much lag.
More info in this tread.
viewtopic.php?f=32&t=915
Not anymore then a slower polling rate or one single usb device at 1ms will slow down the mister.
from what i understand the usb will constantly ask/check the device is there anything new happening and if nothing new nothing will be sent/happen.
Lowering the polling rate to 1ms will tax the mister (probably the ARM cpu) more since it will have to shout anything new more often, but if that is noticeable at all on what the mister is used for is a completely different story.
If more usb devices would mean more lag/tax/etc is probably only if more then one controller presses one or more buttons at the same millisecond meaning maximum 3ms lag added=4ms total for 4 vs 1 controller, and it will be well under the 16ms one screen is shown.
My guess is if so many devices is using the usb bus at once it is the usb bus that will slow down not the rest of the system (as long as nothing in the rest of the system is not wating for any usb device to finish anyway).
The more problematic is probably what chip is in the usb hubs/usb controllers and so on THAT can add much lag.
More info in this tread.
viewtopic.php?f=32&t=915
- Newsdee
- Top Contributor
- Posts: 873
- Joined: Mon May 25, 2020 1:07 am
- Has thanked: 104 times
- Been thanked: 239 times
Re: Does multiple USB devices with 1ms polling slow down the MiSTer?
1ms is basically 1Mhz, and the ARM core runs at 800 Mhz. Even if it used the four Deamonbyte adapters sequentially, it has plenty of capacity to spare.
As grizzly points out, if you experience lag with a controller it's more likely to be a slower chip inside it (i.e. it is not physically capable of running at 1ms / 1Mhz). You could also have odd issues (e.g. button not debouncing) if you have insufficient power to your devices (it's recommended to use a powered USB hub).
As grizzly points out, if you experience lag with a controller it's more likely to be a slower chip inside it (i.e. it is not physically capable of running at 1ms / 1Mhz). You could also have odd issues (e.g. button not debouncing) if you have insufficient power to your devices (it's recommended to use a powered USB hub).
Re: Does multiple USB devices with 1ms polling slow down the MiSTer?
Almost correct, but a period of 1 ms corresponds to a frequency of only 1 kHz, so it is even slower than you say.
Github: github.com/rsn8887
- Newsdee
- Top Contributor
- Posts: 873
- Joined: Mon May 25, 2020 1:07 am
- Has thanked: 104 times
- Been thanked: 239 times
Re: Does multiple USB devices with 1ms polling slow down the MiSTer?
Doh, you're right, got mixed up by an order of magnitude
Re: Does multiple USB devices with 1ms polling slow down the MiSTer?
Three orders of magnitude.
Github: github.com/rsn8887