Some Question about USB and TIMERS
Posted: 06 Apr 2021, 12:06
Hello Forum, hello Chief.
I have questions but no answers.
1) About USB.
1.Is usb 2.0 better for mice than usb 3.0?
2. xhci hand off disabled make usb 3.0 = usb 2.0?
3. If I buy a pci-e usb 2.0 extension will I get more latency than I would use the ports on the motherboard?
2) CSM better in Mousetester, but uefi feels better. placebo or is there an explanation? MBR or GPT?
3) I have experience with hpet disabled in bios. I actually get less input lag. But I want to clarify some points.
I currently have two systems. 1709 and 2009. Both systems have a timer less than 0.5ms.
With bcdedit settings like useplatformclock no, and useplatformtick yes (1709) QPC 3.51
And useplatformclock yes, and useplatformtick yes (2009) ACPI 3.58
I also used the clock yes parameter on 1709 and didn't see any noticeable difference in anything.
but there is a performance problem, if at 1709 in the single-core cpu-z test I get 530 points (on both timers) then on the 2009 assembly I get 430 points, memory latency also suffers. on the 2009 build, this is solved by resetting the timer from 0.5 to 15ms. which affects the delay. I roughly understand why 2009 loses, but at the same time I do not understand why this happens on the same timers ACPI + TSC (clock yes + tick yes). Maybe I'm confused about the timers?
What timer do I use if I turn off clock no and tick no? at 1709 this does the timer 0.50013 but which timer is it?
I do not intend to enable hpet in the BIOS because I will not be able to return to this lag again.
I have questions but no answers.
1) About USB.
1.Is usb 2.0 better for mice than usb 3.0?
2. xhci hand off disabled make usb 3.0 = usb 2.0?
3. If I buy a pci-e usb 2.0 extension will I get more latency than I would use the ports on the motherboard?
2) CSM better in Mousetester, but uefi feels better. placebo or is there an explanation? MBR or GPT?
3) I have experience with hpet disabled in bios. I actually get less input lag. But I want to clarify some points.
I currently have two systems. 1709 and 2009. Both systems have a timer less than 0.5ms.
With bcdedit settings like useplatformclock no, and useplatformtick yes (1709) QPC 3.51
And useplatformclock yes, and useplatformtick yes (2009) ACPI 3.58
I also used the clock yes parameter on 1709 and didn't see any noticeable difference in anything.
but there is a performance problem, if at 1709 in the single-core cpu-z test I get 530 points (on both timers) then on the 2009 assembly I get 430 points, memory latency also suffers. on the 2009 build, this is solved by resetting the timer from 0.5 to 15ms. which affects the delay. I roughly understand why 2009 loses, but at the same time I do not understand why this happens on the same timers ACPI + TSC (clock yes + tick yes). Maybe I'm confused about the timers?
What timer do I use if I turn off clock no and tick no? at 1709 this does the timer 0.50013 but which timer is it?
I do not intend to enable hpet in the BIOS because I will not be able to return to this lag again.