My 8000hz mouse is smooth, but unstable.

[Slender]

I bought a hyperx pulsefire haste 2, it's the best mouse I've ever had, but has the same problems as the viper 8k. At 8000hz the game looks super smooth (xl2566k), but it feels like the mouse is stuck, like I need to put in more effort than at 1000hz and the line is unstable.
mobo/cpu : msi z790 edge d4 / i7-13700kf (configured bios), GGOS / ReviOS
Also, im try internal usb 2.0 Gl usb chip and mcs9990 pcie usb 2.0 card. It feel better then internal GL 3.0 and z790 usb controller but it not fix the problem.

[bumbeen]

Am actually surprised most of the samples are >7khz tbh. It's quite normal to see many result in these testing tools showing lower than configure poll rate when mouse isn't moving fast. I have haste 2 and also DAv3 wired, but i only ever used them at 2000 max. are you saying you have to move mouse further than normally you would have to? I wonder if the game is skipping some input from the mouse? Have you tried changing the core affinity of the usb controller from default to share across all cpu core? This is just a wild guess on my part btw, like I said I have never used 8000hz.

I would be curious what rawaccel shows on graph

[mago]

Razer 8k works better at 500hz for me
higher polling rate will reduce the input lag make it smoother
but after trying to finish aim process, toward the head, it will start stuttering

at 500hz there not having big input lag, different mouse probably will cos the PC, ig
and not that accurate however it better then have stuttering when trying aiming to the head

i use high dpi 3200dpi and 0.45 sense @ csgo
helps with input lag a bit more

[Slender]

Am actually surprised most of the samples are >7khz tbh. It's quite normal to see many result in these testing tools showing lower than configure poll rate when mouse isn't moving fast. I have haste 2 and also DAv3 wired, but i only ever used them at 2000 max. are you saying you have to move mouse further than normally you would have to? I wonder if the game is skipping some input from the mouse? Have you tried changing the core affinity of the usb controller from default to share across all cpu core? This is just a wild guess on my part btw, like I said I have never used 8000hz.

I would be curious what rawaccel shows on graph

yea, im tried interrupt controller affinity = same.

friend with z490 + DA3 8000hz

f8ubuoT.png (146.14 KiB) Viewed 15906 times

[bumbeen]

Am actually surprised most of the samples are >7khz tbh. It's quite normal to see many result in these testing tools showing lower than configure poll rate when mouse isn't moving fast. I have haste 2 and also DAv3 wired, but i only ever used them at 2000 max. are you saying you have to move mouse further than normally you would have to? I wonder if the game is skipping some input from the mouse? Have you tried changing the core affinity of the usb controller from default to share across all cpu core? This is just a wild guess on my part btw, like I said I have never used 8000hz.

I would be curious what rawaccel shows on graph

yea, im tried interrupt controller affinity = same.

friend with z490 + DA3 8000hz
...

Now that's very interesting! I wonder if this is so much more consistent because of motion sync? Although I had read somewhere that motionsync only works at 4khz and below, but danged if that line doesn't look real consistent even with the outliers.

Since I've got both the haste 2 and the dav3 I will test both on my machine and see what my graphs look like. I am on AMD instead of intel though, B650 mobo and 7800x3d. I'll update this post in I guess 20-30 minutes depending how long it takes me to figure out that mouse tester tool!

all at 4800dpi. I opened testufo ghosting test at 1920pps, and started mouse on left side of screen and tried to track UFO to right side of screen, then a quick flick back to the left side of screen and tracked again. I did this 5 times and then stopped the capture for each of these examples:

dav3 2000hz


dav3 8000hz


haste2 2000hz


haste2 8000hz


viper v2 pro 1000hz

[Slender]

Am actually surprised most of the samples are >7khz tbh. It's quite normal to see many result in these testing tools showing lower than configure poll rate when mouse isn't moving fast. I have haste 2 and also DAv3 wired, but i only ever used them at 2000 max. are you saying you have to move mouse further than normally you would have to? I wonder if the game is skipping some input from the mouse? Have you tried changing the core affinity of the usb controller from default to share across all cpu core? This is just a wild guess on my part btw, like I said I have never used 8000hz.

I would be curious what rawaccel shows on graph

yea, im tried interrupt controller affinity = same.

friend with z490 + DA3 8000hz
...

Now that's very interesting! I wonder if this is so much more consistent because of motion sync? Although I had read somewhere that motionsync only works at 4khz and below, but danged if that line doesn't look real consistent even with the outliers.

Since I've got both the haste 2 and the dav3 I will test both on my machine and see what my graphs look like. I am on AMD instead of intel though, B650 mobo and 7800x3d. I'll update this post in I guess 20-30 minutes depending how long it takes me to figure out that mouse tester tool!

all at 4800dpi. I opened testufo ghosting test at 1920pps, and started mouse on left side of screen and tried to track UFO to right side of screen, then a quick flick back to the left side of screen and tracked again. I did this 5 times and then stopped the capture for each of these examples:

dav3 2000hz


dav3 8000hz


haste2 2000hz


haste2 8000hz


viper v2 pro 1000hz

thanks for test, but as i can see, you have very high jitter.
try to stabilize it to 990-1010 on 1000hz.
maybe you use some wrong settings with timers or bios?

[bumbeen]

Wireshark USB capture shows nearly perfectly timed intervals. I think it's just that mousetester software not being that accurate.

from a capture of ~21,000 packets received from viper v2 pro at 2000hz and 4800dpi, I get a nearly perfect 500ms interval and standard deviation of 1.984534860162605e-05. The worst ones are only off by 75 microseconds.

Median of differences: 0.0004999999999988347
Standard Deviation of differences: 1.984534860162605e-05
Outliers of differences:
43 0.000561
44 0.000439
182 0.000566
183 0.000435
184 0.000573
...
20756 0.000414
20955 0.000575
20956 0.000425
21100 0.000565
21101 0.000436

[Slender]

Wireshark USB capture shows nearly perfectly timed intervals. I think it's just that mousetester software not being that accurate.

from a capture of ~21,000 packets received from viper v2 pro at 2000hz and 4800dpi, I get a nearly perfect 500ms interval and standard deviation of 1.984534860162605e-05. The worst ones are only off by 75 microseconds.

Median of differences: 0.0004999999999988347
Standard Deviation of differences: 1.984534860162605e-05
Outliers of differences:
43 0.000561
44 0.000439
182 0.000566
183 0.000435
184 0.000573
...
20756 0.000414
20955 0.000575
20956 0.000425
21100 0.000565
21101 0.000436

can you write a guide how to use that software?

[bumbeen]

you just need to install wireshark, select the usbcap option when installing it. Capture your usb device and make sure to disable "update display of packets during capture". Then you export the packets as a CSV file, then run this python script with your csv:

Code: Select all

import pandas as pd
import numpy as np
from scipy import stats

# Load the data
df = pd.read_csv('razerviper2000hz.csv')

# Ensure that Time column is numeric
df['Time'] = pd.to_numeric(df['Time'], errors='coerce')

# Calculate time differences between consecutive rows
df['Time_diff'] = df['Time'].diff()

# Drop the rows with NaN values in 'Time_diff' column
df = df.dropna(subset=['Time_diff'])

# Calculate the median of differences
median_diff = df['Time_diff'].median()
print('Median of differences:', median_diff)

# Calculate the standard deviation of differences
std_dev_diff = df['Time_diff'].std()
print('Standard Deviation of differences:', std_dev_diff)

# Calculate outliers of differences using the Z-score method
z_scores_diff = np.abs(stats.zscore(df['Time_diff']))
outlier_threshold = 3
outliers_diff = df['Time_diff'][z_scores_diff > outlier_threshold]

print('Outliers of differences:')
print(outliers_diff)

[Chief Blur Buster]

Don't forget to test different USB ports, including a dedicated PCIe USB card just only for the 8KHz mouse.

I usually run my Razer 8KHz at 2-4KHz, since you get 90% of the human visible benefits of 8KHz at just 2Khz. Most systems seem to be unable to keep up at 8KHz, so the sweet spot is 2KHz or 4KHz.