Sensor

PixArt PMW3391—a sensor that's rather unique and only available in Corsair mice as of now. It's based on the PMW3389, so it has very similar specifications, though there are some improvements. This tracking unit is just fantastic, absolutely raw and responsive at even its maximum tracking speed, without any added acceleration, angle snapping, or any sort of filtering. There's some smoothing on and above 2100 CPI, which is usually the case with most newer PixArt sensors, and it's there to reduce jitter on the higher resolution steps.

Now for some specifications—the nominal maximum tracking speed and acceleration values are ridiculously high at 400 IPS and 50 G; there is no way of hitting these values during regular use. Resolution can be set from 100–18,000 CPI in steps of 1 CPI (which is still unique to this sensor). The available polling rates are 1000, 500, 250 and 125 Hz, which respectively translates to nominal response times of 1, 2, 4 and 8 ms. Lift-off distance can be configured in the software's surface calibration option, which I highly recommend doing because the default LOD is unusually high.

Paint Test

There is no jitter on the reasonable CPI steps, and I couldn't detect any unwanted angle snapping or sensor rattle either.

CPI Divergence

CPI divergence is essentially the same as on all PMW3391-equipped mice, which means the measured values are 5–7% higher than the nominal ones. Based on my testing, this is just a tad higher than average, so there's nothing to worry about—if you come from a mouse with a perfect CPI accuracy, you should consider lowering your in-game sensitivity, though.

Perfect Control Speed

Perfect control speed (or PCS for short) is extremely high on this sensor as 400 IPS is over 10.6 m/s. There is absolutely no way of hitting it while using the mouse normally—hitting 400 IPS will still be very challenging on even a 2 m long mouse pad.

Speed Related Accuracy Variance

This test shows the sensor's accuracy at different speeds. You can see me doing a fast swipe to the right before I slowly slide the mouse back to its original position. There is pretty much no acceleration or deceleration here.

Polling Rate & Stability

We are at a point where I can again (for a third time so far) objectively complain about RGB lighting. While I wasn't exactly sure about this on the Ironclaw Wireless, this mouse certainly shows how LEDs can affect performance. With any of the three RGB lighting zones active, the mouse periodically drops its polling rates, which produces serious anomalies on measurements. This means the Glaive RGB Pro is unstable with any lighting, which is just outrageous and shouldn't be overlooked if you ask me.

Once I disabled all customizable LEDs, the polling rate instantly stabilized, and the mouse performed perfectly without any errors. If you remember the Cooler Master MM830 review, this issue was present there as well, and they could solve it with a firmware update—I hope Corsair will follow their example, and I have notified them about the matter.

With disabled RGBs, all available polling rates are nice and stable, without any periodic drops or suspicious outliers.

Input Lag & Smoothing

There is no measurable smoothing or any other delay below 2,100 CPI. Some smoothing kicks in at or above 2,100 CPI, resulting in about +2–3 ms of input lag. It reaches about +10 ms on 18,000 CPI, which is the highest available resolution. For competitive gaming, I would highly advise staying below 2,100 CPI for no smoothing. For anything else, feel free to use whatever you like that's within reasonable bounds. I personally have no clue why peripherals-producing companies still fight this CPI war. In my opinion, anything above 6,000 is ridiculous on even a 4K display, but hey, it is again a matter of personal preference.

Click Latency

Click latency is roughly +4.4 ms when compared to the SteelSeries Ikari, which is considered as the baseline with 0 ms. The data comes from this thread and my own testings (please note that this review's chart only contains right-handed ergonomic mice for comparison). Testing was done with a Logitech G102 and the Glaive, using qsxcv's program.