Review Corsair Sabre RGB Pro Mouse Review: the High CPU Cost of 8,000 Hz

[Admin]

Aimed at eSports professionals, the Sabre RGB Pro hits 8,000 Hz but requires a high-end CPU to do so.

Corsair Sabre RGB Pro Mouse Review: the High CPU Cost of 8,000 Hz : Read more

 

[mac_angel]

Heaven forbid you ever have to deal with Corsair for warranty though. I don't know if they are outsourcing their tech support or what, but they refused to honour the warranty of my quad channel RAM kit with the argument "dual channel is the same"

 

[g-unit1111]

Finally iCue gets a much needed overhaul. It's ridiculously frustrating in its' current form.

 

[thepersonwithaface45]

Finally iCue gets a much needed overhaul. It's ridiculously frustrating in its' current form.

Oh? They updated it?? That could be an article in itself, just looked at the bottom portion of the article and it do be lookin pretty nice. Always had trouble confirming if my changes were saved or not.

 

[g-unit1111]

Oh? They updated it?? That could be an article in itself, just looked at the bottom portion of the article and it do be lookin pretty nice. Always had trouble confirming if my changes were saved or not.

Hopefully the new version is slightly less of a PITA. But it is annoying attempting to use forward / back functions on any Corsair mouse. I bought a Logitech G903 and it's worlds better than my Vengeance M65.

 

[Kamen Rider Blade]

Would pluging this mouse into a dedicated USB Controller Card lessen the CPU load?

 

[Makaveli]

Hopefully the new version is slightly less of a PITA. But it is annoying attempting to use forward / back functions on any Corsair mouse. I bought a Logitech G903 and it's worlds better than my Vengeance M65.

I don't have any issues with forward and back with my corsair mouse.

 

[escksu]

Mouse 8000hz but monitor is 144hz (some models 200hz or more)... but not all game could hit 144 or 200fps...some can but not all.

So regardless of how many times your mouse poll, its still limited by your fps and monitor refresh rate. its not to say that its useless, but diminishing returns.

 

[cryoburner]

An 8,000 Hz polling rate seems a bit excessive. Even at 1,000 Hz, your system will be reading the mouse's position roughly four times for every frame on a 240Hz display. I suppose that could mean up to 24% jitter from one frame to the next on such a high-refresh screen though, depending on whether the most recent update was immediately before the frame was rendered, or a quarter frame prior, which might make movement a little less smooth than it could be. At 2,000 Hz the maximum jitter would drop to 12%, or 6% at 4,000 Hz, and 3% at 8,000 Hz. We are only talking fractions of a millisecond though, on a frame that takes multiple milliseconds to be displayed, so I'm not convinced there would be a noticeable difference between something like 4,000 and 8,000 Hz. And that's assuming the OS is even polling the mouse at a perfect 8,000 Hz rate, which it might not be if other other things are interrupting it.

Would pluging this mouse into a dedicated USB Controller Card lessen the CPU load?

The drivers still needs to read all the data about the mouse's position, button-presses and so on each time it is polled, then possibly process that data in some way before reporting it to the OS, which will in turn store the data in other locations in memory. There are probably a lot of inefficiencies along the way, especially since the standard USB polling rate is just 125 Hz, whereas this is 64 times that. If the OS and drivers were being smart about it, this might all be getting done on a dedicated core on processors with enough cores to spare, though that's probably not happening.

 

[escksu]

Would pluging this mouse into a dedicated USB Controller Card lessen the CPU load?

Nope. Because USB cards are simply just phy controllers on a card. No different from those soldered on your board.

I understand where you are coming from. A dedicated processor to offload task from the cpu (similar to server NIC and raid controllers). However, it may not be desirable as it will introduce additional latencies.

 

[Kamen Rider Blade]

Nope. Because USB cards are simply just phy controllers on a card. No different from those soldered on your board.

I understand where you are coming from. A dedicated processor to offload task from the cpu (similar to server NIC and raid controllers). However, it may not be desirable as it will introduce additional latencies.

I was just hoping for a dedicated USB Processor per USB port, that would allow each port to be a Root Port/Hub.

https://wiki.fractalaudio.com/wiki/index.php?title=CPU_usage

This way, no matter the polling rate, no matter how much data you send through the USB port, the impact on your CPU usage would be little to negligible.

 

[escksu]

I was just hoping for a dedicated USB Processor per USB port, that would allow each port to be a Root Port/Hub.

https://wiki.fractalaudio.com/wiki/index.php?title=CPU_usage

This way, no matter the polling rate, no matter how much data you send through the USB port, the impact on your CPU usage would be little to negligible.

Yes, it is possible to have a PHY controller coupled with a dedicated processor or ASIC (or All in one). Its not that different from existing NIC/RAID controllers. Unfortunately, I don't think anyone makes such a controller. Even if there is, I am not sure if it can handle 8000Hz polling rate as its way beyond USB specification of 125Hz.

If it is made, I believe it will be extremely niche and likely cost more than the mouse (easily USD100 or more).

 

[Kamen Rider Blade]

Yes, it is possible to have a PHY controller coupled with a dedicated processor or ASIC (or All in one). Its not that different from existing NIC/RAID controllers. Unfortunately, I don't think anyone makes such a controller. Even if there is, I am not sure if it can handle 8000Hz polling rate as its way beyond USB specification of 125Hz.

If it is made, I believe it will be extremely niche and likely cost more than the mouse (easily USD100 or more).

Make that UPU (USB Processor Unit) ASIC multi-core where it can handle as many USB Ports as there are on the MotherBoard's Rear IO & Front Panel Headers and I gurantee you that there will be a market for those who want a MoBo with dedicated processing for those ports.

 

[jasonkaler]

They've really messed up their drivers if they need 16% cpu.
I would be worried if a mouse even needs 0.16% cpu.

The only processing needed is Add to X, Add to Y, And the mouse buttons, the original IBM XT at 4Mz could do all this a million times a second
Here they only need to do it 8000 times a second and CPUs are 1000 times faster.

Maybe they should fire their programmers and hire someone that can code in c++
Or get rid of all the stupid bloat.

 

[escksu]

They've really messed up their drivers if they need 16% cpu.
I would be worried if a mouse even needs 0.16% cpu.

The only processing needed is Add to X, Add to Y, And the mouse buttons, the original IBM XT at 4Mz could do all this a million times a second
Here they only need to do it 8000 times a second and CPUs are 1000 times faster.

Maybe they should fire their programmers and hire someone that can code in c++
Or get rid of all the stupid bloat.

I believe its due to the usb interface that is not designed for 8000hz polling rate.

 

[escksu]

Make that UPU (USB Processor Unit) ASIC multi-core where it can handle as many USB Ports as there are on the MotherBoard's Rear IO & Front Panel Headers and I gurantee you that there will be a market for those who want a MoBo with dedicated processing for those ports.

I dont think there will be much of a demand.

 

[fmyhr]

Maybe instead of insane USB polling rates it's time to go back to PS/2 ports, with essentially zero latency. (Check out Wikipedia page...)