Question X570

[BTRY B 529th FA BN]

I'm considering a PCIe USB card for the CPU PCIe X16 lanes but before I do that I have 2 other options I could use to spread out my I/O devices. I currently have the mouse on the CPU USB Host Controller with the KB on the X570 USB Host controller. My dedicated sound card also runs through the X570 chip, as-well-as the onboard Intel i211 NIC. I also have a spare Intel Gigabit CT Adapter NIC; so I can put either the sound card or the spare NIC on the X16 CPU lanes. What would benefit the most, the sound card or the spare NIC? If neither should I just get a PCIe USB card for the CPU lanes?


Thanks for your time.

 

[UsandThem]

USB cards, NICs, and sound cards don't need x16 lanes.

They work just fine in x1 slots. There's no need to manually "spread them out" at all. BTW, do you even have need for an additional NIC add-in card compared to the onboard NIC slots (which many x570 motherboards have two onboard ones)?

 

[BTRY B 529th FA BN]

Well I'd be splitting the CPU PCIe x16 lanes: the point is to take the device with the higher amount of interrupts and put it directly onto the CPU PCIe lanes. I have 2 PCIe x16 slots that are linked directly to the CPU. Obviously it will split the slot down to x8; the GPU is a 3080 Ti.

Yeah it has 2 onboard NICs that are linked through the X570 chipset. The Intel and an Aquantia? even a WiFi but I don't use that. The system is less than 2' away from the router and using wired connection.

AFAIK splitting a KB & Mouse onto different Host controllers will increase mouse accuracy and alleviate Host controller congestion. For an experiment take both your devices and run them on the same USB Host hub and then try them on separate Host controllers during gaming. For me this is noticeable even on two 1K-Hz devices. Some games might not exploit the "problem" but for me it's noticeable. It became more apparent with 2 8K-Hz devices (M & KB) This has been demonstrated via MouseTester in a troubleshooting thread.

So I'm just taking the idea and extrapolating it with the CPU PCIe lanes. Again I have 2 CPU PCIe X16 slots which will reduce down to x8 slots when both are in use. I currently am using the GPU in the main PCIe x16 slot, the second x16 slot is free, and the bottom hardwired x4 slot (X570/PCH) is being used by the Sound card, and the Onboard Intel NIC I'm using is going through the X570/PCH.

 

[Insert_Nickname]

I also have a spare Intel Gigabit CT Adapter NIC; so I can put either the sound card or the spare NIC on the X16 CPU lanes. What would benefit the most, the sound card or the spare NIC?

What board are you using?

Neither benefits at all. They're perfectly good in ordinary PCIe x1 slots, since neither use anywhere near even PCIe x1 1.0 bandwidth (250MB/s). If it's busmastering you're worried about, all PCIe slots are their own busses. So not an issue at all.

I'd go so far as to say what you're suggesting is detrimental to your PCs performance since you're sacrificing 8 of the CPU-to-Graphics lanes for no gain.

 

[BTRY B 529th FA BN]

So with 4 PCIe 4.0 lanes going to the X570 chip how is this any different than a HEDT with it's 128 lanes, I mean of course the lane count but,, What does the X570 platform lose with the X570 chip? IIRC isn't it the 12nm IMC chip?

 

[Leeea]

AFAIK splitting a KB & Mouse onto different Host controllers will increase mouse accuracy and alleviate Host controller congestion. For an experiment take both your devices and run them on the same USB Host hub and then try them on separate Host controllers during gaming. For me this is noticeable even on two 1K-Hz devices. Some games might not exploit the "problem" but for me it's noticeable. It became more apparent with 2 8K-Hz devices (M & KB) This has been demonstrated via MouseTester in a troubleshooting thread.

You are confusing polling rate with bandwidth.

So your standard mouse takes 4 bytes to send all its data. That is wrapped up in a USB packet, which will make that about 8 bytes in size.

An old USB 1.1 connection from the beginning of time communicates at 1,500,000 bytes per second. Which means there is enough bandwidth on the 1990's USB channel to poll a mouse or keyboard at 187500 times per second.

The 1000 Hz ( or times per second ) polling rate is a limitation of the mouse or keyboard.

You could put over a 100 modern gaming mice on a USB port from 1999 with no lag.

Your just seeing a placebo effect.

So with 4 PCIe 4.0 lanes going to the X570 chip how is this any different than a HEDT with it's 128 lanes, I mean of course the lane count but,, What does the X570 platform lose with the X570 chip? IIRC isn't it the 12nm IMC chip?

Lose?

For the home user, nothing.


HEDT is for systems with
3 or more video cards
5 or more NVMe drives.
48 or more USB 3.0 ports
384 or more USB 2.0 ports
40 or more SATA 3 SSDs
6,000,000,000 or more mice and keyboards