News for hackers. Stuff that matters.
Re:Tinygrad: Hacked 4090 driver to enable P2P by wtallis on Friday April 12, 2024 @04:09PM (#40017121)
PCIe p2p transactions that end up routed through the CPU's PCIe root complex still have performance advantages over split transactions using the CPU's DRAM as an intermediate buffer. Separate PCIe switches are not necessary except when the CPU doesn't support routing p2p transactions, which IIRC was not a problem on anything more mainstream than IBM POWER.
Re:Tinygrad: Hacked 4090 driver to enable P2P by numpad0 on Saturday April 13, 2024 @04:53AM (#40021594)
Maybe not strictly necessary, but a separate PCIe backplane just for P2P bandwidth bypasses topology and bottleneck mess[1][2] of PC platform altogether and might be useful. I suspect this was the original premise for NVLink too.

1: https://assets.hardwarezone.com/img/2023/09/pre-meteror-lake...

2: https://www.gigabyte.com/FileUpload/Global/MicroSite/579/inn...