News for hackers. Stuff that matters.

A comment on Task Failed Successfully: Saturating NIC and Disk Bandwidth.

Re:Task Failed Successfully: Saturating NIC and Disk Bandwidth by nycerrrrrrrrrr on Saturday June 27, 2026 @12:12PM (#48699457)
This might be orthogonal to the TLB miss overhead you found, but have you looked at using P2PDMA to transfer directly from the NVMe SSDs to the NIC? Not sure how the CRC calculation would play into that.
[ Parent | on HN ]
Re:Task Failed Successfully: Saturating NIC and Disk Bandwidth by MrCroxx on Sunday June 28, 2026 @04:42AM (#48705549)
Thank you for your reply. This is a long-running service. Without CRC validation, errors caused by partial writes could accumulate over time and affect correctness. Therefore, we adopted this approach.
[ Parent | on HN ]
Re:Task Failed Successfully: Saturating NIC and Disk Bandwidth by nycerrrrrrrrrr on Sunday June 28, 2026 @10:51AM (#48707820)
Depending on your hardware you may be able to do the CRC validation in the NIC. If you're using Intel you could use DSA also, but then you're still copying the data through DRAM.

Just throwing out some ideas, obviously the best solution is the one that you already have working :)

[ Parent | on HN ]