POLYVAL Work Bringing More Performance Gains To Linux Crypto Subsystem
Whenever seeing Linux kernel mailing list patches from Google engineer Eric Biggers it tends to be about performance optimizations to the Linux kernel's cryptography subsystem. That was once again the case on Sunday with the newest patch series providing some nice gains.
A set of nine patches were sent out by Biggers to migrate the POLYVAL code to the lib/crypto area of the kernel and to replace a generic implementation of POLYVAL with a superior implementation.
The key takeaway from this work:
So a nice win from this patch series now under review for potentially making it into the next version of the Linux kernel.
A set of nine patches were sent out by Biggers to migrate the POLYVAL code to the lib/crypto area of the kernel and to replace a generic implementation of POLYVAL with a superior implementation.
The key takeaway from this work:
"Notably, this series improves the performance of HCTR2, since it eliminates unnecessary overhead that was being incurred by accessing POLYVAL via the crypto_shash API. I see a 45% increase in throughput with 64-byte messages, 53% with 128-byte, or 6% with 4096-byte.
It also eliminates the need to explicitly enable the optimized POLYVAL code, as it's now enabled automatically when HCTR2 support is enabled."
👁 POLYVAL improved implementation benchmarks
So a nice win from this patch series now under review for potentially making it into the next version of the Linux kernel.