[PATCH v2 0/1] x86: Tune non-temporal copy threshold for Hygon

[Jiamei Xie]

This patch adjusts the non-temporal copy threshold for Hygon
processors.

Benchmarks on Hygon processors show that the default non-temporal
threshold is higher than ideal for large copy workloads. With the
current threshold, memcpy and memmove may continue to use the temporal
copy path for longer than is beneficial, reducing throughput for large
copies.

The patch lowers the copy non-temporal threshold to 3/8 of the shared
cache size per thread on Hygon, allowing the non-temporal copy path to
be selected earlier. The memset non-temporal threshold is left
unchanged.


Changes in v2:
  - Move the Hygon default threshold into the else branch of the
    x86_non_temporal_threshold tunable handling.
  - Clarify in the comment that the memset threshold has already been
    initialized and is intentionally left unchanged.

xiejiamei (1):
  x86: Lower non-temporal copy threshold for Hygon

 sysdeps/x86/dl-cacheinfo.h | 7 +++++++
 1 file changed, 7 insertions(+)

-- 
2.43.0