From: Sun Yuechi <sunyue...@iscas.ac.cn>
This patch series adds support for the RISC-V Vector (V) extension and
provides an optimized implementation of `rte_lpm_lookupx4` using RVV.
The initialization of vtbl_entry is not fully vectorized here because
doing so would require __riscv_vluxei32_v_u32m1, which is slower
than the scalar approach in this small-scale scenario.
- Test: app/test/lpm_perf_autotest
- Platform: Banana Pi BPI-F3
- SoC: Spacemit X60 (8 cores with Vector extension)
- CPU Frequency: up to 1.6 GHz
- Cache: 256 KiB L1d ×8, 256 KiB L1i ×8, 1 MiB L2 ×2
- Memory: 16 GiB
- Kernel: Linux 6.6.36
- Compiler: GCC 14.2.0 (with RVV intrinsic support)
Test results(LPM LookupX4):
scalar: 5.7 cycles
rvv: 4.6 cycles
v5:
- Updated the commit message and moved test details and thoughts to the cover
letter.
v4:
- Keep the LPM code consistent, use a static inline function instead of runtime
detection.
- Update the commit message.
- Update the .mailmap file.
v3:
- Due to previous SMTP server restrictions, patches could not be sent as a
proper series.
This version re-sends the same patches as a series. No code changes.
v2:
- Improved commit messages.
Sun Yuechi (3):
config/riscv: detect V extension
lib/lpm: R-V V rte_lpm_lookupx4
riscv: override machine_args only when default
.mailmap | 1 +
MAINTAINERS | 2 ++
config/riscv/meson.build | 27 +++++++++++++++++
lib/lpm/meson.build | 1 +
lib/lpm/rte_lpm.h | 2 ++
lib/lpm/rte_lpm_rvv.h | 62 ++++++++++++++++++++++++++++++++++++++++
6 files changed, 95 insertions(+)
create mode 100644 lib/lpm/rte_lpm_rvv.h
--
2.49.0
