On 1/7/2021 7:44 AM, Leyi Rong wrote:
As eal parameter --force-max-simd-bitwidth is already introduced,
to make it more clear when setting rx/tx function, remove
devarg use-latest-supported-vec support.
Signed-off-by: Leyi Rong <leyi.r...@intel.com>
<...>
@@ -3154,20 +3124,33 @@ i40e_set_rx_function(struct rte_eth_dev *dev)
break;
}
}
+
+ if ((rte_cpu_get_flag_enabled(RTE_CPUFLAG_AVX2) == 1 ||
+ rte_cpu_get_flag_enabled(RTE_CPUFLAG_AVX512F) == 1)
&&
+ rte_vect_get_max_simd_bitwidth() >=
RTE_VECT_SIMD_256)
+ use_avx2 = true;
}
Hi Leyi,
The cpu flags, 'RTE_CPUFLAG_AVX2' & 'RTE_CPUFLAG_AVX512F', are only defined for
x86 and causing build error for other architectures.
And what about extracting that logic into a static inline function, this makes
code more clean, and in that function other architectures return 'use_avx2'
false straightaway and can do the checks only for x86?
