* [PATCH 1/2] net/i40e: use common AVX2 Tx function
2026-09-23 13:51 [PATCH 0/2] net/i40e: use common AXV vector code Bruce Richardson
@ 2026-09-23 13:51 ` Bruce Richardson
2026-09-25 10:10 ` Burakov, Anatoly
2026-09-23 13:51 ` [PATCH 2/2] net/i40e: use common AVX-512 " Bruce Richardson
2026-09-25 14:33 ` [PATCH 0/2] net/i40e: use common AXV vector code Bruce Richardson
2 siblings, 1 reply; 6+ messages in thread
From: Bruce Richardson @ 2026-09-23 13:51 UTC (permalink / raw)
To: dev; +Cc: Bruce Richardson
Replace the standard (non-offloading) AVX2 Tx function with the one from
the intel/common folder. Both old and new code paths have identical
behaviour.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
---
drivers/net/intel/i40e/i40e_rxtx_vec_avx2.c | 121 +-------------------
1 file changed, 3 insertions(+), 118 deletions(-)
diff --git a/drivers/net/intel/i40e/i40e_rxtx_vec_avx2.c b/drivers/net/intel/i40e/i40e_rxtx_vec_avx2.c
index 88de303349..dd06e258ad 100644
--- a/drivers/net/intel/i40e/i40e_rxtx_vec_avx2.c
+++ b/drivers/net/intel/i40e/i40e_rxtx_vec_avx2.c
@@ -13,6 +13,7 @@
#include "i40e_rxtx_vec_common.h"
#include "../common/rx_vec_x86.h"
+#include "../common/tx_vec_x86.h"
#include <rte_vect.h>
@@ -680,122 +681,6 @@ i40e_recv_scattered_pkts_vec_avx2(void *rx_queue, struct rte_mbuf **rx_pkts,
}
-static inline void
-vtx1(volatile struct ci_tx_desc *txdp,
- struct rte_mbuf *pkt, uint64_t flags)
-{
- uint64_t high_qw = (CI_TX_DESC_DTYPE_DATA |
- ((uint64_t)flags << CI_TXD_QW1_CMD_S) |
- ((uint64_t)pkt->data_len << CI_TXD_QW1_TX_BUF_SZ_S));
-
- __m128i descriptor = _mm_set_epi64x(high_qw,
- pkt->buf_iova + pkt->data_off);
- _mm_store_si128(RTE_CAST_PTR(__m128i *, txdp), descriptor);
-}
-
-static inline void
-vtx(volatile struct ci_tx_desc *txdp,
- struct rte_mbuf **pkt, uint16_t nb_pkts, uint64_t flags)
-{
- const uint64_t hi_qw_tmpl = (CI_TX_DESC_DTYPE_DATA | (flags << CI_TXD_QW1_CMD_S));
-
- /* if unaligned on 32-bit boundary, do one to align */
- if (((uintptr_t)txdp & 0x1F) != 0 && nb_pkts != 0) {
- vtx1(txdp, *pkt, flags);
- nb_pkts--; txdp++; pkt++;
- }
-
- /* do two at a time while possible, in bursts */
- for (; nb_pkts > 3; txdp += 4, pkt += 4, nb_pkts -= 4) {
- uint64_t hi_qw3 = hi_qw_tmpl |
- ((uint64_t)pkt[3]->data_len << CI_TXD_QW1_TX_BUF_SZ_S);
- uint64_t hi_qw2 = hi_qw_tmpl |
- ((uint64_t)pkt[2]->data_len << CI_TXD_QW1_TX_BUF_SZ_S);
- uint64_t hi_qw1 = hi_qw_tmpl |
- ((uint64_t)pkt[1]->data_len << CI_TXD_QW1_TX_BUF_SZ_S);
- uint64_t hi_qw0 = hi_qw_tmpl |
- ((uint64_t)pkt[0]->data_len << CI_TXD_QW1_TX_BUF_SZ_S);
-
- __m256i desc2_3 = _mm256_set_epi64x(
- hi_qw3, pkt[3]->buf_iova + pkt[3]->data_off,
- hi_qw2, pkt[2]->buf_iova + pkt[2]->data_off);
- __m256i desc0_1 = _mm256_set_epi64x(
- hi_qw1, pkt[1]->buf_iova + pkt[1]->data_off,
- hi_qw0, pkt[0]->buf_iova + pkt[0]->data_off);
- _mm256_store_si256(RTE_CAST_PTR(__m256i *, txdp + 2), desc2_3);
- _mm256_store_si256(RTE_CAST_PTR(__m256i *, txdp), desc0_1);
- }
-
- /* do any last ones */
- while (nb_pkts) {
- vtx1(txdp, *pkt, flags);
- txdp++; pkt++; nb_pkts--;
- }
-}
-
-static inline uint16_t
-i40e_xmit_fixed_burst_vec_avx2(void *tx_queue, struct rte_mbuf **tx_pkts,
- uint16_t nb_pkts)
-{
- struct ci_tx_queue *txq = (struct ci_tx_queue *)tx_queue;
- volatile struct ci_tx_desc *txdp;
- struct ci_tx_entry_vec *txep;
- uint16_t n, nb_commit, tx_id;
- uint64_t flags = CI_TX_DESC_CMD_DEFAULT;
- uint64_t rs = CI_TX_DESC_CMD_RS | CI_TX_DESC_CMD_DEFAULT;
-
- if (txq->nb_tx_free < txq->tx_free_thresh)
- ci_tx_free_bufs_vec(txq, i40e_tx_desc_done, false);
-
- nb_commit = nb_pkts = (uint16_t)RTE_MIN(txq->nb_tx_free, nb_pkts);
- if (unlikely(nb_pkts == 0))
- return 0;
-
- tx_id = txq->tx_tail;
- txdp = &txq->ci_tx_ring[tx_id];
- txep = &txq->sw_ring_vec[tx_id];
-
- txq->nb_tx_free = (uint16_t)(txq->nb_tx_free - nb_pkts);
-
- n = (uint16_t)(txq->nb_tx_desc - tx_id);
- if (nb_commit >= n) {
- ci_tx_backlog_entry_vec(txep, tx_pkts, n);
-
- vtx(txdp, tx_pkts, n - 1, flags);
- tx_pkts += (n - 1);
- txdp += (n - 1);
-
- vtx1(txdp, *tx_pkts++, rs);
-
- nb_commit = (uint16_t)(nb_commit - n);
-
- tx_id = 0;
- txq->tx_next_rs = (uint16_t)(txq->tx_rs_thresh - 1);
-
- /* avoid reach the end of ring */
- txdp = &txq->ci_tx_ring[tx_id];
- txep = &txq->sw_ring_vec[tx_id];
- }
-
- ci_tx_backlog_entry_vec(txep, tx_pkts, nb_commit);
-
- vtx(txdp, tx_pkts, nb_commit, flags);
-
- tx_id = (uint16_t)(tx_id + nb_commit);
- if (tx_id > txq->tx_next_rs) {
- txq->ci_tx_ring[txq->tx_next_rs].cmd_type_offset_bsz |=
- rte_cpu_to_le_64(((uint64_t)CI_TX_DESC_CMD_RS) << CI_TXD_QW1_CMD_S);
- txq->tx_next_rs =
- (uint16_t)(txq->tx_next_rs + txq->tx_rs_thresh);
- }
-
- txq->tx_tail = tx_id;
-
- I40E_PCI_REG_WC_WRITE(txq->qtx_tail, txq->tx_tail);
-
- return nb_pkts;
-}
-
uint16_t
i40e_xmit_pkts_vec_avx2(void *tx_queue, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
@@ -808,8 +693,8 @@ i40e_xmit_pkts_vec_avx2(void *tx_queue, struct rte_mbuf **tx_pkts,
/* cross rs_thresh boundary is not allowed */
num = (uint16_t)RTE_MIN(nb_pkts, txq->tx_rs_thresh);
- ret = i40e_xmit_fixed_burst_vec_avx2(tx_queue, &tx_pkts[nb_tx],
- num);
+ ret = ci_xmit_fixed_burst_vec_avx2(tx_queue, &tx_pkts[nb_tx], num,
+ false, CI_TAG_IN_DATA_DESC, CI_TAG_IN_DATA_DESC);
nb_tx += ret;
nb_pkts -= ret;
if (ret < num)
--
2.53.0
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH 2/2] net/i40e: use common AVX-512 Tx function
2026-09-23 13:51 [PATCH 0/2] net/i40e: use common AXV vector code Bruce Richardson
2026-09-23 13:51 ` [PATCH 1/2] net/i40e: use common AVX2 Tx function Bruce Richardson
@ 2026-09-23 13:51 ` Bruce Richardson
2026-09-25 10:11 ` Burakov, Anatoly
2026-09-25 14:33 ` [PATCH 0/2] net/i40e: use common AXV vector code Bruce Richardson
2 siblings, 1 reply; 6+ messages in thread
From: Bruce Richardson @ 2026-09-23 13:51 UTC (permalink / raw)
To: dev; +Cc: Bruce Richardson
Replace the standard (non-offloading) AVX-512 Tx function with the one
from the intel/common folder. Both old and new code paths have identical
behaviour.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
---
drivers/net/intel/i40e/i40e_rxtx_vec_avx512.c | 113 +-----------------
1 file changed, 3 insertions(+), 110 deletions(-)
diff --git a/drivers/net/intel/i40e/i40e_rxtx_vec_avx512.c b/drivers/net/intel/i40e/i40e_rxtx_vec_avx512.c
index 0c2e699a95..ea2e6dc024 100644
--- a/drivers/net/intel/i40e/i40e_rxtx_vec_avx512.c
+++ b/drivers/net/intel/i40e/i40e_rxtx_vec_avx512.c
@@ -13,6 +13,7 @@
#include "i40e_rxtx_vec_common.h"
#include "../common/rx_vec_x86.h"
+#include "../common/tx_vec_x86.h"
#include <rte_vect.h>
@@ -749,114 +750,6 @@ i40e_recv_scattered_pkts_vec_avx512(void *rx_queue,
rx_pkts + retval, nb_pkts);
}
-static inline void
-vtx1(volatile struct ci_tx_desc *txdp, struct rte_mbuf *pkt, uint64_t flags)
-{
- uint64_t high_qw = (CI_TX_DESC_DTYPE_DATA |
- ((uint64_t)flags << CI_TXD_QW1_CMD_S) |
- ((uint64_t)pkt->data_len << CI_TXD_QW1_TX_BUF_SZ_S));
-
- __m128i descriptor = _mm_set_epi64x(high_qw,
- pkt->buf_iova + pkt->data_off);
- _mm_store_si128(RTE_CAST_PTR(__m128i *, txdp), descriptor);
-}
-
-static inline void
-vtx(volatile struct ci_tx_desc *txdp,
- struct rte_mbuf **pkt, uint16_t nb_pkts, uint64_t flags)
-{
- const uint64_t hi_qw_tmpl = (CI_TX_DESC_DTYPE_DATA | (flags << CI_TXD_QW1_CMD_S));
-
- for (; nb_pkts > 3; txdp += 4, pkt += 4, nb_pkts -= 4) {
- uint64_t hi_qw3 = hi_qw_tmpl |
- ((uint64_t)pkt[3]->data_len << CI_TXD_QW1_TX_BUF_SZ_S);
- uint64_t hi_qw2 = hi_qw_tmpl |
- ((uint64_t)pkt[2]->data_len << CI_TXD_QW1_TX_BUF_SZ_S);
- uint64_t hi_qw1 = hi_qw_tmpl |
- ((uint64_t)pkt[1]->data_len << CI_TXD_QW1_TX_BUF_SZ_S);
- uint64_t hi_qw0 = hi_qw_tmpl |
- ((uint64_t)pkt[0]->data_len << CI_TXD_QW1_TX_BUF_SZ_S);
-
- __m512i desc0_3 =
- _mm512_set_epi64
- (hi_qw3, pkt[3]->buf_iova + pkt[3]->data_off,
- hi_qw2, pkt[2]->buf_iova + pkt[2]->data_off,
- hi_qw1, pkt[1]->buf_iova + pkt[1]->data_off,
- hi_qw0, pkt[0]->buf_iova + pkt[0]->data_off);
- _mm512_storeu_si512(RTE_CAST_PTR(void *, txdp), desc0_3);
- }
-
- /* do any last ones */
- while (nb_pkts) {
- vtx1(txdp, *pkt, flags);
- txdp++; pkt++; nb_pkts--;
- }
-}
-
-static inline uint16_t
-i40e_xmit_fixed_burst_vec_avx512(void *tx_queue, struct rte_mbuf **tx_pkts,
- uint16_t nb_pkts)
-{
- struct ci_tx_queue *txq = (struct ci_tx_queue *)tx_queue;
- volatile struct ci_tx_desc *txdp;
- struct ci_tx_entry_vec *txep;
- uint16_t n, nb_commit, tx_id;
- uint64_t flags = CI_TX_DESC_CMD_DEFAULT;
- uint64_t rs = CI_TX_DESC_CMD_RS | CI_TX_DESC_CMD_DEFAULT;
-
- if (txq->nb_tx_free < txq->tx_free_thresh)
- ci_tx_free_bufs_vec(txq, i40e_tx_desc_done, false);
-
- nb_commit = nb_pkts = (uint16_t)RTE_MIN(txq->nb_tx_free, nb_pkts);
- if (unlikely(nb_pkts == 0))
- return 0;
-
- tx_id = txq->tx_tail;
- txdp = &txq->ci_tx_ring[tx_id];
- txep = (void *)txq->sw_ring;
- txep += tx_id;
-
- txq->nb_tx_free = (uint16_t)(txq->nb_tx_free - nb_pkts);
-
- n = (uint16_t)(txq->nb_tx_desc - tx_id);
- if (nb_commit >= n) {
- ci_tx_backlog_entry_vec(txep, tx_pkts, n);
-
- vtx(txdp, tx_pkts, n - 1, flags);
- tx_pkts += (n - 1);
- txdp += (n - 1);
-
- vtx1(txdp, *tx_pkts++, rs);
-
- nb_commit = (uint16_t)(nb_commit - n);
-
- tx_id = 0;
- txq->tx_next_rs = (uint16_t)(txq->tx_rs_thresh - 1);
-
- /* avoid reach the end of ring */
- txdp = txq->ci_tx_ring;
- txep = (void *)txq->sw_ring;
- }
-
- ci_tx_backlog_entry_vec(txep, tx_pkts, nb_commit);
-
- vtx(txdp, tx_pkts, nb_commit, flags);
-
- tx_id = (uint16_t)(tx_id + nb_commit);
- if (tx_id > txq->tx_next_rs) {
- txq->ci_tx_ring[txq->tx_next_rs].cmd_type_offset_bsz |=
- rte_cpu_to_le_64(((uint64_t)CI_TX_DESC_CMD_RS) << CI_TXD_QW1_CMD_S);
- txq->tx_next_rs =
- (uint16_t)(txq->tx_next_rs + txq->tx_rs_thresh);
- }
-
- txq->tx_tail = tx_id;
-
- I40E_PCI_REG_WC_WRITE(txq->qtx_tail, txq->tx_tail);
-
- return nb_pkts;
-}
-
uint16_t
i40e_xmit_pkts_vec_avx512(void *tx_queue, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
@@ -869,8 +762,8 @@ i40e_xmit_pkts_vec_avx512(void *tx_queue, struct rte_mbuf **tx_pkts,
/* cross rs_thresh boundary is not allowed */
num = (uint16_t)RTE_MIN(nb_pkts, txq->tx_rs_thresh);
- ret = i40e_xmit_fixed_burst_vec_avx512
- (tx_queue, &tx_pkts[nb_tx], num);
+ ret = ci_xmit_fixed_burst_vec_avx512(tx_queue, &tx_pkts[nb_tx], num,
+ false, CI_TAG_IN_DATA_DESC, CI_TAG_IN_DATA_DESC);
nb_tx += ret;
nb_pkts -= ret;
if (ret < num)
--
2.53.0
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH 0/2] net/i40e: use common AXV vector code
2026-09-23 13:51 [PATCH 0/2] net/i40e: use common AXV vector code Bruce Richardson
2026-09-23 13:51 ` [PATCH 1/2] net/i40e: use common AVX2 Tx function Bruce Richardson
2026-09-23 13:51 ` [PATCH 2/2] net/i40e: use common AVX-512 " Bruce Richardson
@ 2026-09-25 14:33 ` Bruce Richardson
2 siblings, 0 replies; 6+ messages in thread
From: Bruce Richardson @ 2026-09-25 14:33 UTC (permalink / raw)
To: dev
On Wed, Sep 23, 2026 at 02:51:44PM +0100, Bruce Richardson wrote:
> With the AVX2 and AVX-512 code paths from iavf and ice consolidated
> into common folder, we can reuse that code from i40e and reduce our
> code footprint a bit.
>
> Bruce Richardson (2):
> net/i40e: use common AVX2 Tx function
> net/i40e: use common AVX-512 Tx function
>
> drivers/net/intel/i40e/i40e_rxtx_vec_avx2.c | 121 +-----------------
> drivers/net/intel/i40e/i40e_rxtx_vec_avx512.c | 113 +---------------
> 2 files changed, 6 insertions(+), 228 deletions(-)
>
Patches applied to dpdk-next-net-intel
/Bruce
^ permalink raw reply [flat|nested] 6+ messages in thread