examples/l3fwd: add NEON implementation

[dpdk.git] / examples / l3fwd / l3fwd_em_hlm.h

diff --git a/examples/l3fwd/l3fwd_em_hlm.h b/examples/l3fwd/l3fwd_em_hlm.h
index 9fb5ff6..6fa81c4 100644 (file)
--- a/examples/l3fwd/l3fwd_em_hlm.h
+++ b/examples/l3fwd/l3fwd_em_hlm.h
@@ -35,8 +35,13 @@
 #ifndef __L3FWD_EM_HLM_H__
 #define __L3FWD_EM_HLM_H__
 
+#if defined RTE_ARCH_X86
 #include "l3fwd_sse.h"
 #include "l3fwd_em_hlm_sse.h"
+#elif defined RTE_MACHINE_CPUFLAG_NEON
+#include "l3fwd_neon.h"
+#include "l3fwd_em_hlm_neon.h"
+#endif
 
 static __rte_always_inline void
 em_get_dst_port_ipv4x8(struct lcore_conf *qconf, struct rte_mbuf *m[8],
@@ -238,7 +243,7 @@ static inline void
 l3fwd_em_send_packets(int nb_rx, struct rte_mbuf **pkts_burst,
                uint8_t portid, struct lcore_conf *qconf)
 {
-       int32_t j;
+       int32_t i, j, pos;
        uint16_t dst_port[MAX_PKT_BURST];
 
        /*
@@ -247,6 +252,11 @@ l3fwd_em_send_packets(int nb_rx, struct rte_mbuf **pkts_burst,
         */
        int32_t n = RTE_ALIGN_FLOOR(nb_rx, 8);
 
+       for (j = 0; j < 8 && j < nb_rx; j++) {
+               rte_prefetch0(rte_pktmbuf_mtod(pkts_burst[j],
+                                              struct ether_hdr *) + 1);
+       }
+
        for (j = 0; j < n; j += 8) {
 
                uint32_t pkt_type =
@@ -263,6 +273,11 @@ l3fwd_em_send_packets(int nb_rx, struct rte_mbuf **pkts_burst,
                uint32_t tcp_or_udp = pkt_type &
                        (RTE_PTYPE_L4_TCP | RTE_PTYPE_L4_UDP);
 
+               for (i = 0, pos = j + 8; i < 8 && pos < nb_rx; i++, pos++) {
+                       rte_prefetch0(rte_pktmbuf_mtod(pkts_burst[pos],
+                                                      struct ether_hdr *) + 1);
+               }
+
                if (tcp_or_udp && (l3_type == RTE_PTYPE_L3_IPV4)) {
 
                        em_get_dst_port_ipv4x8(qconf, &pkts_burst[j], portid,