[Qemu-devel] [PATCH v4 04/10] target/mips: Unroll loops in helpers for MSA logic instructions

[Aleksandar Markovic <aleksandar.markovic@rt-rk.com>]

From: Aleksandar Markovic <amarkovic@wavecomp.com>

Unroll loops in helpers for MSA logic instructions for better
performance.

Signed-off-by: Aleksandar Markovic <amarkovic@wavecomp.com>
Reviewed-by: Aleksandar Rikalo <arikalo@wavecomp.com>
Message-Id: <1559745316-1454-4-git-send-email-aleksandar.markovic@rt-rk.com>
---
 target/mips/msa_helper.c | 44 ++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 40 insertions(+), 4 deletions(-)

diff --git a/target/mips/msa_helper.c b/target/mips/msa_helper.c
index 7b73e22..a861155 100644
--- a/target/mips/msa_helper.c
+++ b/target/mips/msa_helper.c
@@ -268,10 +268,6 @@ void helper_msa_ ## FUNC(CPUMIPSState *env, uint32_t wd, uint32_t ws,   \
     }                                                                   \
 }
 
-MSA_FN_VECTOR(and_v, pwd->d[i], pws->d[i] & pwt->d[i])
-MSA_FN_VECTOR(or_v, pwd->d[i], pws->d[i] | pwt->d[i])
-MSA_FN_VECTOR(nor_v, pwd->d[i], ~(pws->d[i] | pwt->d[i]))
-MSA_FN_VECTOR(xor_v, pwd->d[i], pws->d[i] ^ pwt->d[i])
 MSA_FN_VECTOR(bmnz_v, pwd->d[i],
         BIT_MOVE_IF_NOT_ZERO(pwd->d[i], pws->d[i], pwt->d[i], DF_DOUBLE))
 MSA_FN_VECTOR(bmz_v, pwd->d[i],
@@ -283,6 +279,46 @@ MSA_FN_VECTOR(bsel_v, pwd->d[i],
 #undef BIT_SELECT
 #undef MSA_FN_VECTOR
 
+void helper_msa_and_v(CPUMIPSState *env, uint32_t wd, uint32_t ws, uint32_t wt)
+{
+    wr_t *pwd = &(env->active_fpu.fpr[wd].wr);
+    wr_t *pws = &(env->active_fpu.fpr[ws].wr);
+    wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
+
+    pwd->d[0] = pws->d[0] & pwt->d[0];
+    pwd->d[1] = pws->d[1] & pwt->d[1];
+}
+
+void helper_msa_or_v(CPUMIPSState *env, uint32_t wd, uint32_t ws, uint32_t wt)
+{
+    wr_t *pwd = &(env->active_fpu.fpr[wd].wr);
+    wr_t *pws = &(env->active_fpu.fpr[ws].wr);
+    wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
+
+    pwd->d[0] = pws->d[0] | pwt->d[0];
+    pwd->d[1] = pws->d[1] | pwt->d[1];
+}
+
+void helper_msa_nor_v(CPUMIPSState *env, uint32_t wd, uint32_t ws, uint32_t wt)
+{
+    wr_t *pwd = &(env->active_fpu.fpr[wd].wr);
+    wr_t *pws = &(env->active_fpu.fpr[ws].wr);
+    wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
+
+    pwd->d[0] = ~(pws->d[0] | pwt->d[0]);
+    pwd->d[1] = ~(pws->d[1] | pwt->d[1]);
+}
+
+void helper_msa_xor_v(CPUMIPSState *env, uint32_t wd, uint32_t ws, uint32_t wt)
+{
+    wr_t *pwd = &(env->active_fpu.fpr[wd].wr);
+    wr_t *pws = &(env->active_fpu.fpr[ws].wr);
+    wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
+
+    pwd->d[0] = pws->d[0] ^ pwt->d[0];
+    pwd->d[1] = pws->d[1] ^ pwt->d[1];
+}
+
 static inline int64_t msa_addv_df(uint32_t df, int64_t arg1, int64_t arg2)
 {
     return arg1 + arg2;
-- 
2.7.4