All of lore.kernel.org
 help / color / mirror / Atom feed
From: Luiz Augusto von Dentz <luiz.dentz@gmail.com>
To: linux-bluetooth@vger.kernel.org
Subject: [PATCH SBC 2/3] sbc/sbc_primitives_sse: Optimize sbc_analyze_4s
Date: Tue, 11 Aug 2020 11:16:22 -0700	[thread overview]
Message-ID: <20200811181623.3683374-2-luiz.dentz@gmail.com> (raw)
In-Reply-To: <20200811181623.3683374-1-luiz.dentz@gmail.com>

From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>

This makes use 128 bit XMM registers whenever possible.

$ time src/sbcenc_mmx -s 4 sin_4m.au > /dev/null
real    0m1.073s
user    0m1.039s
sys     0m0.030s

===  After ====

$ time src/sbcenc -s 4 sin_4m.au > /dev/null
real    0m1.049s
user    0m1.000s
sys     0m0.047s
---
 sbc/sbc_primitives_sse.c | 58 +++++++++++++++++-----------------------
 1 file changed, 25 insertions(+), 33 deletions(-)

diff --git a/sbc/sbc_primitives_sse.c b/sbc/sbc_primitives_sse.c
index c2b729a..6471bd5 100644
--- a/sbc/sbc_primitives_sse.c
+++ b/sbc/sbc_primitives_sse.c
@@ -38,48 +38,40 @@
 static inline void sbc_analyze_four_sse(const int16_t *in, int32_t *out,
 					const FIXED_T *consts)
 {
-	static const SBC_ALIGNED int32_t round_c[2] = {
+	static const SBC_ALIGNED int32_t round_c[4] = {
+		1 << (SBC_PROTO_FIXED4_SCALE - 1),
+		1 << (SBC_PROTO_FIXED4_SCALE - 1),
 		1 << (SBC_PROTO_FIXED4_SCALE - 1),
 		1 << (SBC_PROTO_FIXED4_SCALE - 1),
 	};
 	__asm__ volatile (
-		"movq        (%0), %%mm0\n"
-		"movq       8(%0), %%mm1\n"
-		"pmaddwd     (%1), %%mm0\n"
-		"pmaddwd    8(%1), %%mm1\n"
-		"paddd       (%2), %%mm0\n"
-		"paddd       (%2), %%mm1\n"
+		"movdqu      (%0), %%xmm0\n"
+		"pmaddwd     (%1), %%xmm0\n"
+		"paddd       (%2), %%xmm0\n"
 		"\n"
-		"movq      16(%0), %%mm2\n"
-		"movq      24(%0), %%mm3\n"
-		"pmaddwd   16(%1), %%mm2\n"
-		"pmaddwd   24(%1), %%mm3\n"
-		"paddd      %%mm2, %%mm0\n"
-		"paddd      %%mm3, %%mm1\n"
+		"movdqu    16(%0), %%xmm1\n"
+		"pmaddwd   16(%1), %%xmm1\n"
+		"paddd     %%xmm1, %%xmm0\n"
 		"\n"
-		"movq      32(%0), %%mm2\n"
-		"movq      40(%0), %%mm3\n"
-		"pmaddwd   32(%1), %%mm2\n"
-		"pmaddwd   40(%1), %%mm3\n"
-		"paddd      %%mm2, %%mm0\n"
-		"paddd      %%mm3, %%mm1\n"
+		"movdqu    32(%0), %%xmm1\n"
+		"pmaddwd   32(%1), %%xmm1\n"
+		"paddd     %%xmm1, %%xmm0\n"
 		"\n"
-		"movq      48(%0), %%mm2\n"
-		"movq      56(%0), %%mm3\n"
-		"pmaddwd   48(%1), %%mm2\n"
-		"pmaddwd   56(%1), %%mm3\n"
-		"paddd      %%mm2, %%mm0\n"
-		"paddd      %%mm3, %%mm1\n"
+		"movdqu    48(%0), %%xmm1\n"
+		"pmaddwd   48(%1), %%xmm1\n"
+		"paddd     %%xmm1, %%xmm0\n"
 		"\n"
-		"movq      64(%0), %%mm2\n"
-		"movq      72(%0), %%mm3\n"
-		"pmaddwd   64(%1), %%mm2\n"
-		"pmaddwd   72(%1), %%mm3\n"
-		"paddd      %%mm2, %%mm0\n"
-		"paddd      %%mm3, %%mm1\n"
+		"movdqu    64(%0), %%xmm1\n"
+		"pmaddwd   64(%1), %%xmm1\n"
+		"paddd     %%xmm1, %%xmm0\n"
+		"\n"
+		"psrad         %4, %%xmm0\n"
+		"\n"
+		"movdqa    %%xmm0, %%xmm1\n"
+		"punpckhqdq %%xmm1, %%xmm1\n"
+		"movdq2q   %%xmm0, %%mm0\n"
+		"movdq2q   %%xmm1, %%mm1\n"
 		"\n"
-		"psrad         %4, %%mm0\n"
-		"psrad         %4, %%mm1\n"
 		"packssdw   %%mm0, %%mm0\n"
 		"packssdw   %%mm1, %%mm1\n"
 		"\n"
-- 
2.26.2


  reply	other threads:[~2020-08-11 18:16 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-08-11 18:16 [PATCH SBC 1/3] sbc: Add initial code for SSE primitives Luiz Augusto von Dentz
2020-08-11 18:16 ` Luiz Augusto von Dentz [this message]
2020-08-11 18:16 ` [PATCH SBC 3/3] sbc/sbc_primitives_sse: Optimize sbc_analyze_8s Luiz Augusto von Dentz
2020-08-12 12:48 ` [PATCH SBC 1/3] sbc: Add initial code for SSE primitives Marcel Holtmann
2020-08-14 20:56   ` Luiz Augusto von Dentz

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200811181623.3683374-2-luiz.dentz@gmail.com \
    --to=luiz.dentz@gmail.com \
    --cc=linux-bluetooth@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.