From: Richard Henderson <rth@twiddle.net>
To: qemu-devel@nongnu.org
Cc: peter.maydell@linaro.org, vijay.kilari@gmail.com,
qemu-arm@nongnu.org, qemu-ppc@nongnu.org, pbonzini@redhat.com,
David Gibson <david@gibson.dropbear.id.au>
Subject: [Qemu-arm] [PATCH 7/7] cutils: Rewrite ppc buffer zero checking
Date: Tue, 23 Aug 2016 21:17:59 -0700 [thread overview]
Message-ID: <1472012279-20581-8-git-send-email-rth@twiddle.net> (raw)
In-Reply-To: <1472012279-20581-1-git-send-email-rth@twiddle.net>
GCC versions through 6 do a poor job with the indexed addressing,
and (for ppc64le) issues unnecessary xxswapd insns.
Cc: qemu-ppc@nongnu.org
Cc: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Richard Henderson <rth@twiddle.net>
---
util/cutils.c | 41 +++++++++++++++++++++++++++++++++++++++--
1 file changed, 39 insertions(+), 2 deletions(-)
diff --git a/util/cutils.c b/util/cutils.c
index fe860e8..30fac02 100644
--- a/util/cutils.c
+++ b/util/cutils.c
@@ -254,8 +254,45 @@ static bool select_accel_int(const void *buf, size_t len)
#undef pixel
#undef bool
#define bool _Bool
-#define DO_ZERO(X) vec_all_eq(X, (__vector unsigned char){ 0 })
-ACCEL_BUFFER_ZERO(buffer_zero_ppc, 128, __vector unsigned char, DO_ZERO)
+
+static bool __attribute__((noinline))
+buffer_zero_ppc(const void *buf, size_t len)
+{
+ typedef unsigned char vec __attribute__((vector_size(16)));
+ const vec *p = buf;
+ const vec *end = buf + len;
+ vec t0, t1, t2, t3, zero = (vec){ 0 };
+
+ do {
+ p += 8;
+ __builtin_prefetch(p);
+ /* ??? GCC6 does poorly with power64le; extra xxswap. */
+ __asm volatile("lvebx %0,%4,%5\n\t"
+ "lvebx %1,%4,%6\n\t"
+ "lvebx %2,%4,%7\n\t"
+ "lvebx %3,%4,%8\n\t"
+ "vor %0,%0,%1\n\t"
+ "vor %1,%2,%3\n\t"
+ "lvebx %2,%4,%9\n\t"
+ "lvebx %3,%4,%10\n\t"
+ "vor %0,%0,%1\n\t"
+ "vor %1,%2,%3\n\t"
+ "lvebx %2,%4,%11\n\t"
+ "lvebx %3,%4,%12\n\t"
+ "vor %0,%0,%1\n\t"
+ "vor %1,%2,%3\n\t"
+ "vor %0,%0,%1"
+ : "=v"(t0), "=v"(t1), "=v"(t2), "=v"(t3)
+ : "b"(p), "b"(-8 * 16), "b"(-7 * 16),
+ "b"(-6 * 16), "b"(-5 * 16),
+ "b"(-4 * 16), "b"(-3 * 16),
+ "b"(-2 * 16), "b"(-1 * 16));
+ if (unlikely(vec_any_ne(t0, zero))) {
+ return false;
+ }
+ } while (p < end);
+ return true;
+}
static bool select_accel_fn(const void *buf, size_t len)
{
--
2.7.4
WARNING: multiple messages have this Message-ID (diff)
From: Richard Henderson <rth@twiddle.net>
To: qemu-devel@nongnu.org
Cc: vijay.kilari@gmail.com, qemu-arm@nongnu.org, pbonzini@redhat.com,
peter.maydell@linaro.org, qemu-ppc@nongnu.org,
David Gibson <david@gibson.dropbear.id.au>
Subject: [Qemu-devel] [PATCH 7/7] cutils: Rewrite ppc buffer zero checking
Date: Tue, 23 Aug 2016 21:17:59 -0700 [thread overview]
Message-ID: <1472012279-20581-8-git-send-email-rth@twiddle.net> (raw)
In-Reply-To: <1472012279-20581-1-git-send-email-rth@twiddle.net>
GCC versions through 6 do a poor job with the indexed addressing,
and (for ppc64le) issues unnecessary xxswapd insns.
Cc: qemu-ppc@nongnu.org
Cc: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Richard Henderson <rth@twiddle.net>
---
util/cutils.c | 41 +++++++++++++++++++++++++++++++++++++++--
1 file changed, 39 insertions(+), 2 deletions(-)
diff --git a/util/cutils.c b/util/cutils.c
index fe860e8..30fac02 100644
--- a/util/cutils.c
+++ b/util/cutils.c
@@ -254,8 +254,45 @@ static bool select_accel_int(const void *buf, size_t len)
#undef pixel
#undef bool
#define bool _Bool
-#define DO_ZERO(X) vec_all_eq(X, (__vector unsigned char){ 0 })
-ACCEL_BUFFER_ZERO(buffer_zero_ppc, 128, __vector unsigned char, DO_ZERO)
+
+static bool __attribute__((noinline))
+buffer_zero_ppc(const void *buf, size_t len)
+{
+ typedef unsigned char vec __attribute__((vector_size(16)));
+ const vec *p = buf;
+ const vec *end = buf + len;
+ vec t0, t1, t2, t3, zero = (vec){ 0 };
+
+ do {
+ p += 8;
+ __builtin_prefetch(p);
+ /* ??? GCC6 does poorly with power64le; extra xxswap. */
+ __asm volatile("lvebx %0,%4,%5\n\t"
+ "lvebx %1,%4,%6\n\t"
+ "lvebx %2,%4,%7\n\t"
+ "lvebx %3,%4,%8\n\t"
+ "vor %0,%0,%1\n\t"
+ "vor %1,%2,%3\n\t"
+ "lvebx %2,%4,%9\n\t"
+ "lvebx %3,%4,%10\n\t"
+ "vor %0,%0,%1\n\t"
+ "vor %1,%2,%3\n\t"
+ "lvebx %2,%4,%11\n\t"
+ "lvebx %3,%4,%12\n\t"
+ "vor %0,%0,%1\n\t"
+ "vor %1,%2,%3\n\t"
+ "vor %0,%0,%1"
+ : "=v"(t0), "=v"(t1), "=v"(t2), "=v"(t3)
+ : "b"(p), "b"(-8 * 16), "b"(-7 * 16),
+ "b"(-6 * 16), "b"(-5 * 16),
+ "b"(-4 * 16), "b"(-3 * 16),
+ "b"(-2 * 16), "b"(-1 * 16));
+ if (unlikely(vec_any_ne(t0, zero))) {
+ return false;
+ }
+ } while (p < end);
+ return true;
+}
static bool select_accel_fn(const void *buf, size_t len)
{
--
2.7.4
next prev parent reply other threads:[~2016-08-24 4:20 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-08-24 4:17 [Qemu-arm] [PATCH 0/7] Improve buffer_is_zero Richard Henderson
2016-08-24 4:17 ` [Qemu-devel] " Richard Henderson
2016-08-24 4:17 ` [Qemu-devel] [PATCH 1/7] cutils: Remove SPLAT macro Richard Henderson
2016-08-24 4:17 ` Richard Henderson
2016-08-24 4:17 ` [Qemu-devel] [PATCH 2/7] cutils: Export only buffer_is_zero Richard Henderson
2016-08-24 4:17 ` Richard Henderson
2016-08-24 8:37 ` [Qemu-arm] " Dr. David Alan Gilbert
2016-08-24 8:37 ` Dr. David Alan Gilbert
2016-08-24 4:17 ` [Qemu-devel] [PATCH 3/7] cutils: Rearrange buffer_is_zero acceleration Richard Henderson
2016-08-24 4:17 ` Richard Henderson
2016-08-24 4:17 ` [Qemu-devel] [PATCH 4/7] cutils: Add generic prefetch Richard Henderson
2016-08-24 4:17 ` Richard Henderson
2016-08-24 4:17 ` [Qemu-arm] [PATCH 5/7] cutils: Rewrite x86 buffer zero checking Richard Henderson
2016-08-24 4:17 ` [Qemu-devel] " Richard Henderson
2016-08-24 4:17 ` [Qemu-arm] [PATCH 6/7] cutils: Rewrite aarch64 " Richard Henderson
2016-08-24 4:17 ` [Qemu-devel] " Richard Henderson
2016-08-24 4:17 ` Richard Henderson [this message]
2016-08-24 4:17 ` [Qemu-devel] [PATCH 7/7] cutils: Rewrite ppc " Richard Henderson
2016-08-24 4:30 ` [Qemu-arm] [Qemu-devel] [PATCH 0/7] Improve buffer_is_zero no-reply
2016-08-24 4:30 ` no-reply
2016-08-24 4:38 ` [Qemu-arm] " Paolo Bonzini
2016-08-24 4:38 ` [Qemu-devel] " Paolo Bonzini
2016-08-24 14:53 ` [Qemu-arm] " Richard Henderson
2016-08-24 14:53 ` Richard Henderson
2016-08-24 14:59 ` [Qemu-arm] " Paolo Bonzini
2016-08-24 14:59 ` Paolo Bonzini
2016-08-24 8:34 ` [Qemu-arm] " Dr. David Alan Gilbert
2016-08-24 8:34 ` Dr. David Alan Gilbert
2016-08-24 10:26 ` Adam Richter
2016-08-24 10:26 ` Adam Richter
2016-08-24 10:52 ` [Qemu-arm] " Peter Maydell
2016-08-24 10:52 ` Peter Maydell
2016-08-24 11:45 ` [Qemu-arm] " Paolo Bonzini
2016-08-24 11:45 ` Paolo Bonzini
2016-08-24 12:22 ` [Qemu-arm] " Peter Maydell
2016-08-24 12:22 ` Peter Maydell
2016-08-25 6:37 ` [Qemu-arm] " Vijay Kilari
2016-08-25 6:37 ` [Qemu-devel] " Vijay Kilari
2016-08-25 8:04 ` [Qemu-arm] " Vijay Kilari
2016-08-25 8:04 ` [Qemu-devel] " Vijay Kilari
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=1472012279-20581-8-git-send-email-rth@twiddle.net \
--to=rth@twiddle.net \
--cc=david@gibson.dropbear.id.au \
--cc=pbonzini@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=qemu-arm@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@nongnu.org \
--cc=vijay.kilari@gmail.com \
/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.