From 27c33ab73fd03d659654c49967a081214daf6ac2 Mon Sep 17 00:00:00 2001
From: Morph <39850852+Morph1984@users.noreply.github.com>
Date: Tue, 21 Mar 2023 21:28:38 -0400
Subject: x64: Add MicroSleep

MicroSleep allows the processor to pause for a "short" amount of time (in the microsecond range). This is useful for spin-waiting that does not require nanosecond precision.
This uses the new TPAUSE instruction introduced on Intel's newest processors as part of the waitpkg instructions. For CPUs that do not support waitpkg instructions, this is equivalent to yield().

Co-Authored-By: liamwhite <liamwhite@users.noreply.github.com>
---
 src/common/x64/cpu_wait.cpp | 72 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 72 insertions(+)
 create mode 100644 src/common/x64/cpu_wait.cpp

(limited to 'src/common/x64/cpu_wait.cpp')

diff --git a/src/common/x64/cpu_wait.cpp b/src/common/x64/cpu_wait.cpp
new file mode 100644
index 000000000..1fab0bfe8
--- /dev/null
+++ b/src/common/x64/cpu_wait.cpp
@@ -0,0 +1,72 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <thread>
+
+#ifdef _MSC_VER
+#include <intrin.h>
+#endif
+
+#include "common/x64/cpu_detect.h"
+#include "common/x64/cpu_wait.h"
+
+namespace Common::X64 {
+
+#ifdef _MSC_VER
+__forceinline static u64 FencedRDTSC() {
+    _mm_lfence();
+    _ReadWriteBarrier();
+    const u64 result = __rdtsc();
+    _mm_lfence();
+    _ReadWriteBarrier();
+    return result;
+}
+
+__forceinline static void TPAUSE() {
+    // 100,000 cycles is a reasonable amount of time to wait to save on CPU resources.
+    // For reference:
+    // At 1 GHz, 100K cycles is 100us
+    // At 2 GHz, 100K cycles is 50us
+    // At 4 GHz, 100K cycles is 25us
+    static constexpr auto PauseCycles = 100'000;
+    _tpause(0, FencedRDTSC() + PauseCycles);
+}
+#else
+static u64 FencedRDTSC() {
+    u64 result;
+    asm volatile("lfence\n\t"
+                 "rdtsc\n\t"
+                 "shl $32, %%rdx\n\t"
+                 "or %%rdx, %0\n\t"
+                 "lfence"
+                 : "=a"(result)
+                 :
+                 : "rdx", "memory", "cc");
+    return result;
+}
+
+static void TPAUSE() {
+    // 100,000 cycles is a reasonable amount of time to wait to save on CPU resources.
+    // For reference:
+    // At 1 GHz, 100K cycles is 100us
+    // At 2 GHz, 100K cycles is 50us
+    // At 4 GHz, 100K cycles is 25us
+    static constexpr auto PauseCycles = 100'000;
+    const auto tsc = FencedRDTSC() + PauseCycles;
+    const auto eax = static_cast<u32>(tsc & 0xFFFFFFFF);
+    const auto edx = static_cast<u32>(tsc >> 32);
+    asm volatile("tpause %0" : : "r"(0), "d"(edx), "a"(eax));
+}
+#endif
+
+void MicroSleep() {
+    static const bool has_waitpkg = GetCPUCaps().waitpkg;
+
+    if (has_waitpkg) {
+        TPAUSE();
+    } else {
+        std::this_thread::yield();
+    }
+}
+
+} // namespace Common::X64
-- 
cgit v1.2.3


From 981bc8aa1c924eabc55f63f9671324c548c72d59 Mon Sep 17 00:00:00 2001
From: Morph <39850852+Morph1984@users.noreply.github.com>
Date: Wed, 22 Mar 2023 16:40:41 -0400
Subject: x64: Simplify RDTSC on non-MSVC compilers

Co-Authored-By: liamwhite <liamwhite@users.noreply.github.com>
---
 src/common/x64/cpu_wait.cpp | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)

(limited to 'src/common/x64/cpu_wait.cpp')

diff --git a/src/common/x64/cpu_wait.cpp b/src/common/x64/cpu_wait.cpp
index 1fab0bfe8..cfeef6a3d 100644
--- a/src/common/x64/cpu_wait.cpp
+++ b/src/common/x64/cpu_wait.cpp
@@ -33,16 +33,13 @@ __forceinline static void TPAUSE() {
 }
 #else
 static u64 FencedRDTSC() {
-    u64 result;
+    u64 eax;
+    u64 edx;
     asm volatile("lfence\n\t"
                  "rdtsc\n\t"
-                 "shl $32, %%rdx\n\t"
-                 "or %%rdx, %0\n\t"
-                 "lfence"
-                 : "=a"(result)
-                 :
-                 : "rdx", "memory", "cc");
-    return result;
+                 "lfence\n\t"
+                 : "=a"(eax), "=d"(edx));
+    return (edx << 32) | eax;
 }
 
 static void TPAUSE() {
-- 
cgit v1.2.3