소스 검색

x64: cpu_wait: Make use of MWAITX in MicroSleep

MWAITX is equivalent to UMWAIT on Intel's Alder Lake CPUs.
We can emulate TPAUSE by using MONITORX in conjunction with MWAITX to wait for 100K cycles.
Morph 3 년 전
부모
커밋
3d868baaa4
1개의 변경된 파일21개의 추가작업 그리고 12개의 파일을 삭제
  1. 21 12
      src/common/x64/cpu_wait.cpp

+ 21 - 12
src/common/x64/cpu_wait.cpp

@@ -13,24 +13,30 @@
 
 namespace Common::X64 {
 
+namespace {
+
+// 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
+constexpr auto PauseCycles = 100'000U;
+
+} // Anonymous namespace
+
 #ifdef _MSC_VER
 __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);
 }
+
+__forceinline static void MWAITX() {
+    // monitor_var should be aligned to a cache line.
+    alignas(64) u64 monitor_var{};
+    _mm_monitorx(&monitor_var, 0, 0);
+    _mm_mwaitx(/* extensions*/ 2, /* hints */ 0, /* cycles */ PauseCycles);
+}
 #else
 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);
@@ -40,9 +46,12 @@ static void TPAUSE() {
 
 void MicroSleep() {
     static const bool has_waitpkg = GetCPUCaps().waitpkg;
+    static const bool has_monitorx = GetCPUCaps().monitorx;
 
     if (has_waitpkg) {
         TPAUSE();
+    } else if (has_monitorx) {
+        MWAITX();
     } else {
         std::this_thread::yield();
     }