From f207239349aefe5dd93c4f797a0c18211ee6b53a Mon Sep 17 00:00:00 2001
From: Wunkolo <Wunkolo@gmail.com>
Date: Fri, 4 Feb 2022 22:51:17 -0800
Subject: [PATCH] [x64] Add `kX64EmitAVX512VBMI` feature-flag and detection

Allows access to byte-element 2-register permutations(32-byte look up
tables) and for 64-bit multi-shifts.
Particularly adding this to accelerate the assembly of our `PERMUTE`
opcode.
---
 src/xenia/cpu/backend/x64/x64_backend.cc | 1 +
 src/xenia/cpu/backend/x64/x64_emitter.cc | 1 +
 src/xenia/cpu/backend/x64/x64_emitter.h  | 2 ++
 3 files changed, 4 insertions(+)

diff --git a/src/xenia/cpu/backend/x64/x64_backend.cc b/src/xenia/cpu/backend/x64/x64_backend.cc
index 31e1dc9fd..5355d1470 100644
--- a/src/xenia/cpu/backend/x64/x64_backend.cc
+++ b/src/xenia/cpu/backend/x64/x64_backend.cc
@@ -42,6 +42,7 @@ DEFINE_int32(x64_extension_mask, -1,
              "  512 = AVX512VL\n"
              " 1024 = AVX512BW\n"
              " 2048 = AVX512DQ\n"
+             " 4096 = AVX512VBMI\n"
              "   -1 = Detect and utilize all possible processor features\n",
              "x64");
 
diff --git a/src/xenia/cpu/backend/x64/x64_emitter.cc b/src/xenia/cpu/backend/x64/x64_emitter.cc
index 97b14e03e..034946d03 100644
--- a/src/xenia/cpu/backend/x64/x64_emitter.cc
+++ b/src/xenia/cpu/backend/x64/x64_emitter.cc
@@ -98,6 +98,7 @@ X64Emitter::X64Emitter(X64Backend* backend, XbyakAllocator* allocator)
   TEST_EMIT_FEATURE(kX64EmitAVX512VL, Xbyak::util::Cpu::tAVX512VL);
   TEST_EMIT_FEATURE(kX64EmitAVX512BW, Xbyak::util::Cpu::tAVX512BW);
   TEST_EMIT_FEATURE(kX64EmitAVX512DQ, Xbyak::util::Cpu::tAVX512DQ);
+  TEST_EMIT_FEATURE(kX64EmitAVX512VBMI, Xbyak::util::Cpu::tAVX512_VBMI);
 
 #undef TEST_EMIT_FEATURE
 }
diff --git a/src/xenia/cpu/backend/x64/x64_emitter.h b/src/xenia/cpu/backend/x64/x64_emitter.h
index 247d6175c..3cf1df7d7 100644
--- a/src/xenia/cpu/backend/x64/x64_emitter.h
+++ b/src/xenia/cpu/backend/x64/x64_emitter.h
@@ -140,6 +140,8 @@ enum X64EmitterFeatureFlags {
   kX64EmitAVX512BW = 1 << 10,
   kX64EmitAVX512DQ = 1 << 11,
 
+  kX64EmitAVX512VBMI = 1 << 12,
+
   kX64EmitAVX512Ortho = kX64EmitAVX512F | kX64EmitAVX512VL,
   kX64EmitAVX512Ortho64 = kX64EmitAVX512Ortho | kX64EmitAVX512DQ
 };