Skip to content

8367611: Enable vblendvp[sd] on Future ECore #27354

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 4 commits into
base: master
Choose a base branch
from
+10 −2
Open

8367611: Enable vblendvp[sd] on Future ECore #27354

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
cddc0a4
Use blend instruction when destination is same as source for certain …
missa-prime Sep 17, 2025
078a295
Add file missing from last commit
missa-prime Sep 18, 2025
88ac773
Use function to identify Darkmont cores instead of global flag
missa-prime Sep 18, 2025
5e5c027
Combine boolean flags to avoid ambiguity
missa-prime Sep 19, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 4 additions & 2 deletions src/hotspot/cpu/x86/macroAssembler_x86.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2897,7 +2897,8 @@ void MacroAssembler::vbroadcastss(XMMRegister dst, AddressLiteral src, int vecto
// vblendvps(XMMRegister dst, XMMRegister nds, XMMRegister src, XMMRegister mask, int vector_len, bool compute_mask = true, XMMRegister scratch = xnoreg)
void MacroAssembler::vblendvps(XMMRegister dst, XMMRegister src1, XMMRegister src2, XMMRegister mask, int vector_len, bool compute_mask, XMMRegister scratch) {
// WARN: Allow dst == (src1|src2), mask == scratch
bool blend_emulation = EnableX86ECoreOpts && UseAVX > 1;
bool blend_emulation = EnableX86ECoreOpts && UseAVX > 1 &&
!(VM_Version::is_intel_darkmont() && (dst == src1)); // partially fixed on Darkmont
bool scratch_available = scratch != xnoreg && scratch != src1 && scratch != src2 && scratch != dst;
bool dst_available = dst != mask && (dst != src1 || dst != src2);
if (blend_emulation && scratch_available && dst_available) {
Expand All @@ -2921,7 +2922,8 @@ void MacroAssembler::vblendvps(XMMRegister dst, XMMRegister src1, XMMRegister sr
// vblendvpd(XMMRegister dst, XMMRegister nds, XMMRegister src, XMMRegister mask, int vector_len, bool compute_mask = true, XMMRegister scratch = xnoreg)
void MacroAssembler::vblendvpd(XMMRegister dst, XMMRegister src1, XMMRegister src2, XMMRegister mask, int vector_len, bool compute_mask, XMMRegister scratch) {
// WARN: Allow dst == (src1|src2), mask == scratch
bool blend_emulation = EnableX86ECoreOpts && UseAVX > 1;
bool blend_emulation = EnableX86ECoreOpts && UseAVX > 1 &&
!(VM_Version::is_intel_darkmont() && (dst == src1)); // partially fixed on Darkmont
bool scratch_available = scratch != xnoreg && scratch != src1 && scratch != src2 && scratch != dst && (!compute_mask || scratch != mask);
bool dst_available = dst != mask && (dst != src1 || dst != src2);
if (blend_emulation && scratch_available && dst_available) {
Expand Down
4 changes: 4 additions & 0 deletions src/hotspot/cpu/x86/vm_version_x86.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2081,6 +2081,10 @@ bool VM_Version::is_intel_cascade_lake() {
return is_intel_skylake() && _stepping >= 5;
}

bool VM_Version::is_intel_darkmont() {
return is_intel() && is_intel_server_family() && (_model == 0xCC || _model == 0xDD);
}

// avx3_threshold() sets the threshold at which 64-byte instructions are used
// for implementing the array copy and clear operations.
// The Intel platforms that supports the serialize instruction
Expand Down
2 changes: 2 additions & 0 deletions src/hotspot/cpu/x86/vm_version_x86.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -923,6 +923,8 @@ class VM_Version : public Abstract_VM_Version {

static bool is_intel_cascade_lake();

static bool is_intel_darkmont();

static int avx3_threshold();

static bool is_intel_tsc_synched_at_init();
Expand Down