Skip to content
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

Zero register optimization for AVX-512-VBMI #14241

Merged
merged 2 commits into from
Aug 28, 2023
Merged

Zero register optimization for AVX-512-VBMI #14241

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
eabb2e7
Zero register optimization for AVX-512-VBMI
Whatcookie Jul 21, 2023
4796d54
Merge branch 'master' into shufb_utopia
elad335 Aug 6, 2023
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
50 changes: 50 additions & 0 deletions rpcs3/Emu/Cell/SPURecompiler.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8638,6 +8638,20 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
{
if (data == v128::from8p(data._u8[0]))
{
if (m_use_avx512_icl)
{
if (perm_only)
{
set_vr(op.rt4, vperm2b256to128(as, b, c));
return;
}

const auto m = gf2p8affineqb(c, build<u8[16]>(0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20), 0x7f);
const auto mm = select(noncast<s8[16]>(m) >= 0, splat<u8[16]>(0), m);
const auto ab = vperm2b256to128(as, b, c);
set_vr(op.rt4, select(noncast<s8[16]>(c) >= 0, ab, mm));
return;
}
// See above
const auto x = pshufb(build<u8[16]>(0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0x80, 0x80), (c >> 4));
const auto ax = pshufb(as, c);
Expand Down Expand Up @@ -8672,6 +8686,42 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator

if (m_use_avx512_icl && (op.ra != op.rb || m_interp_magn))
{
if (auto [ok, data] = get_const_vector(b.value, m_pos); ok)
{
if (data == v128::from8p(data._u8[0]))
{
if (perm_only)
{
set_vr(op.rt4, vperm2b256to128(a, b, eval(c ^ 0xf)));
return;
}

const auto m = gf2p8affineqb(c, build<u8[16]>(0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20), 0x7f);
const auto mm = select(noncast<s8[16]>(m) >= 0, splat<u8[16]>(0), m);
const auto ab = vperm2b256to128(a, b, eval(c ^ 0xf));
set_vr(op.rt4, select(noncast<s8[16]>(c) >= 0, ab, mm));
return;
}
}

if (auto [ok, data] = get_const_vector(a.value, m_pos); ok)
{
if (data == v128::from8p(data._u8[0]))
{
if (perm_only)
{
set_vr(op.rt4, vperm2b256to128(b, a, eval(c ^ 0x1f)));
return;
}

const auto m = gf2p8affineqb(c, build<u8[16]>(0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20), 0x7f);
const auto mm = select(noncast<s8[16]>(m) >= 0, splat<u8[16]>(0), m);
const auto ab = vperm2b256to128(b, a, eval(c ^ 0x1f));
set_vr(op.rt4, select(noncast<s8[16]>(c) >= 0, ab, mm));
return;
}
}

if (perm_only)
{
set_vr(op.rt4, vperm2b(a, b, eval(c ^ 0xf)));
Expand Down