[PATCH v3 2/2] Support Intel AMX-AVX512

Haochen Jiang haochen.jiang@intel.com
Thu Jan 9 06:00:52 GMT 2025


Note: The encoding issue on tcvtrowps2[bf16,ph][h,l] for Imm8
seems more and more unlikely to change. It is not a good design and 
could be treated as somehow a mistake (A real mistake would be the
insn get implemented wrongly with unexpected long latency, so I called
it somehow a mistake), but too late to change, which is really a shame.

---

Changes in v3:

  - Adjust the GPR mode handle in OP_VEX and combining the switch
    with the previous switch since they are all doing switch on
    bytemode.

---

This patch will support AMX-AVX512. In disassmbler, we pull out all
GPR mode out of the vex length switch to make it more general.

---

Changes in v2:

  - Pull out all GPR mode out of vex length switch in OP_VEX to make
    it more general.
  - Remove invalid test for 32-bit.
  - Reuse VexGdq for operands.
  - Update the mnemonics from tcvtrowps2pbf16[h,l] to tcvtrowps2bf16[h,l]
    according to ISE056.

---

gas/ChangeLog:

	* NEWS: Mention the full support on DMR AMX ISAs.
	* config/tc-i386.c: Add amx_avx512.
	* doc/c-i386.texi: Document .amx_avx512.
	* testsuite/gas/i386/x86-64.exp: Run AMX-AVX512 tests.
	* testsuite/gas/i386/x86-64-amx-avx512-intel.d: New test.
	* testsuite/gas/i386/x86-64-amx-avx512.d: Ditto.
	* testsuite/gas/i386/x86-64-amx-avx512.s: Ditto.

opcodes/ChangeLog:

	* i386-dis-evex-len.h: Add EVEX_LEN_0F384A_X86_64_W_0,
	EVEX_LEN_0F386D_X86_64_W_0, EVEX_LEN_0F3A07_X86_64_W_0,
	EVEX_LEN_0F3A77_X86_64_W_0.
	* i386-dis-evex-prefix.h: Add PREFIX_EVEX_0F384A_W_0_L_2,
	PREFIX_EVEX_0F386D_W_0_L_2, PREFIX_EVEX_0F3A07_W_0_L_2,
	PREFIX_EVEX_0F3A77_W_0_L_2.
	* i386-dis-evex-w.h: Add EVEX_W_0F384A_X86_64, EVEX_W_0F386D_X86_64,
	EVEX_W_0F3A07_X86_64, EVEX_W_0F3A77_X86_64.
	* i386-dis-evex-x86-64.h: Add X86_64_EVEX_0F384A, X86_64_EVEX_0F386D,
	X86_64_EVEX_0F3A07, X86_64_EVEX_0F3A77.
	* i386-dis-evex.h: Ditto.
	* i386-dis.c (EVEX_LEN_0F384A_X86_64_W_0): New.
	(EVEX_LEN_0F386D_X86_64_W_0): Ditto.
	(EVEX_LEN_0F3A07_X86_64_W_0): Ditto.
	(EVEX_LEN_0F3A77_X86_64_W_0): Ditto.
	(MOD_EVEX_0F384A_X86_64_W_0): Ditto.
	(MOD_EVEX_0F386D_X86_64_W_0): Ditto.
	(MOD_EVEX_0F3A07_X86_64_W_0): Ditto.
	(MOD_EVEX_0F3A77_X86_64_W_0): Ditto.
	(PREFIX_EVEX_0F384A_W_0_L_2): Ditto.
	(PREFIX_EVEX_0F386D_W_0_L_2): Ditto.
	(PREFIX_EVEX_0F3A07_W_0_L_2): Ditto.
	(PREFIX_EVEX_0F3A77_W_0_L_2): Ditto.
	(EVEX_W_0F384A_X86_64): Ditto.
	(EVEX_W_0F386D_X86_64): Ditto.
	(EVEX_W_0F3A07_X86_64): Ditto.
	(EVEX_W_0F3A77_X86_64): Ditto.
	(X86_64_EVEX_0F384A): Ditto.
	(X86_64_EVEX_0F386D): Ditto.
	(X86_64_EVEX_0F3A07): Ditto.
	(X86_64_EVEX_0F3A77): Ditto.
	(OP_VEX): Pull out all GPR mode out of the vector length switch.
	* i386-gen.c (isa_dependencies): Add AMX-AVX512.
	(cpu_flags): Ditto.
	* i386-init.h: Regenerated.
	* i386-mnem.h: Ditto.
	* i386-opc.h (CpuAMX_AVX512): New.
	(i386_cpu_flags): Add cpuamx_avx512.
	* i386-opc.tbl: Add AMX-AVX512 instructions.
	* i386-tbl.h: Regenerated.
---
 gas/NEWS                                      |    3 +
 gas/config/tc-i386.c                          |    1 +
 gas/doc/c-i386.texi                           |    4 +-
 .../gas/i386/x86-64-amx-avx512-intel.d        |   35 +
 gas/testsuite/gas/i386/x86-64-amx-avx512.d    |   34 +
 gas/testsuite/gas/i386/x86-64-amx-avx512.s    |   55 +
 gas/testsuite/gas/i386/x86-64.exp             |    2 +
 opcodes/i386-dis-evex-len.h                   |   23 +
 opcodes/i386-dis-evex-prefix.h                |   27 +
 opcodes/i386-dis-evex-w.h                     |   12 +
 opcodes/i386-dis-evex-x86-64.h                |   15 +
 opcodes/i386-dis-evex.h                       |    6 +-
 opcodes/i386-dis.c                            |   50 +-
 opcodes/i386-gen.c                            |    3 +
 opcodes/i386-init.h                           |  722 ++---
 opcodes/i386-mnem.h                           | 2570 +++++++++--------
 opcodes/i386-opc.h                            |    3 +
 opcodes/i386-opc.tbl                          |   15 +
 opcodes/i386-tbl.h                            |  409 ++-
 19 files changed, 2207 insertions(+), 1782 deletions(-)
 create mode 100644 gas/testsuite/gas/i386/x86-64-amx-avx512-intel.d
 create mode 100644 gas/testsuite/gas/i386/x86-64-amx-avx512.d
 create mode 100644 gas/testsuite/gas/i386/x86-64-amx-avx512.s

diff --git a/gas/NEWS b/gas/NEWS
index 2b31fdb1515..920fc40c7de 100644
--- a/gas/NEWS
+++ b/gas/NEWS
@@ -1,5 +1,8 @@
 -*- text -*-
 
+* Add support for the x86 Intel Dimaond Rapids AMX instructions, including
+AMX-AVX512, AMX-FP8, AMX-MOVRS, AMX-TF32 and AMX-TRANSPOSE.
+
 * Add support for the x86 Intel AVX10.2 instructions.
 
 * Add support for the x86 Intel SM4 AVX10.2 instructions.
diff --git a/gas/config/tc-i386.c b/gas/config/tc-i386.c
index 81597568443..d274625b0b9 100644
--- a/gas/config/tc-i386.c
+++ b/gas/config/tc-i386.c
@@ -1186,6 +1186,7 @@ static const arch_entry cpu_arch[] =
   SUBARCH (amx_tf32, AMX_TF32, ANY_AMX_TF32, false),
   SUBARCH (amx_fp8, AMX_FP8, ANY_AMX_FP8, false),
   SUBARCH (amx_movrs, AMX_MOVRS, ANY_AMX_MOVRS, false),
+  SUBARCH (amx_avx512, AMX_AVX512, ANY_AMX_AVX512, false),
   SUBARCH (amx_tile, AMX_TILE, ANY_AMX_TILE, false),
   SUBARCH (movdiri, MOVDIRI, MOVDIRI, false),
   SUBARCH (movdir64b, MOVDIR64B, MOVDIR64B, false),
diff --git a/gas/doc/c-i386.texi b/gas/doc/c-i386.texi
index ddf0393ec95..bffc381b96c 100644
--- a/gas/doc/c-i386.texi
+++ b/gas/doc/c-i386.texi
@@ -232,6 +232,7 @@ accept various extension mnemonics.  For example,
 @code{amx_tf32},
 @code{amx_fp8}
 @code{amx_movrs},
+@code{amx_avx512},
 @code{amx_tile},
 @code{vmx},
 @code{vmfunc},
@@ -1705,7 +1706,8 @@ supported on the CPU specified.  The choices for @var{cpu_type} are:
 @item @samp{.movdiri} @tab @samp{.movdir64b} @tab @samp{.enqcmd} @tab @samp{.tsxldtrk}
 @item @samp{.amx_int8} @tab @samp{.amx_bf16} @tab @samp{.amx_fp16}
 @item @samp{.amx_complex} @tab @samp{.amx_transpose} @tab @samp{.amx_tf32}
-@item @samp{.amx_fp8} @tab @samp{.amx_movrs} @tab @samp{.amx_tile}
+@item @samp{.amx_fp8} @tab @samp{.amx_movrs} @tab @samp{.amx_avx512}
+@item @samp{.amx_tile}
 @item @samp{.kl} @tab @samp{.widekl} @tab @samp{.uintr} @tab @samp{.hreset}
 @item @samp{.3dnow} @tab @samp{.3dnowa} @tab @samp{.sse4a} @tab @samp{.sse5}
 @item @samp{.syscall} @tab @samp{.rdtscp} @tab @samp{.svme}
diff --git a/gas/testsuite/gas/i386/x86-64-amx-avx512-intel.d b/gas/testsuite/gas/i386/x86-64-amx-avx512-intel.d
new file mode 100644
index 00000000000..33e6d01a558
--- /dev/null
+++ b/gas/testsuite/gas/i386/x86-64-amx-avx512-intel.d
@@ -0,0 +1,35 @@
+#objdump: -dw -Mintel
+#name: x86_64 AMX-AVX512 insns (Intel disassembly)
+#source: x86-64-amx-avx512.s
+
+.*: +file format .*
+
+Disassembly of section \.text:
+
+#...
+[a-f0-9]+ <_intel>:
+\s*[a-f0-9]+:\s*62 62 6e 48 4a f5\s+tcvtrowd2ps zmm30,tmm5,edx
+\s*[a-f0-9]+:\s*62 62 6e 48 4a f2\s+tcvtrowd2ps zmm30,tmm2,edx
+\s*[a-f0-9]+:\s*62 63 7e 48 07 f5 7b\s+tcvtrowd2ps zmm30,tmm5,0x7b
+\s*[a-f0-9]+:\s*62 63 7e 48 07 f2 7b\s+tcvtrowd2ps zmm30,tmm2,0x7b
+\s*[a-f0-9]+:\s*62 62 6f 48 6d f5\s+tcvtrowps2bf16h zmm30,tmm5,edx
+\s*[a-f0-9]+:\s*62 62 6f 48 6d f2\s+tcvtrowps2bf16h zmm30,tmm2,edx
+\s*[a-f0-9]+:\s*62 63 7f 48 07 f5 7b\s+tcvtrowps2bf16h zmm30,tmm5,0x7b
+\s*[a-f0-9]+:\s*62 63 7f 48 07 f2 7b\s+tcvtrowps2bf16h zmm30,tmm2,0x7b
+\s*[a-f0-9]+:\s*62 62 6e 48 6d f5\s+tcvtrowps2bf16l zmm30,tmm5,edx
+\s*[a-f0-9]+:\s*62 62 6e 48 6d f2\s+tcvtrowps2bf16l zmm30,tmm2,edx
+\s*[a-f0-9]+:\s*62 63 7e 48 77 f5 7b\s+tcvtrowps2bf16l zmm30,tmm5,0x7b
+\s*[a-f0-9]+:\s*62 63 7e 48 77 f2 7b\s+tcvtrowps2bf16l zmm30,tmm2,0x7b
+\s*[a-f0-9]+:\s*62 62 6c 48 6d f5\s+tcvtrowps2phh zmm30,tmm5,edx
+\s*[a-f0-9]+:\s*62 62 6c 48 6d f2\s+tcvtrowps2phh zmm30,tmm2,edx
+\s*[a-f0-9]+:\s*62 63 7c 48 07 f5 7b\s+tcvtrowps2phh zmm30,tmm5,0x7b
+\s*[a-f0-9]+:\s*62 63 7c 48 07 f2 7b\s+tcvtrowps2phh zmm30,tmm2,0x7b
+\s*[a-f0-9]+:\s*62 62 6d 48 6d f5\s+tcvtrowps2phl zmm30,tmm5,edx
+\s*[a-f0-9]+:\s*62 62 6d 48 6d f2\s+tcvtrowps2phl zmm30,tmm2,edx
+\s*[a-f0-9]+:\s*62 63 7f 48 77 f5 7b\s+tcvtrowps2phl zmm30,tmm5,0x7b
+\s*[a-f0-9]+:\s*62 63 7f 48 77 f2 7b\s+tcvtrowps2phl zmm30,tmm2,0x7b
+\s*[a-f0-9]+:\s*62 62 6d 48 4a f5\s+tilemovrow zmm30,tmm5,edx
+\s*[a-f0-9]+:\s*62 62 6d 48 4a f2\s+tilemovrow zmm30,tmm2,edx
+\s*[a-f0-9]+:\s*62 63 7d 48 07 f5 7b\s+tilemovrow zmm30,tmm5,0x7b
+\s*[a-f0-9]+:\s*62 63 7d 48 07 f2 7b\s+tilemovrow zmm30,tmm2,0x7b
+#pass
diff --git a/gas/testsuite/gas/i386/x86-64-amx-avx512.d b/gas/testsuite/gas/i386/x86-64-amx-avx512.d
new file mode 100644
index 00000000000..d2f8ac6e51e
--- /dev/null
+++ b/gas/testsuite/gas/i386/x86-64-amx-avx512.d
@@ -0,0 +1,34 @@
+#objdump: -dw
+#name: x86_64 AMX-AVX512 insns
+#source: x86-64-amx-avx512.s
+
+.*: +file format .*
+
+Disassembly of section \.text:
+
+0+ <_start>:
+\s*[a-f0-9]+:\s*62 62 6e 48 4a f5\s+tcvtrowd2ps %edx,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 62 6e 48 4a f2\s+tcvtrowd2ps %edx,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 63 7e 48 07 f5 7b\s+tcvtrowd2ps \$0x7b,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 63 7e 48 07 f2 7b\s+tcvtrowd2ps \$0x7b,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 62 6f 48 6d f5\s+tcvtrowps2bf16h %edx,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 62 6f 48 6d f2\s+tcvtrowps2bf16h %edx,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 63 7f 48 07 f5 7b\s+tcvtrowps2bf16h \$0x7b,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 63 7f 48 07 f2 7b\s+tcvtrowps2bf16h \$0x7b,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 62 6e 48 6d f5\s+tcvtrowps2bf16l %edx,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 62 6e 48 6d f2\s+tcvtrowps2bf16l %edx,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 63 7e 48 77 f5 7b\s+tcvtrowps2bf16l \$0x7b,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 63 7e 48 77 f2 7b\s+tcvtrowps2bf16l \$0x7b,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 62 6c 48 6d f5\s+tcvtrowps2phh %edx,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 62 6c 48 6d f2\s+tcvtrowps2phh %edx,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 63 7c 48 07 f5 7b\s+tcvtrowps2phh \$0x7b,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 63 7c 48 07 f2 7b\s+tcvtrowps2phh \$0x7b,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 62 6d 48 6d f5\s+tcvtrowps2phl %edx,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 62 6d 48 6d f2\s+tcvtrowps2phl %edx,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 63 7f 48 77 f5 7b\s+tcvtrowps2phl \$0x7b,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 63 7f 48 77 f2 7b\s+tcvtrowps2phl \$0x7b,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 62 6d 48 4a f5\s+tilemovrow %edx,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 62 6d 48 4a f2\s+tilemovrow %edx,%tmm2,%zmm30
+\s*[a-f0-9]+:\s*62 63 7d 48 07 f5 7b\s+tilemovrow \$0x7b,%tmm5,%zmm30
+\s*[a-f0-9]+:\s*62 63 7d 48 07 f2 7b\s+tilemovrow \$0x7b,%tmm2,%zmm30
+#pass
diff --git a/gas/testsuite/gas/i386/x86-64-amx-avx512.s b/gas/testsuite/gas/i386/x86-64-amx-avx512.s
new file mode 100644
index 00000000000..6df493430a0
--- /dev/null
+++ b/gas/testsuite/gas/i386/x86-64-amx-avx512.s
@@ -0,0 +1,55 @@
+# Check 64bit AMX-AVX512 instructions
+
+	.text
+_start:
+	tcvtrowd2ps	%edx, %tmm5, %zmm30
+	tcvtrowd2ps	%edx, %tmm2, %zmm30
+	tcvtrowd2ps	$123, %tmm5, %zmm30
+	tcvtrowd2ps	$123, %tmm2, %zmm30
+	tcvtrowps2bf16h	%edx, %tmm5, %zmm30
+	tcvtrowps2bf16h	%edx, %tmm2, %zmm30
+	tcvtrowps2bf16h	$123, %tmm5, %zmm30
+	tcvtrowps2bf16h	$123, %tmm2, %zmm30
+	tcvtrowps2bf16l	%edx, %tmm5, %zmm30
+	tcvtrowps2bf16l	%edx, %tmm2, %zmm30
+	tcvtrowps2bf16l	$123, %tmm5, %zmm30
+	tcvtrowps2bf16l	$123, %tmm2, %zmm30
+	tcvtrowps2phh	%edx, %tmm5, %zmm30
+	tcvtrowps2phh	%edx, %tmm2, %zmm30
+	tcvtrowps2phh	$123, %tmm5, %zmm30
+	tcvtrowps2phh	$123, %tmm2, %zmm30
+	tcvtrowps2phl	%edx, %tmm5, %zmm30
+	tcvtrowps2phl	%edx, %tmm2, %zmm30
+	tcvtrowps2phl	$123, %tmm5, %zmm30
+	tcvtrowps2phl	$123, %tmm2, %zmm30
+	tilemovrow	%edx, %tmm5, %zmm30
+	tilemovrow	%edx, %tmm2, %zmm30
+	tilemovrow	$123, %tmm5, %zmm30
+	tilemovrow	$123, %tmm2, %zmm30
+
+_intel:
+	.intel_syntax noprefix
+	tcvtrowd2ps	zmm30, tmm5, edx
+	tcvtrowd2ps	zmm30, tmm2, edx
+	tcvtrowd2ps	zmm30, tmm5, 123
+	tcvtrowd2ps	zmm30, tmm2, 123
+	tcvtrowps2bf16h	zmm30, tmm5, edx
+	tcvtrowps2bf16h	zmm30, tmm2, edx
+	tcvtrowps2bf16h	zmm30, tmm5, 123
+	tcvtrowps2bf16h	zmm30, tmm2, 123
+	tcvtrowps2bf16l	zmm30, tmm5, edx
+	tcvtrowps2bf16l	zmm30, tmm2, edx
+	tcvtrowps2bf16l	zmm30, tmm5, 123
+	tcvtrowps2bf16l	zmm30, tmm2, 123
+	tcvtrowps2phh	zmm30, tmm5, edx
+	tcvtrowps2phh	zmm30, tmm2, edx
+	tcvtrowps2phh	zmm30, tmm5, 123
+	tcvtrowps2phh	zmm30, tmm2, 123
+	tcvtrowps2phl	zmm30, tmm5, edx
+	tcvtrowps2phl	zmm30, tmm2, edx
+	tcvtrowps2phl	zmm30, tmm5, 123
+	tcvtrowps2phl	zmm30, tmm2, 123
+	tilemovrow	zmm30, tmm5, edx
+	tilemovrow	zmm30, tmm2, edx
+	tilemovrow	zmm30, tmm5, 123
+	tilemovrow	zmm30, tmm2, 123
diff --git a/gas/testsuite/gas/i386/x86-64.exp b/gas/testsuite/gas/i386/x86-64.exp
index c99488935f5..95c78f4ed7a 100644
--- a/gas/testsuite/gas/i386/x86-64.exp
+++ b/gas/testsuite/gas/i386/x86-64.exp
@@ -538,6 +538,8 @@ run_dump_test "x86-64-amx-fp8-bad"
 run_dump_test "x86-64-amx-movrs"
 run_dump_test "x86-64-amx-movrs-intel"
 run_list_test "x86-64-amx-movrs-inval"
+run_dump_test "x86-64-amx-avx512"
+run_dump_test "x86-64-amx-avx512-intel"
 run_dump_test "x86-64-clzero"
 run_dump_test "x86-64-mwaitx-bdver4"
 run_list_test "x86-64-mwaitx-reg"
diff --git a/opcodes/i386-dis-evex-len.h b/opcodes/i386-dis-evex-len.h
index 2b4361f7ae6..434e051bb63 100644
--- a/opcodes/i386-dis-evex-len.h
+++ b/opcodes/i386-dis-evex-len.h
@@ -47,6 +47,8 @@ static const struct dis386 evex_len_table[][3] = {
   /* EVEX_LEN_0F384A_X86_64_W_0 */
   {
     { X86_64_EVEX_PFX_TABLE (PREFIX_VEX_0F384A_X86_64_W_0_L_0) },
+    { Bad_Opcode },
+    { PREFIX_TABLE (PREFIX_EVEX_0F384A_X86_64_W_0_L_2) },
   },
 
   /* EVEX_LEN_0F385A */
@@ -63,6 +65,13 @@ static const struct dis386 evex_len_table[][3] = {
     { VEX_W_TABLE (EVEX_W_0F385B_L_2) },
   },
 
+  /* EVEX_LEN_0F386D_X86_64_W_0_M_1 */
+  {
+    { Bad_Opcode },
+    { Bad_Opcode },
+    { PREFIX_TABLE (PREFIX_EVEX_0F386D_X86_64_W_0_L_2) },
+  },
+
   /* EVEX_LEN_0F38C6 */
   {
     { Bad_Opcode },
@@ -91,6 +100,13 @@ static const struct dis386 evex_len_table[][3] = {
     { VEX_W_TABLE (VEX_W_0F3A01_L_1) },
   },
 
+  /* EVEX_LEN_0F3A07_X86_64_W_0 */
+  {
+    { Bad_Opcode },
+    { Bad_Opcode },
+    { PREFIX_TABLE (PREFIX_EVEX_0F3A07_X86_64_W_0_L_2) },
+  },
+
   /* EVEX_LEN_0F3A18 */
   {
     { Bad_Opcode },
@@ -161,6 +177,13 @@ static const struct dis386 evex_len_table[][3] = {
     { VEX_W_TABLE (EVEX_W_0F3A43_L_n) },
   },
 
+  /* EVEX_LEN_0F3A77_X86_64_W_0 */
+  {
+    { Bad_Opcode },
+    { Bad_Opcode },
+    { PREFIX_TABLE (PREFIX_EVEX_0F3A77_X86_64_W_0_L_2) },
+  },
+
   /* EVEX_LEN_MAP5_6E */
   {
     { PREFIX_TABLE (PREFIX_EVEX_MAP5_6E_L_0) },
diff --git a/opcodes/i386-dis-evex-prefix.h b/opcodes/i386-dis-evex-prefix.h
index 16fb2698390..a5b652a3420 100644
--- a/opcodes/i386-dis-evex-prefix.h
+++ b/opcodes/i386-dis-evex-prefix.h
@@ -243,6 +243,12 @@
     { VEX_W_TABLE (EVEX_W_0F383A_P_1) },
     { "%XEvpminuw",	{ XM, Vex, EXx }, 0 },
   },
+  /* PREFIX_EVEX_0F384A_W_0_L_2 */
+  {
+    { Bad_Opcode },
+    { "tcvtrowd2ps",	{ XM, Rtmm, VexGdq }, 0 },
+    { "tilemovrow",	{ XM, Rtmm, VexGdq }, 0 },
+  },
   /* PREFIX_EVEX_0F3852 */
   {
     { "vdpphp%XS",	{ XM, Vex, EXx }, 0 },
@@ -264,6 +270,13 @@
     { Bad_Opcode },
     { "vp2intersectY%DQ", { MaskG, Vex, EXx, EXxEVexS }, 0 },
   },
+  /* PREFIX_EVEX_0F386D_W_0_L_2 */
+  {
+    { "tcvtrowps2phh",	{ XM, Rtmm, VexGdq }, 0 },
+    { "tcvtrowps2bf16l",	{ XM, Rtmm, VexGdq }, 0 },
+    { "tcvtrowps2phl",	{ XM, Rtmm, VexGdq }, 0 },
+    { "tcvtrowps2bf16h",	{ XM, Rtmm, VexGdq }, 0 },
+  },
   /* PREFIX_EVEX_0F3872 */
   {
     { Bad_Opcode },
@@ -306,6 +319,13 @@
     { "%XEvfmsub213s%XW",	{ XMScalar, VexScalar, EXdq, EXxEVexR }, 0 },
     { "v4fnmadds%XS",	{ XMScalar, VexScalar, Mxmm }, 0 },
   },
+  /* PREFIX_EVEX_0F3A07_W_0_L_2 */
+  {
+    { "tcvtrowps2phh",	{ XM, Rtmm, Ib }, 0 },
+    { "tcvtrowd2ps",	{ XM, Rtmm, Ib }, 0 },
+    { "tilemovrow",	{ XM, Rtmm, Ib }, 0 },
+    { "tcvtrowps2bf16h",	{ XM, Rtmm, Ib }, 0 },
+  },
   /* PREFIX_EVEX_0F3A08 */
   {
     { "vrndscalep%XH",  { XM, EXxh, EXxEVexS, Ib }, 0 },
@@ -377,6 +397,13 @@
     { Bad_Opcode },
     { "vfpclasss%XW",	{ MaskG, EXdq, Ib }, 0 },
   },
+  /* PREFIX_EVEX_0F3A77_W_0_L_2 */
+  {
+    { Bad_Opcode },
+    { "tcvtrowps2bf16l",	{ XM, Rtmm, Ib }, 0 },
+    { Bad_Opcode },
+    { "tcvtrowps2phl",	{ XM, Rtmm, Ib }, 0 },
+  },
   /* PREFIX_EVEX_0F3AC2 */
   {
     { "vcmpp%XH", { MaskG, Vex, EXxh, EXxEVexS, CMP }, 0 },
diff --git a/opcodes/i386-dis-evex-w.h b/opcodes/i386-dis-evex-w.h
index 8c73320d633..c15469210a5 100644
--- a/opcodes/i386-dis-evex-w.h
+++ b/opcodes/i386-dis-evex-w.h
@@ -365,6 +365,10 @@
     { "vbroadcasti32x8",	{ XM, Mymm }, PREFIX_DATA },
     { "vbroadcasti64x4",	{ XM, Mymm }, PREFIX_DATA },
   },
+  /* EVEX_W_0F386D_X86_64 */
+  {
+    { EVEX_LEN_TABLE (EVEX_LEN_0F386D_X86_64_W_0) },
+  },
   /* EVEX_W_0F3870 */
   {
     { Bad_Opcode },
@@ -388,6 +392,10 @@
     { Bad_Opcode },
     { "vpmultishiftqb",	{ XM, Vex, EXx }, PREFIX_DATA },
   },
+  /* EVEX_W_0F3A07_X86_64 */
+  {
+    { EVEX_LEN_TABLE (EVEX_LEN_0F3A07_X86_64_W_0) },
+  },
   /* EVEX_W_0F3A18_L_n */
   {
     { "vinsertf32x4",	{ XM, Vex, EXxmm, Ib }, PREFIX_DATA },
@@ -456,6 +464,10 @@
     { Bad_Opcode },
     { "vpshrdw",   { XM, Vex, EXx, Ib }, 0 },
   },
+  /* EVEX_W_0F3A77_X86_64 */
+  {
+    { EVEX_LEN_TABLE (EVEX_LEN_0F3A77_X86_64_W_0) },
+  },
   /* EVEX_W_MAP4_8F_R_0 */
   {
     { "pop2", { { PUSH2_POP2_Fixup, q_mode}, Eq }, NO_PREFIX },
diff --git a/opcodes/i386-dis-evex-x86-64.h b/opcodes/i386-dis-evex-x86-64.h
index 9b0d7132268..8df4ba07bc5 100644
--- a/opcodes/i386-dis-evex-x86-64.h
+++ b/opcodes/i386-dis-evex-x86-64.h
@@ -3,3 +3,18 @@
     { Bad_Opcode },
     { VEX_W_TABLE (EVEX_W_0F384A_X86_64) },
   },
+  /* X86_64_EVEX_0F386D */
+  {
+    { Bad_Opcode },
+    { VEX_W_TABLE (EVEX_W_0F386D_X86_64) },
+  },
+  /* X86_64_EVEX_0F3A07 */
+  {
+    { Bad_Opcode },
+    { VEX_W_TABLE (EVEX_W_0F3A07_X86_64) },
+  },
+  /* X86_64_EVEX_0F3A77 */
+  {
+    { Bad_Opcode },
+    { VEX_W_TABLE (EVEX_W_0F3A77_X86_64) },
+  },
diff --git a/opcodes/i386-dis-evex.h b/opcodes/i386-dis-evex.h
index fdb4e07f2e1..895d617fed2 100644
--- a/opcodes/i386-dis-evex.h
+++ b/opcodes/i386-dis-evex.h
@@ -415,7 +415,7 @@ static const struct dis386 evex_table[][256] = {
     { Bad_Opcode },
     { Bad_Opcode },
     { Bad_Opcode },
-    { Bad_Opcode },
+    { X86_64_TABLE (X86_64_EVEX_0F386D) },
     { X86_64_EVEX_FROM_VEX_TABLE (X86_64_VEX_0F386E) },
     { X86_64_EVEX_FROM_VEX_TABLE (X86_64_VEX_0F386F) },
     /* 70 */
@@ -591,7 +591,7 @@ static const struct dis386 evex_table[][256] = {
     { VEX_W_TABLE (VEX_W_0F3A04) },
     { "%XEvpermilp%XD", { XM, EXx, Ib }, PREFIX_DATA },
     { Bad_Opcode },
-    { Bad_Opcode },
+    { X86_64_TABLE (X86_64_EVEX_0F3A07) },
     /* 08 */
     { PREFIX_TABLE (PREFIX_EVEX_0F3A08) },
     { "vrndscalep%XD", { XM, EXx, EXxEVexS, Ib }, PREFIX_DATA },
@@ -717,7 +717,7 @@ static const struct dis386 evex_table[][256] = {
     { Bad_Opcode },
     { Bad_Opcode },
     { Bad_Opcode },
-    { Bad_Opcode },
+    { X86_64_TABLE (X86_64_EVEX_0F3A77) },
     /* 78 */
     { Bad_Opcode },
     { Bad_Opcode },
diff --git a/opcodes/i386-dis.c b/opcodes/i386-dis.c
index adbd93c0489..bd00897c88f 100644
--- a/opcodes/i386-dis.c
+++ b/opcodes/i386-dis.c
@@ -1207,9 +1207,11 @@ enum
   PREFIX_EVEX_0F3838,
   PREFIX_EVEX_0F3839,
   PREFIX_EVEX_0F383A,
+  PREFIX_EVEX_0F384A_X86_64_W_0_L_2,
   PREFIX_EVEX_0F3852,
   PREFIX_EVEX_0F3853,
   PREFIX_EVEX_0F3868,
+  PREFIX_EVEX_0F386D_X86_64_W_0_L_2,
   PREFIX_EVEX_0F3872,
   PREFIX_EVEX_0F3874,
   PREFIX_EVEX_0F389A,
@@ -1217,6 +1219,7 @@ enum
   PREFIX_EVEX_0F38AA,
   PREFIX_EVEX_0F38AB,
 
+  PREFIX_EVEX_0F3A07_X86_64_W_0_L_2,
   PREFIX_EVEX_0F3A08,
   PREFIX_EVEX_0F3A0A,
   PREFIX_EVEX_0F3A26,
@@ -1228,6 +1231,7 @@ enum
   PREFIX_EVEX_0F3A57,
   PREFIX_EVEX_0F3A66,
   PREFIX_EVEX_0F3A67,
+  PREFIX_EVEX_0F3A77_X86_64_W_0_L_2,
   PREFIX_EVEX_0F3AC2,
 
   PREFIX_EVEX_MAP4_4x,
@@ -1374,6 +1378,11 @@ enum
   X86_64_VEX_MAP7_F8_L_0_W_0_R_0,
 
   X86_64_EVEX_0F384A,
+  X86_64_EVEX_0F386D,
+  X86_64_EVEX_0F3A07,
+  X86_64_EVEX_0F3A77,
+
+  X86_64_EVEX_MAP5_6F_M_0,
 };
 
 enum
@@ -1571,10 +1580,12 @@ enum
   EVEX_LEN_0F384A_X86_64_W_0,
   EVEX_LEN_0F385A,
   EVEX_LEN_0F385B,
+  EVEX_LEN_0F386D_X86_64_W_0,
   EVEX_LEN_0F38C6,
   EVEX_LEN_0F38C7,
   EVEX_LEN_0F3A00,
   EVEX_LEN_0F3A01,
+  EVEX_LEN_0F3A07_X86_64_W_0,
   EVEX_LEN_0F3A18,
   EVEX_LEN_0F3A19,
   EVEX_LEN_0F3A1A,
@@ -1585,6 +1596,7 @@ enum
   EVEX_LEN_0F3A3A,
   EVEX_LEN_0F3A3B,
   EVEX_LEN_0F3A43,
+  EVEX_LEN_0F3A77_X86_64_W_0,
 
   EVEX_LEN_MAP5_6E,
   EVEX_LEN_MAP5_7E,
@@ -1804,12 +1816,14 @@ enum
   EVEX_W_0F3859,
   EVEX_W_0F385A_L_n,
   EVEX_W_0F385B_L_2,
+  EVEX_W_0F386D_X86_64,
   EVEX_W_0F3870,
   EVEX_W_0F3872_P_2,
   EVEX_W_0F387A,
   EVEX_W_0F387B,
   EVEX_W_0F3883,
 
+  EVEX_W_0F3A07_X86_64,
   EVEX_W_0F3A18_L_n,
   EVEX_W_0F3A19_L_n,
   EVEX_W_0F3A1A_L_2,
@@ -1824,6 +1838,7 @@ enum
   EVEX_W_0F3A43_L_n,
   EVEX_W_0F3A70,
   EVEX_W_0F3A72,
+  EVEX_W_0F3A77_X86_64,
 
   EVEX_W_MAP4_8F_R_0,
   EVEX_W_MAP4_F8_P1_M_1,
@@ -14030,6 +14045,25 @@ OP_VEX (instr_info *ins, int bytemode, int sizeflag ATTRIBUTE_UNUSED)
 	}
 
       return true;
+
+    case v_mode:
+    case dq_mode:
+      if (ins->rex & REX_W)
+	oappend_register (ins, att_names64[reg]);
+      else if (bytemode == v_mode
+	       && !(sizeflag & DFLAG))
+	oappend_register (ins, att_names16[reg]);
+      else
+	oappend_register (ins, att_names32[reg]);
+      return true;
+
+    case b_mode:
+      oappend_register (ins, att_names8rex[reg]);
+      return true;
+
+    case q_mode:
+      oappend_register (ins, att_names64[reg]);
+      return true;
     }
 
   switch (ins->vex.length)
@@ -14041,22 +14075,6 @@ OP_VEX (instr_info *ins, int bytemode, int sizeflag ATTRIBUTE_UNUSED)
 	  names = att_names_xmm;
 	  ins->evex_used |= EVEX_len_used;
 	  break;
-	case v_mode:
-	case dq_mode:
-	  if (ins->rex & REX_W)
-	    names = att_names64;
-	  else if (bytemode == v_mode
-		   && !(sizeflag & DFLAG))
-	    names = att_names16;
-	  else
-	    names = att_names32;
-	  break;
-	case b_mode:
-	  names = att_names8rex;
-	  break;
-	case q_mode:
-	  names = att_names64;
-	  break;
 	case mask_bd_mode:
 	case mask_mode:
 	  if (reg > 0x7)
diff --git a/opcodes/i386-gen.c b/opcodes/i386-gen.c
index d2ecea2ecd5..d7e0b29bfbd 100644
--- a/opcodes/i386-gen.c
+++ b/opcodes/i386-gen.c
@@ -273,6 +273,8 @@ static const dependency isa_dependencies[] =
     "AMX_TILE" },
   { "AMX_MOVRS",
     "AMX_TILE" },
+  { "AMX_AVX512",
+    "AMX_TILE|AVX10_2" },
   { "KL",
     "SSE2" },
   { "WIDEKL",
@@ -443,6 +445,7 @@ static bitfield cpu_flags[] =
   BITFIELD (AMX_TF32),
   BITFIELD (AMX_FP8),
   BITFIELD (AMX_MOVRS),
+  BITFIELD (AMX_AVX512),
   BITFIELD (AMX_TILE),
   BITFIELD (MOVDIRI),
   BITFIELD (MOVDIR64B),
diff --git a/opcodes/i386-opc.h b/opcodes/i386-opc.h
index 77dad3159dd..cc303e0e818 100644
--- a/opcodes/i386-opc.h
+++ b/opcodes/i386-opc.h
@@ -256,6 +256,8 @@ enum i386_cpu
   CpuAMX_FP8,
   /* AMX-MOVRS Instructions support required.  */
   CpuAMX_MOVRS,
+  /* AMX-AVX512 Instructions support required.  */
+  CpuAMX_AVX512,
   /* AMX-TILE instructions required */
   CpuAMX_TILE,
   /* GFNI instructions required */
@@ -509,6 +511,7 @@ typedef union i386_cpu_flags
       unsigned int cpuamx_tf32:1;
       unsigned int cpuamx_fp8:1;
       unsigned int cpuamx_movrs:1;
+      unsigned int cpuamx_avx512:1;
       unsigned int cpuamx_tile:1;
       unsigned int cpugfni:1;
       unsigned int cpuvaes:1;
diff --git a/opcodes/i386-opc.tbl b/opcodes/i386-opc.tbl
index b28ceb5be19..731d94bc128 100644
--- a/opcodes/i386-opc.tbl
+++ b/opcodes/i386-opc.tbl
@@ -3243,6 +3243,21 @@ t2rpntlvw<z>rs<loc>, 0x<z:pfx>f8 | <loc:opc>, AMX_TRANSPOSE&APX_F(AMX_MOVRS), Si
 tileloaddrs, 0xf24a, APX_F(AMX_MOVRS), Sibmem|Vex128|EVex128|Space0F38|VexW0|NoSuf, { Unspecified|BaseIndex, RegTMM }
 tileloaddrst1, 0x664a, APX_F(AMX_MOVRS), Sibmem|Vex128|EVex128|Space0F38|VexW0|NoSuf, { Unspecified|BaseIndex, RegTMM }
 
+tcvtrowd2ps, 0xf34a, AMX_AVX512, Modrm|EVex512|Space0F38|Src2VVVV|VexW0|NoSuf, { Reg32, RegTMM, RegZMM }
+tcvtrowd2ps, 0xf307, AMX_AVX512, Modrm|EVex512|Space0F3A|VexW0|NoSuf, { Imm8, RegTMM, RegZMM }
+
+tcvtrowps2bf16h, 0xf26d, AMX_AVX512, Modrm|EVex512|Space0F38|Src2VVVV|VexW0|NoSuf, { Reg32, RegTMM, RegZMM }
+tcvtrowps2bf16h, 0xf207, AMX_AVX512, Modrm|EVex512|Space0F3A|VexW0|NoSuf, { Imm8, RegTMM, RegZMM }
+tcvtrowps2bf16l, 0xf36d, AMX_AVX512, Modrm|EVex512|Space0F38|Src2VVVV|VexW0|NoSuf, { Reg32, RegTMM, RegZMM }
+tcvtrowps2bf16l, 0xf377, AMX_AVX512, Modrm|EVex512|Space0F3A|VexW0|NoSuf, { Imm8, RegTMM, RegZMM }
+tcvtrowps2phh, 0x6d, AMX_AVX512, Modrm|EVex512|Space0F38|Src2VVVV|VexW0|NoSuf, { Reg32, RegTMM, RegZMM }
+tcvtrowps2phh, 0x07, AMX_AVX512, Modrm|EVex512|Space0F3A|VexW0|NoSuf, { Imm8, RegTMM, RegZMM }
+tcvtrowps2phl, 0x666d, AMX_AVX512, Modrm|EVex512|Space0F38|Src2VVVV|VexW0|NoSuf, { Reg32, RegTMM, RegZMM }
+tcvtrowps2phl, 0xf277, AMX_AVX512, Modrm|EVex512|Space0F3A|VexW0|NoSuf, { Imm8, RegTMM, RegZMM }
+
+tilemovrow, 0x664a, AMX_AVX512, Modrm|EVex512|Space0F38|Src2VVVV|VexW0|NoSuf, { Reg32, RegTMM, RegZMM }
+tilemovrow, 0x6607, AMX_AVX512, Modrm|EVex512|Space0F3A|VexW0|NoSuf, { Imm8, RegTMM, RegZMM }
+
 // AMX instructions end.
 
 // KEYLOCKER instructions.
-- 
2.31.1



More information about the Binutils mailing list