RE: [PING] [PATCH 1/1] gdb, i386: Enable AVX512-bfloat16 for i386 targets.

RE: [PING] [PATCH 1/1] gdb, i386: Enable AVX512-bfloat16 for i386 targets.

[Willgerodt, Felix]

Kindly pinging,

Felix

> -----Original Message-----
> From: Willgerodt, Felix <felix.willgerodt@intel.com>
> Sent: Donnerstag, 29. Juli 2021 15:05
> To: gdb-patches@sourceware.org
> Cc: Willgerodt, Felix <felix.willgerodt@intel.com>
> Subject: [PATCH 1/1] gdb, i386: Enable AVX512-bfloat16 for i386 targets.
> 
> Values of type bfloat16 can also be used on 32-bit targets, which was missed
> in the original enablement.  This also adjusts the testcase to pass with
> "unix/-m32", where only the lower 8 AVX registers are available.
> 
> gdb/ChangeLog:
> 2021-07-21  Felix Willgerodt  <Felix.Willgerodt@intel.com>
> 
> 	* features/i386/32bit-sse.xml: Add bfloat16 type.
> 	* features/i386/32bit-sse.c: Regenerated.
> 
> gdb/testsuite/ChangeLog:
> 2021-07-21  Felix Willgerodt  <Felix.Willgerodt@intel.com>
> 
> 	* gdb.arch/x86-avx512bf16.exp: Only use x/z/ymm 0-7.
> ---
>  gdb/features/i386/32bit-sse.c             |  5 +++++
>  gdb/features/i386/32bit-sse.xml           |  2 ++
>  gdb/testsuite/gdb.arch/x86-avx512bf16.exp | 12 ++++++------
>  3 files changed, 13 insertions(+), 6 deletions(-)
> 
> diff --git a/gdb/features/i386/32bit-sse.c b/gdb/features/i386/32bit-sse.c
> index e5eaf8feb59..b6d5fbaf0fc 100644
> --- a/gdb/features/i386/32bit-sse.c
> +++ b/gdb/features/i386/32bit-sse.c
> @@ -10,6 +10,9 @@ create_feature_i386_32bit_sse (struct target_desc
> *result, long regnum)
> 
>    feature = tdesc_create_feature (result, "org.gnu.gdb.i386.sse");
>    tdesc_type *element_type;
> +  element_type = tdesc_named_type (feature, "bfloat16");
> +  tdesc_create_vector (feature, "v8bf16", element_type, 8);
> +
>    element_type = tdesc_named_type (feature, "ieee_single");
>    tdesc_create_vector (feature, "v4f", element_type, 4);
> 
> @@ -31,6 +34,8 @@ create_feature_i386_32bit_sse (struct target_desc
> *result, long regnum)
>    tdesc_type_with_fields *type_with_fields;
>    type_with_fields = tdesc_create_union (feature, "vec128");
>    tdesc_type *field_type;
> +  field_type = tdesc_named_type (feature, "v8bf16");
> +  tdesc_add_field (type_with_fields, "v8_bfloat16", field_type);
>    field_type = tdesc_named_type (feature, "v4f");
>    tdesc_add_field (type_with_fields, "v4_float", field_type);
>    field_type = tdesc_named_type (feature, "v2d");
> diff --git a/gdb/features/i386/32bit-sse.xml b/gdb/features/i386/32bit-
> sse.xml
> index 7bf3e0f3828..8710622e944 100644
> --- a/gdb/features/i386/32bit-sse.xml
> +++ b/gdb/features/i386/32bit-sse.xml
> @@ -7,6 +7,7 @@
> 
>  <!DOCTYPE feature SYSTEM "gdb-target.dtd">
>  <feature name="org.gnu.gdb.i386.sse">
> +  <vector id="v8bf16" type="bfloat16" count="8"/>
>    <vector id="v4f" type="ieee_single" count="4"/>
>    <vector id="v2d" type="ieee_double" count="2"/>
>    <vector id="v16i8" type="int8" count="16"/>
> @@ -14,6 +15,7 @@
>    <vector id="v4i32" type="int32" count="4"/>
>    <vector id="v2i64" type="int64" count="2"/>
>    <union id="vec128">
> +    <field name="v8_bfloat16" type="v8bf16"/>
>      <field name="v4_float" type="v4f"/>
>      <field name="v2_double" type="v2d"/>
>      <field name="v16_int8" type="v16i8"/>
> diff --git a/gdb/testsuite/gdb.arch/x86-avx512bf16.exp
> b/gdb/testsuite/gdb.arch/x86-avx512bf16.exp
> index 20528169c64..528e79f02df 100644
> --- a/gdb/testsuite/gdb.arch/x86-avx512bf16.exp
> +++ b/gdb/testsuite/gdb.arch/x86-avx512bf16.exp
> @@ -64,11 +64,11 @@ gdb_test "print \$zmm6.v32_bfloat16\[1\]" "= 68.5"
>  # Test setting of bfloat values
>  gdb_test_no_output "set var \$xmm0.v8_bfloat16\[0\] = 32.25" \
>      "set %xmm0.v8_bfloat16\[0\]"
> -gdb_test_no_output "set var \$ymm8.v16_bfloat16\[1\] = 33.5" \
> -    "set %ymm8.v16_bfloat16\[1\]"
> -gdb_test_no_output "set var \$zmm16.v32_bfloat16\[2\] = 22.75" \
> -    "set %zmm16.v32_bfloat16\[2\]"
> +gdb_test_no_output "set var \$ymm4.v16_bfloat16\[1\] = 33.5" \
> +    "set %ymm4.v16_bfloat16\[1\]"
> +gdb_test_no_output "set var \$zmm7.v32_bfloat16\[2\] = 22.75" \
> +    "set %zmm7.v32_bfloat16\[2\]"
> 
>  gdb_test "p \$xmm0.v8_bfloat16\[0\]" "= 32.25"
> -gdb_test "p \$ymm8.v16_bfloat16\[1\]" "= 33.5"
> -gdb_test "p \$zmm16.v32_bfloat16\[2\]" "= 22.75"
> +gdb_test "p \$ymm4.v16_bfloat16\[1\]" "= 33.5"
> +gdb_test "p \$zmm7.v32_bfloat16\[2\]" "= 22.75"
> --
> 2.25.4

Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva  
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928