diff mbox series

[5/5] x86: support AVX10.1 vector size restrictions

Message ID 990c83c3-0776-efdd-e162-5c367f4ebdc2@suse.com
State New
Headers
Series x86: AVX10.1 (alternative attempt) |

Checks

Context Check Description
linaro-tcwg-bot/tcwg_binutils_build--master-arm success Testing passed
linaro-tcwg-bot/tcwg_binutils_build--master-aarch64 success Testing passed
linaro-tcwg-bot/tcwg_binutils_check--master-aarch64 success Testing passed
linaro-tcwg-bot/tcwg_binutils_check--master-arm success Testing passed

Commit Message

Jan Beulich Aug. 25, 2023, 12:47 p.m. UTC 
  Recognize "/<number>" suffixes on both -march=+avx10.1 and the
corresponding .arch directive, setting an upper bound on the vector size
that insns may use. Such a restriction can be reset by setting a new base
architecture, by using a suffix-less form, by disabling AVX10, or by
enabling any other VEX/EVEX-based vector extension.

While for most insns we can suppress their use with too wide operands
via registers becoming unavailable (or in Intel syntax memory operand
size specifiers not being recognized), mask register insns have to have
their minimum required vector size specified in a new attribute. (Of
course this new attribute could also be used on other insns.)

Note that .insn continues to be permitted to emit EVEX{512,256} (and
VEX256 ones) encodings regardless of vector size restrictions in place.
Of course these can't be expressed using zmm (or ymm) operands then,
but need using the EVEX.512.* forms (broadcast forms may be usable right
now, but this may go away so shouldn't be relied upon). This is why no
assertions should be added to build_{e,}vex_prefix().
---
It is unclear whether Vsz is a good name for the new attribute: The spec
leaves open how 256-bit embedded rounding is going to be expressed. Yet
that may require some similar attribute ...

Comments

H.J. Lu Aug. 29, 2023, 4:26 p.m. UTC | #1 
On Fri, Aug 25, 2023 at 5:48 AM Jan Beulich <jbeulich@suse.com> wrote:
>
> Recognize "/<number>" suffixes on both -march=+avx10.1 and the
> corresponding .arch directive, setting an upper bound on the vector size
> that insns may use. Such a restriction can be reset by setting a new base
> architecture, by using a suffix-less form, by disabling AVX10, or by
> enabling any other VEX/EVEX-based vector extension.
>
> While for most insns we can suppress their use with too wide operands
> via registers becoming unavailable (or in Intel syntax memory operand
> size specifiers not being recognized), mask register insns have to have
> their minimum required vector size specified in a new attribute. (Of
> course this new attribute could also be used on other insns.)
>
> Note that .insn continues to be permitted to emit EVEX{512,256} (and
> VEX256 ones) encodings regardless of vector size restrictions in place.
> Of course these can't be expressed using zmm (or ymm) operands then,
> but need using the EVEX.512.* forms (broadcast forms may be usable right
> now, but this may go away so shouldn't be relied upon). This is why no
> assertions should be added to build_{e,}vex_prefix().
> ---
> It is unclear whether Vsz is a good name for the new attribute: The spec
> leaves open how 256-bit embedded rounding is going to be expressed. Yet
> that may require some similar attribute ...
>
> --- a/gas/NEWS
> +++ b/gas/NEWS
> @@ -1,5 +1,7 @@
>  -*- text -*-
>
> +* Add support for Intel AVX10.1.
> +
>  * Add support for Intel PBNDKB instructions.
>
>  * Add support for Intel SM4 instructions.
> --- a/gas/doc/c-i386.texi
> +++ b/gas/doc/c-i386.texi
> @@ -213,6 +213,9 @@ accept various extension mnemonics.  For
>  @code{sm4},
>  @code{pbndkb},
>  @code{avx10.1},
> +@code{avx10.1/512},
> +@code{avx10.1/256},
> +@code{avx10.1/128},
>  @code{amx_int8},
>  @code{amx_bf16},
>  @code{amx_fp16},
> @@ -267,7 +270,11 @@ accept various extension mnemonics.  For
>  @code{svme} and
>  @code{padlock}.
>  Note that these extension mnemonics can be prefixed with @code{no} to revoke
> -the respective (and any dependent) functionality.
> +the respective (and any dependent) functionality.  Note further that the
> +suffixes permitted on @code{-march=avx10.<N>} enforce a vector length
> +restriction, i.e. despite these otherwise being "enabling" options, using
> +these suffixes will disable all insns with wider vector or mask register
> +operands.
>
>  When the @code{.arch} directive is used with @option{-march}, the
>  @code{.arch} directive will take precedent.
> @@ -1673,6 +1680,12 @@ an unconditional jump to the target.
>
>  Note that the sub-architecture specifiers (starting with a dot) can be prefixed
>  with @code{no} to revoke the respective (and any dependent) functionality.
> +Note further that @samp{.avx10.<N>} can be suffixed with a vector length
> +restriction (@samp{/256} or @samp{/128}, with @samp{/512} simply restoring the
> +default).  Despite these otherwise being "enabling" specifiers, using these
> +suffixes will disable all insns with wider vector or mask register operands.
> +On SVR4-derived platforms, the separator character @samp{/} can be replaced by
> +@samp{:}.
>
>  Following the CPU architecture (but not a sub-architecture, which are those
>  starting with a dot), you may specify @samp{jumps} or @samp{nojumps} to

Although CPUID bits in AVX10 spec may leave an impression that 128-bit,
256-bit and 512-bit vectors may be enabled independently.  But it also says

A “converged” version of Intel AVX10 with maximum vector lengths of 256
bits and 32-bit opmask registers will be supported across all Intel processors,
while 512-bit vector registers and 64-bit opmasks will continue to be supported
on some P-core processors.

Adding avx10.1/128 isn't necessary.
Jan Beulich Aug. 30, 2023, 7:57 a.m. UTC | #2 
On 29.08.2023 18:26, H.J. Lu wrote:
> On Fri, Aug 25, 2023 at 5:48 AM Jan Beulich <jbeulich@suse.com> wrote:
>> @@ -1673,6 +1680,12 @@ an unconditional jump to the target.
>>
>>  Note that the sub-architecture specifiers (starting with a dot) can be prefixed
>>  with @code{no} to revoke the respective (and any dependent) functionality.
>> +Note further that @samp{.avx10.<N>} can be suffixed with a vector length
>> +restriction (@samp{/256} or @samp{/128}, with @samp{/512} simply restoring the
>> +default).  Despite these otherwise being "enabling" specifiers, using these
>> +suffixes will disable all insns with wider vector or mask register operands.
>> +On SVR4-derived platforms, the separator character @samp{/} can be replaced by
>> +@samp{:}.
>>
>>  Following the CPU architecture (but not a sub-architecture, which are those
>>  starting with a dot), you may specify @samp{jumps} or @samp{nojumps} to
> 
> Although CPUID bits in AVX10 spec may leave an impression that 128-bit,
> 256-bit and 512-bit vectors may be enabled independently.  But it also says
> 
> A “converged” version of Intel AVX10 with maximum vector lengths of 256
> bits and 32-bit opmask registers will be supported across all Intel processors,
> while 512-bit vector registers and 64-bit opmasks will continue to be supported
> on some P-core processors.
> 
> Adding avx10.1/128 isn't necessary.

I agree it isn't necessary, but as expressed before I view it as desirable.
Apart from the sentence you quoted the spec later also says "There are
currently no plans to support an Intel AVX10/128 implementation." For my
choice of also supporting the 128-bit restriction I'd like to put emphasis
on "currently". I think I said before that emulation environments (qemu,
sde to name just two well-known examples) are free to implement such
further restricted ISAs without then becoming out-of-spec.

Plus supporting this mode right away has made me make certain adjustments
in what I'd call more clean a way, which I view as desirable as well.

Jan
H.J. Lu Aug. 30, 2023, 3:25 p.m. UTC | #3 
On Wed, Aug 30, 2023 at 12:57 AM Jan Beulich <jbeulich@suse.com> wrote:
>
> On 29.08.2023 18:26, H.J. Lu wrote:
> > On Fri, Aug 25, 2023 at 5:48 AM Jan Beulich <jbeulich@suse.com> wrote:
> >> @@ -1673,6 +1680,12 @@ an unconditional jump to the target.
> >>
> >>  Note that the sub-architecture specifiers (starting with a dot) can be prefixed
> >>  with @code{no} to revoke the respective (and any dependent) functionality.
> >> +Note further that @samp{.avx10.<N>} can be suffixed with a vector length
> >> +restriction (@samp{/256} or @samp{/128}, with @samp{/512} simply restoring the
> >> +default).  Despite these otherwise being "enabling" specifiers, using these
> >> +suffixes will disable all insns with wider vector or mask register operands.
> >> +On SVR4-derived platforms, the separator character @samp{/} can be replaced by
> >> +@samp{:}.
> >>
> >>  Following the CPU architecture (but not a sub-architecture, which are those
> >>  starting with a dot), you may specify @samp{jumps} or @samp{nojumps} to
> >
> > Although CPUID bits in AVX10 spec may leave an impression that 128-bit,
> > 256-bit and 512-bit vectors may be enabled independently.  But it also says
> >
> > A “converged” version of Intel AVX10 with maximum vector lengths of 256
> > bits and 32-bit opmask registers will be supported across all Intel processors,
> > while 512-bit vector registers and 64-bit opmasks will continue to be supported
> > on some P-core processors.
> >
> > Adding avx10.1/128 isn't necessary.
>
> I agree it isn't necessary, but as expressed before I view it as desirable.
> Apart from the sentence you quoted the spec later also says "There are
> currently no plans to support an Intel AVX10/128 implementation." For my
> choice of also supporting the 128-bit restriction I'd like to put emphasis
> on "currently". I think I said before that emulation environments (qemu,
> sde to name just two well-known examples) are free to implement such
> further restricted ISAs without then becoming out-of-spec.
>
> Plus supporting this mode right away has made me make certain adjustments
> in what I'd call more clean a way, which I view as desirable as well.
>

Since AVX10 spec doesn't specify if mask registers should be limited to
16 bits for AVX10/128, doing it in assembler is premature.
Jan Beulich Aug. 30, 2023, 4:16 p.m. UTC | #4 
On 30.08.2023 17:25, H.J. Lu wrote:
> On Wed, Aug 30, 2023 at 12:57 AM Jan Beulich <jbeulich@suse.com> wrote:
>>
>> On 29.08.2023 18:26, H.J. Lu wrote:
>>> On Fri, Aug 25, 2023 at 5:48 AM Jan Beulich <jbeulich@suse.com> wrote:
>>>> @@ -1673,6 +1680,12 @@ an unconditional jump to the target.
>>>>
>>>>  Note that the sub-architecture specifiers (starting with a dot) can be prefixed
>>>>  with @code{no} to revoke the respective (and any dependent) functionality.
>>>> +Note further that @samp{.avx10.<N>} can be suffixed with a vector length
>>>> +restriction (@samp{/256} or @samp{/128}, with @samp{/512} simply restoring the
>>>> +default).  Despite these otherwise being "enabling" specifiers, using these
>>>> +suffixes will disable all insns with wider vector or mask register operands.
>>>> +On SVR4-derived platforms, the separator character @samp{/} can be replaced by
>>>> +@samp{:}.
>>>>
>>>>  Following the CPU architecture (but not a sub-architecture, which are those
>>>>  starting with a dot), you may specify @samp{jumps} or @samp{nojumps} to
>>>
>>> Although CPUID bits in AVX10 spec may leave an impression that 128-bit,
>>> 256-bit and 512-bit vectors may be enabled independently.  But it also says
>>>
>>> A “converged” version of Intel AVX10 with maximum vector lengths of 256
>>> bits and 32-bit opmask registers will be supported across all Intel processors,
>>> while 512-bit vector registers and 64-bit opmasks will continue to be supported
>>> on some P-core processors.
>>>
>>> Adding avx10.1/128 isn't necessary.
>>
>> I agree it isn't necessary, but as expressed before I view it as desirable.
>> Apart from the sentence you quoted the spec later also says "There are
>> currently no plans to support an Intel AVX10/128 implementation." For my
>> choice of also supporting the 128-bit restriction I'd like to put emphasis
>> on "currently". I think I said before that emulation environments (qemu,
>> sde to name just two well-known examples) are free to implement such
>> further restricted ISAs without then becoming out-of-spec.
>>
>> Plus supporting this mode right away has made me make certain adjustments
>> in what I'd call more clean a way, which I view as desirable as well.
> 
> Since AVX10 spec doesn't specify if mask registers should be limited to
> 16 bits for AVX10/128, doing it in assembler is premature.

It's hard to see why they would remain wider. The more that they were 16
bits only in AVX512F.

Plus of course nobody needs to use the options to enforce the 128-bit
limit. The way I've coded it, it matches what the specification says.

Jan
H.J. Lu Aug. 30, 2023, 6 p.m. UTC | #5 
On Wed, Aug 30, 2023 at 9:16 AM Jan Beulich <jbeulich@suse.com> wrote:
>
> On 30.08.2023 17:25, H.J. Lu wrote:
> > On Wed, Aug 30, 2023 at 12:57 AM Jan Beulich <jbeulich@suse.com> wrote:
> >>
> >> On 29.08.2023 18:26, H.J. Lu wrote:
> >>> On Fri, Aug 25, 2023 at 5:48 AM Jan Beulich <jbeulich@suse.com> wrote:
> >>>> @@ -1673,6 +1680,12 @@ an unconditional jump to the target.
> >>>>
> >>>>  Note that the sub-architecture specifiers (starting with a dot) can be prefixed
> >>>>  with @code{no} to revoke the respective (and any dependent) functionality.
> >>>> +Note further that @samp{.avx10.<N>} can be suffixed with a vector length
> >>>> +restriction (@samp{/256} or @samp{/128}, with @samp{/512} simply restoring the
> >>>> +default).  Despite these otherwise being "enabling" specifiers, using these
> >>>> +suffixes will disable all insns with wider vector or mask register operands.
> >>>> +On SVR4-derived platforms, the separator character @samp{/} can be replaced by
> >>>> +@samp{:}.
> >>>>
> >>>>  Following the CPU architecture (but not a sub-architecture, which are those
> >>>>  starting with a dot), you may specify @samp{jumps} or @samp{nojumps} to
> >>>
> >>> Although CPUID bits in AVX10 spec may leave an impression that 128-bit,
> >>> 256-bit and 512-bit vectors may be enabled independently.  But it also says
> >>>
> >>> A “converged” version of Intel AVX10 with maximum vector lengths of 256
> >>> bits and 32-bit opmask registers will be supported across all Intel processors,
> >>> while 512-bit vector registers and 64-bit opmasks will continue to be supported
> >>> on some P-core processors.
> >>>
> >>> Adding avx10.1/128 isn't necessary.
> >>
> >> I agree it isn't necessary, but as expressed before I view it as desirable.
> >> Apart from the sentence you quoted the spec later also says "There are
> >> currently no plans to support an Intel AVX10/128 implementation." For my
> >> choice of also supporting the 128-bit restriction I'd like to put emphasis
> >> on "currently". I think I said before that emulation environments (qemu,
> >> sde to name just two well-known examples) are free to implement such
> >> further restricted ISAs without then becoming out-of-spec.
> >>
> >> Plus supporting this mode right away has made me make certain adjustments
> >> in what I'd call more clean a way, which I view as desirable as well.
> >
> > Since AVX10 spec doesn't specify if mask registers should be limited to
> > 16 bits for AVX10/128, doing it in assembler is premature.
>
> It's hard to see why they would remain wider. The more that they were 16
> bits only in AVX512F.
>
> Plus of course nobody needs to use the options to enforce the 128-bit
> limit. The way I've coded it, it matches what the specification says.
>

AVX10 spec only has

Quadword opmask instructions will only be supported on processors
supporting vector lengths of 512 bits.

It doesn't say anything about 32-bit mask.   32-bit mask can be useful
even with 16 byte vector.
Frager, Neal via Binutils Aug. 31, 2023, 5:56 a.m. UTC | #6 
> > >> I agree it isn't necessary, but as expressed before I view it as desirable.
> > >> Apart from the sentence you quoted the spec later also says "There are
> > >> currently no plans to support an Intel AVX10/128 implementation." For my
> > >> choice of also supporting the 128-bit restriction I'd like to put emphasis
> > >> on "currently". I think I said before that emulation environments (qemu,
> > >> sde to name just two well-known examples) are free to implement such
> > >> further restricted ISAs without then becoming out-of-spec.
> > >>
> > >> Plus supporting this mode right away has made me make certain adjustments
> > >> in what I'd call more clean a way, which I view as desirable as well.
> > >
> > > Since AVX10 spec doesn't specify if mask registers should be limited to
> > > 16 bits for AVX10/128, doing it in assembler is premature.
> >
> > It's hard to see why they would remain wider. The more that they were 16
> > bits only in AVX512F.
> >
> > Plus of course nobody needs to use the options to enforce the 128-bit
> > limit. The way I've coded it, it matches what the specification says.
> >
> 
> AVX10 spec only has
> 
> Quadword opmask instructions will only be supported on processors
> supporting vector lengths of 512 bits.
> 
> It doesn't say anything about 32-bit mask.   32-bit mask can be useful
> even with 16 byte vector.
> 
> --
> H.J.

The concern form my side is if there is an extreme case that overloads
registers, we might need to spill 32-bit register to 32-bit mask register
in the compiler.

Another minor concern is if there is finally a AVX10/128, although I do
not see that could happen, if we get a wrong choice here, it will take
some more time to correct the final assembler on the user side, which
I mean on the real OS.

However, I suppose both ok for me whether to allow 32-bit mask since
AVX10/128 is nowhere near in the future and it is a toy code to play with.
We could be some kind of conservative at first by just allowing 16-bit
mask register. Also, the code change is quite easy and no much worry on
changing that.

Thx,
Haochen
Jan Beulich Aug. 31, 2023, 7:18 a.m. UTC | #7 
On 31.08.2023 07:56, Jiang, Haochen wrote:
>>>>> I agree it isn't necessary, but as expressed before I view it as desirable.
>>>>> Apart from the sentence you quoted the spec later also says "There are
>>>>> currently no plans to support an Intel AVX10/128 implementation." For my
>>>>> choice of also supporting the 128-bit restriction I'd like to put emphasis
>>>>> on "currently". I think I said before that emulation environments (qemu,
>>>>> sde to name just two well-known examples) are free to implement such
>>>>> further restricted ISAs without then becoming out-of-spec.
>>>>>
>>>>> Plus supporting this mode right away has made me make certain adjustments
>>>>> in what I'd call more clean a way, which I view as desirable as well.
>>>>
>>>> Since AVX10 spec doesn't specify if mask registers should be limited to
>>>> 16 bits for AVX10/128, doing it in assembler is premature.
>>>
>>> It's hard to see why they would remain wider. The more that they were 16
>>> bits only in AVX512F.
>>>
>>> Plus of course nobody needs to use the options to enforce the 128-bit
>>> limit. The way I've coded it, it matches what the specification says.
>>>
>>
>> AVX10 spec only has
>>
>> Quadword opmask instructions will only be supported on processors
>> supporting vector lengths of 512 bits.
>>
>> It doesn't say anything about 32-bit mask.   32-bit mask can be useful
>> even with 16 byte vector.

How's that any different for 64-bit mask with 32-byte vector?

> The concern form my side is if there is an extreme case that overloads
> registers, we might need to spill 32-bit register to 32-bit mask register
> in the compiler.

How's that any different for spilling of 64-bit registers?

> Another minor concern is if there is finally a AVX10/128, although I do
> not see that could happen, if we get a wrong choice here, it will take
> some more time to correct the final assembler on the user side, which
> I mean on the real OS.
> 
> However, I suppose both ok for me whether to allow 32-bit mask since
> AVX10/128 is nowhere near in the future and it is a toy code to play with.
> We could be some kind of conservative at first by just allowing 16-bit
> mask register. Also, the code change is quite easy and no much worry on
> changing that.

Exactly: I'd rather be overly restrictive initially (which people can
easily work around by using .arch suitably around individual insns)
rather than being too permissive and then failing to flag mistakes.

Jan
Frager, Neal via Binutils Sept. 1, 2023, 6:21 a.m. UTC | #8 
> >>
> >> AVX10 spec only has
> >>
> >> Quadword opmask instructions will only be supported on processors
> >> supporting vector lengths of 512 bits.
> >>
> >> It doesn't say anything about 32-bit mask.   32-bit mask can be useful
> >> even with 16 byte vector.
> 
> How's that any different for 64-bit mask with 32-byte vector?
> 
> > The concern form my side is if there is an extreme case that overloads
> > registers, we might need to spill 32-bit register to 32-bit mask
> > register in the compiler.
> 
> How's that any different for spilling of 64-bit registers?
>

In GCC, under 256 bit mode, we will ban them. But since there is no definition
for 128 bit mode in GCC for a long time, it will leave some space of undefined.

Actually, the cost for that spill is really high. So it should be little concern, not a
blocking issue.

Thx,
Haochen
diff mbox series

Patch

--- a/gas/NEWS
+++ b/gas/NEWS
@@ -1,5 +1,7 @@ 
 -*- text -*-
 
+* Add support for Intel AVX10.1.
+
 * Add support for Intel PBNDKB instructions.
 
 * Add support for Intel SM4 instructions.
--- a/gas/doc/c-i386.texi
+++ b/gas/doc/c-i386.texi
@@ -213,6 +213,9 @@  accept various extension mnemonics.  For
 @code{sm4},
 @code{pbndkb},
 @code{avx10.1},
+@code{avx10.1/512},
+@code{avx10.1/256},
+@code{avx10.1/128},
 @code{amx_int8},
 @code{amx_bf16},
 @code{amx_fp16},
@@ -267,7 +270,11 @@  accept various extension mnemonics.  For
 @code{svme} and
 @code{padlock}.
 Note that these extension mnemonics can be prefixed with @code{no} to revoke
-the respective (and any dependent) functionality.
+the respective (and any dependent) functionality.  Note further that the
+suffixes permitted on @code{-march=avx10.<N>} enforce a vector length
+restriction, i.e. despite these otherwise being "enabling" options, using
+these suffixes will disable all insns with wider vector or mask register
+operands.
 
 When the @code{.arch} directive is used with @option{-march}, the
 @code{.arch} directive will take precedent.
@@ -1673,6 +1680,12 @@  an unconditional jump to the target.
 
 Note that the sub-architecture specifiers (starting with a dot) can be prefixed
 with @code{no} to revoke the respective (and any dependent) functionality.
+Note further that @samp{.avx10.<N>} can be suffixed with a vector length
+restriction (@samp{/256} or @samp{/128}, with @samp{/512} simply restoring the
+default).  Despite these otherwise being "enabling" specifiers, using these
+suffixes will disable all insns with wider vector or mask register operands.
+On SVR4-derived platforms, the separator character @samp{/} can be replaced by
+@samp{:}.
 
 Following the CPU architecture (but not a sub-architecture, which are those
 starting with a dot), you may specify @samp{jumps} or @samp{nojumps} to
--- a/gas/config/tc-i386.c
+++ b/gas/config/tc-i386.c
@@ -131,6 +131,7 @@  typedef struct
   unsigned int len:8;		/* arch string length */
   bool skip:1;			/* show_arch should skip this. */
   enum processor_type type;	/* arch type */
+  enum { vsz_none, vsz_set, vsz_reset } vsz; /* vector size control */
   i386_cpu_flags enable;		/* cpu feature enable flags */
   i386_cpu_flags disable;	/* cpu feature disable flags */
 }
@@ -841,6 +842,10 @@  static unsigned int sse2avx;
 /* Encode aligned vector move as unaligned vector move.  */
 static unsigned int use_unaligned_vector_move;
 
+/* Maximum permitted vector size. */
+#define VSZ_DEFAULT VSZ512
+static unsigned int vector_size = VSZ_DEFAULT;
+
 /* Encode scalar AVX instructions with specific vector length.  */
 static enum
   {
@@ -969,11 +974,14 @@  const relax_typeS md_relax_table[] =
 };
 
 #define ARCH(n, t, f, s) \
-  { STRING_COMMA_LEN (#n), s, PROCESSOR_ ## t, CPU_ ## f ## _FLAGS, \
+  { STRING_COMMA_LEN (#n), s, PROCESSOR_ ## t, vsz_none, CPU_ ## f ## _FLAGS, \
     CPU_NONE_FLAGS }
 #define SUBARCH(n, e, d, s) \
-  { STRING_COMMA_LEN (#n), s, PROCESSOR_NONE, CPU_ ## e ## _FLAGS, \
+  { STRING_COMMA_LEN (#n), s, PROCESSOR_NONE, vsz_none, CPU_ ## e ## _FLAGS, \
     CPU_ ## d ## _FLAGS }
+#define VECARCH(n, e, d, v) \
+  { STRING_COMMA_LEN (#n), false, PROCESSOR_NONE, vsz_ ## v, \
+    CPU_ ## e ## _FLAGS, CPU_ ## d ## _FLAGS }
 
 static const arch_entry cpu_arch[] =
 {
@@ -1035,15 +1043,15 @@  static const arch_entry cpu_arch[] =
   SUBARCH (sse4.1, SSE4_1, ANY_SSE4_1, false),
   SUBARCH (sse4.2, SSE4_2, ANY_SSE4_2, false),
   SUBARCH (sse4, SSE4_2, ANY_SSE4_1, false),
-  SUBARCH (avx, AVX, ANY_AVX, false),
-  SUBARCH (avx2, AVX2, ANY_AVX2, false),
-  SUBARCH (avx512f, AVX512F, ANY_AVX512F, false),
-  SUBARCH (avx512cd, AVX512CD, ANY_AVX512CD, false),
-  SUBARCH (avx512er, AVX512ER, ANY_AVX512ER, false),
-  SUBARCH (avx512pf, AVX512PF, ANY_AVX512PF, false),
-  SUBARCH (avx512dq, AVX512DQ, ANY_AVX512DQ, false),
-  SUBARCH (avx512bw, AVX512BW, ANY_AVX512BW, false),
-  SUBARCH (avx512vl, AVX512VL, ANY_AVX512VL, false),
+  VECARCH (avx, AVX, ANY_AVX, reset),
+  VECARCH (avx2, AVX2, ANY_AVX2, reset),
+  VECARCH (avx512f, AVX512F, ANY_AVX512F, reset),
+  VECARCH (avx512cd, AVX512CD, ANY_AVX512CD, reset),
+  VECARCH (avx512er, AVX512ER, ANY_AVX512ER, reset),
+  VECARCH (avx512pf, AVX512PF, ANY_AVX512PF, reset),
+  VECARCH (avx512dq, AVX512DQ, ANY_AVX512DQ, reset),
+  VECARCH (avx512bw, AVX512BW, ANY_AVX512BW, reset),
+  VECARCH (avx512vl, AVX512VL, ANY_AVX512VL, reset),
   SUBARCH (monitor, MONITOR, MONITOR, false),
   SUBARCH (vmx, VMX, ANY_VMX, false),
   SUBARCH (vmfunc, VMFUNC, ANY_VMFUNC, false),
@@ -1095,15 +1103,15 @@  static const arch_entry cpu_arch[] =
   SUBARCH (prefetchwt1, PREFETCHWT1, PREFETCHWT1, false),
   SUBARCH (se1, SE1, SE1, false),
   SUBARCH (clwb, CLWB, CLWB, false),
-  SUBARCH (avx512ifma, AVX512IFMA, ANY_AVX512IFMA, false),
-  SUBARCH (avx512vbmi, AVX512VBMI, ANY_AVX512VBMI, false),
-  SUBARCH (avx512_4fmaps, AVX512_4FMAPS, ANY_AVX512_4FMAPS, false),
-  SUBARCH (avx512_4vnniw, AVX512_4VNNIW, ANY_AVX512_4VNNIW, false),
-  SUBARCH (avx512_vpopcntdq, AVX512_VPOPCNTDQ, ANY_AVX512_VPOPCNTDQ, false),
-  SUBARCH (avx512_vbmi2, AVX512_VBMI2, ANY_AVX512_VBMI2, false),
-  SUBARCH (avx512_vnni, AVX512_VNNI, ANY_AVX512_VNNI, false),
-  SUBARCH (avx512_bitalg, AVX512_BITALG, ANY_AVX512_BITALG, false),
-  SUBARCH (avx_vnni, AVX_VNNI, ANY_AVX_VNNI, false),
+  VECARCH (avx512ifma, AVX512IFMA, ANY_AVX512IFMA, reset),
+  VECARCH (avx512vbmi, AVX512VBMI, ANY_AVX512VBMI, reset),
+  VECARCH (avx512_4fmaps, AVX512_4FMAPS, ANY_AVX512_4FMAPS, reset),
+  VECARCH (avx512_4vnniw, AVX512_4VNNIW, ANY_AVX512_4VNNIW, reset),
+  VECARCH (avx512_vpopcntdq, AVX512_VPOPCNTDQ, ANY_AVX512_VPOPCNTDQ, reset),
+  VECARCH (avx512_vbmi2, AVX512_VBMI2, ANY_AVX512_VBMI2, reset),
+  VECARCH (avx512_vnni, AVX512_VNNI, ANY_AVX512_VNNI, reset),
+  VECARCH (avx512_bitalg, AVX512_BITALG, ANY_AVX512_BITALG, reset),
+  VECARCH (avx_vnni, AVX_VNNI, ANY_AVX_VNNI, reset),
   SUBARCH (clzero, CLZERO, CLZERO, false),
   SUBARCH (mwaitx, MWAITX, MWAITX, false),
   SUBARCH (ospke, OSPKE, ANY_OSPKE, false),
@@ -1112,8 +1120,8 @@  static const arch_entry cpu_arch[] =
   SUBARCH (ibt, IBT, IBT, false),
   SUBARCH (shstk, SHSTK, SHSTK, false),
   SUBARCH (gfni, GFNI, ANY_GFNI, false),
-  SUBARCH (vaes, VAES, ANY_VAES, false),
-  SUBARCH (vpclmulqdq, VPCLMULQDQ, ANY_VPCLMULQDQ, false),
+  VECARCH (vaes, VAES, ANY_VAES, reset),
+  VECARCH (vpclmulqdq, VPCLMULQDQ, ANY_VPCLMULQDQ, reset),
   SUBARCH (wbnoinvd, WBNOINVD, WBNOINVD, false),
   SUBARCH (pconfig, PCONFIG, PCONFIG, false),
   SUBARCH (waitpkg, WAITPKG, WAITPKG, false),
@@ -1125,9 +1133,9 @@  static const arch_entry cpu_arch[] =
   SUBARCH (amx_tile, AMX_TILE, ANY_AMX_TILE, false),
   SUBARCH (movdiri, MOVDIRI, MOVDIRI, false),
   SUBARCH (movdir64b, MOVDIR64B, MOVDIR64B, false),
-  SUBARCH (avx512_bf16, AVX512_BF16, ANY_AVX512_BF16, false),
-  SUBARCH (avx512_vp2intersect, AVX512_VP2INTERSECT,
-	   ANY_AVX512_VP2INTERSECT, false),
+  VECARCH (avx512_bf16, AVX512_BF16, ANY_AVX512_BF16, reset),
+  VECARCH (avx512_vp2intersect, AVX512_VP2INTERSECT,
+	   ANY_AVX512_VP2INTERSECT, reset),
   SUBARCH (tdx, TDX, TDX, false),
   SUBARCH (enqcmd, ENQCMD, ENQCMD, false),
   SUBARCH (serialize, SERIALIZE, SERIALIZE, false),
@@ -1139,24 +1147,24 @@  static const arch_entry cpu_arch[] =
   SUBARCH (widekl, WIDEKL, ANY_WIDEKL, false),
   SUBARCH (uintr, UINTR, UINTR, false),
   SUBARCH (hreset, HRESET, HRESET, false),
-  SUBARCH (avx512_fp16, AVX512_FP16, ANY_AVX512_FP16, false),
+  VECARCH (avx512_fp16, AVX512_FP16, ANY_AVX512_FP16, reset),
   SUBARCH (prefetchi, PREFETCHI, PREFETCHI, false),
-  SUBARCH (avx_ifma, AVX_IFMA, ANY_AVX_IFMA, false),
-  SUBARCH (avx_vnni_int8, AVX_VNNI_INT8, ANY_AVX_VNNI_INT8, false),
+  VECARCH (avx_ifma, AVX_IFMA, ANY_AVX_IFMA, reset),
+  VECARCH (avx_vnni_int8, AVX_VNNI_INT8, ANY_AVX_VNNI_INT8, reset),
   SUBARCH (cmpccxadd, CMPCCXADD, CMPCCXADD, false),
   SUBARCH (wrmsrns, WRMSRNS, WRMSRNS, false),
   SUBARCH (msrlist, MSRLIST, MSRLIST, false),
-  SUBARCH (avx_ne_convert, AVX_NE_CONVERT, ANY_AVX_NE_CONVERT, false),
+  VECARCH (avx_ne_convert, AVX_NE_CONVERT, ANY_AVX_NE_CONVERT, reset),
   SUBARCH (rao_int, RAO_INT, RAO_INT, false),
   SUBARCH (rmpquery, RMPQUERY, ANY_RMPQUERY, false),
   SUBARCH (fred, FRED, ANY_FRED, false),
   SUBARCH (lkgs, LKGS, ANY_LKGS, false),
-  SUBARCH (avx_vnni_int16, AVX_VNNI_INT16, ANY_AVX_VNNI_INT16, false),
-  SUBARCH (sha512, SHA512, ANY_SHA512, false),
-  SUBARCH (sm3, SM3, ANY_SM3, false),
-  SUBARCH (sm4, SM4, ANY_SM4, false),
+  VECARCH (avx_vnni_int16, AVX_VNNI_INT16, ANY_AVX_VNNI_INT16, reset),
+  VECARCH (sha512, SHA512, ANY_SHA512, reset),
+  VECARCH (sm3, SM3, ANY_SM3, reset),
+  VECARCH (sm4, SM4, ANY_SM4, reset),
   SUBARCH (pbndkb, PBNDKB, PBNDKB, false),
-  SUBARCH (avx10.1, AVX10_1, ANY_AVX512F, false),
+  VECARCH (avx10.1, AVX10_1, ANY_AVX512F, set),
 };
 
 #undef SUBARCH
@@ -2790,6 +2798,7 @@  set_cpu_arch (int dummy ATTRIBUTE_UNUSED
     i386_cpu_flags isa_flags;
     enum processor_type isa;
     enum flag_code flag_code;
+    unsigned int vector_size;
     char stackop_size;
     bool no_cond_jump_promotion;
   } arch_stack_entry;
@@ -2825,6 +2834,7 @@  set_cpu_arch (int dummy ATTRIBUTE_UNUSED
       top->isa = cpu_arch_isa;
       top->isa_flags = cpu_arch_isa_flags;
       top->flag_code = flag_code;
+      top->vector_size = vector_size;
       top->stackop_size = stackop_size;
       top->no_cond_jump_promotion = no_cond_jump_promotion;
 
@@ -2865,6 +2875,7 @@  set_cpu_arch (int dummy ATTRIBUTE_UNUSED
 	  cpu_arch_flags = top->flags;
 	  cpu_arch_isa = top->isa;
 	  cpu_arch_isa_flags = top->isa_flags;
+	  vector_size = top->vector_size;
 	  no_cond_jump_promotion = top->no_cond_jump_promotion;
 
 	  XDELETE (top);
@@ -2905,6 +2916,8 @@  set_cpu_arch (int dummy ATTRIBUTE_UNUSED
 	      cpu_arch_tune_flags = cpu_arch_isa_flags;
 	    }
 
+	  vector_size = VSZ_DEFAULT;
+
 	  j = ARRAY_SIZE (cpu_arch) + 1;
 	}
     }
@@ -2939,6 +2952,9 @@  set_cpu_arch (int dummy ATTRIBUTE_UNUSED
 		  cpu_arch_tune = cpu_arch_isa;
 		  cpu_arch_tune_flags = cpu_arch_isa_flags;
 		}
+
+	      vector_size = VSZ_DEFAULT;
+
 	      pre_386_16bit_warned = false;
 	      break;
 	    }
@@ -2959,6 +2975,38 @@  set_cpu_arch (int dummy ATTRIBUTE_UNUSED
 	      = cpu_flags_or (cpu_arch_isa_flags, cpu_arch[j].enable);
 
 	  (void) restore_line_pointer (e);
+
+	  switch (cpu_arch[j].vsz)
+	    {
+	    default:
+	      break;
+
+	    case vsz_set:
+#ifdef SVR4_COMMENT_CHARS
+	      if (*input_line_pointer == ':' || *input_line_pointer == '/')
+#else
+	      if (*input_line_pointer == '/')
+#endif
+		{
+		  ++input_line_pointer;
+		  switch (get_absolute_expression ())
+		    {
+		    case 512: vector_size = VSZ512; break;
+		    case 256: vector_size = VSZ256; break;
+		    case 128: vector_size = VSZ128; break;
+		    default:
+		      as_bad (_("Unrecognized vector size specifier"));
+		      ignore_rest_of_line ();
+		      return;
+		    }
+		  break;
+		}
+		/* Fall through.  */
+	    case vsz_reset:
+	      vector_size = VSZ_DEFAULT;
+	      break;
+	    }
+
 	  demand_empty_rest_of_line ();
 	  return;
 	}
@@ -2979,6 +3027,9 @@  set_cpu_arch (int dummy ATTRIBUTE_UNUSED
 		cpu_arch_isa_flags = flags;
 	      }
 
+	    if (cpu_arch[j].vsz == vsz_set)
+	      vector_size = VSZ_DEFAULT;
+
 	    (void) restore_line_pointer (e);
 	    demand_empty_rest_of_line ();
 	    return;
@@ -5431,9 +5482,11 @@  md_assemble (char *line)
 	case RegSIMD:
 	  if (i.tm.operand_types[j].bitfield.tmmword)
 	    i.xstate |= xstate_tmm;
-	  else if (i.tm.operand_types[j].bitfield.zmmword)
+	  else if (i.tm.operand_types[j].bitfield.zmmword
+		   && vector_size >= VSZ512)
 	    i.xstate |= xstate_zmm;
-	  else if (i.tm.operand_types[j].bitfield.ymmword)
+	  else if (i.tm.operand_types[j].bitfield.ymmword
+		   && vector_size >= VSZ256)
 	    i.xstate |= xstate_ymm;
 	  else if (i.tm.operand_types[j].bitfield.xmmword)
 	    i.xstate |= xstate_xmm;
@@ -6606,9 +6659,13 @@  check_VecOperands (const insn_template *
 	  type.bitfield.xmmword = 1;
 	  break;
 	case 32:
+	  if (vector_size < VSZ256)
+	    goto bad_broadcast;
 	  type.bitfield.ymmword = 1;
 	  break;
 	case 64:
+	  if (vector_size < VSZ512)
+	    goto bad_broadcast;
 	  type.bitfield.zmmword = 1;
 	  break;
 	default:
@@ -6820,6 +6877,19 @@  VEX_check_encoding (const insn_template
       return 1;
     }
 
+  /* Vector size restrictions.  */
+  if ((vector_size < VSZ512
+       && (t->opcode_modifier.evex == EVEX512
+	   || t->opcode_modifier.vsz >= VSZ512))
+      || (vector_size < VSZ256
+	  && (t->opcode_modifier.evex == EVEX256
+	      || t->opcode_modifier.vex == VEX256
+	      || t->opcode_modifier.vsz >= VSZ256)))
+    {
+      i.error = unsupported;
+      return 1;
+    }
+
   if (i.vec_encoding == vex_encoding_evex)
     {
       /* This instruction must be encoded with EVEX prefix.  */
@@ -7608,8 +7678,27 @@  process_suffix (void)
 
 	  for (op = 0; op < i.tm.operands; ++op)
 	    {
-	      if (is_evex_encoding (&i.tm)
-		  && !cpu_arch_flags.bitfield.cpuavx512vl)
+	      if (vector_size < VSZ512)
+		{
+		  i.tm.operand_types[op].bitfield.zmmword = 0;
+		  if (vector_size < VSZ256)
+		    {
+		      i.tm.operand_types[op].bitfield.ymmword = 0;
+		      if (i.tm.operand_types[op].bitfield.xmmword
+			  && (i.tm.opcode_modifier.evex == EVEXDYN
+			      || (!i.tm.opcode_modifier.evex
+				  && is_evex_encoding (&i.tm))))
+			i.tm.opcode_modifier.evex = EVEX128;
+		    }
+		  else if (i.tm.operand_types[op].bitfield.ymmword
+			   && !i.tm.operand_types[op].bitfield.xmmword
+			   && (i.tm.opcode_modifier.evex == EVEXDYN
+			       || (!i.tm.opcode_modifier.evex
+				   && is_evex_encoding (&i.tm))))
+		    i.tm.opcode_modifier.evex = EVEX256;
+		}
+	      else if (is_evex_encoding (&i.tm)
+		       && !cpu_arch_flags.bitfield.cpuavx512vl)
 		{
 		  if (i.tm.operand_types[op].bitfield.ymmword)
 		    i.tm.operand_types[op].bitfield.xmmword = 0;
@@ -13857,6 +13946,12 @@  static bool check_register (const reg_en
 	}
     }
 
+  if (vector_size < VSZ512 && r->reg_type.bitfield.zmmword)
+    return false;
+
+  if (vector_size < VSZ256 && r->reg_type.bitfield.ymmword)
+    return false;
+
   if (r->reg_type.bitfield.tmmword
       && (!cpu_arch_flags.bitfield.cpuamx_tile
           || flag_code != CODE_64BIT))
@@ -14355,13 +14450,21 @@  md_parse_option (int c, const char *arg)
 	arch++;
       do
 	{
+	  char *vsz;
+
 	  if (*arch == '.')
 	    as_fatal (_("invalid -march= option: `%s'"), arg);
 	  next = strchr (arch, '+');
 	  if (next)
 	    *next++ = '\0';
+	  vsz = strchr (arch, '/');
+	  if (vsz)
+	    *vsz++ = '\0';
 	  for (j = 0; j < ARRAY_SIZE (cpu_arch); j++)
 	    {
+	      if (vsz && cpu_arch[j].vsz != vsz_set)
+		continue;
+
 	      if (arch == saved && cpu_arch[j].type != PROCESSOR_NONE
 	          && strcmp (arch, cpu_arch[j].name) == 0)
 		{
@@ -14380,6 +14483,7 @@  md_parse_option (int c, const char *arg)
 		      cpu_arch_tune = cpu_arch_isa;
 		      cpu_arch_tune_flags = cpu_arch_isa_flags;
 		    }
+		  vector_size = VSZ_DEFAULT;
 		  break;
 		}
 	      else if (cpu_arch[j].type == PROCESSOR_NONE
@@ -14402,6 +14506,37 @@  md_parse_option (int c, const char *arg)
 		    cpu_arch_isa_flags
 		      = cpu_flags_or (cpu_arch_isa_flags,
 				      cpu_arch[j].enable);
+
+		  switch (cpu_arch[j].vsz)
+		    {
+		    default:
+		      break;
+
+		    case vsz_set:
+		      if (vsz)
+			{
+			  char *end;
+			  unsigned long val = strtoul (vsz, &end, 0);
+
+			  if (*end)
+			    val = 0;
+			  switch (val)
+			    {
+			    case 512: vector_size = VSZ512; break;
+			    case 256: vector_size = VSZ256; break;
+			    case 128: vector_size = VSZ128; break;
+			    default:
+			      as_warn (_("Unrecognized vector size specifier ignored"));
+			      break;
+			    }
+			  break;
+			}
+			/* Fall through.  */
+		    case vsz_reset:
+		      vector_size = VSZ_DEFAULT;
+		      break;
+		    }
+
 		  break;
 		}
 	    }
@@ -14423,6 +14558,8 @@  md_parse_option (int c, const char *arg)
 			cpu_arch_flags = flags;
 			cpu_arch_isa_flags = flags;
 		      }
+		    if (cpu_arch[j].vsz == vsz_set)
+		      vector_size = VSZ_DEFAULT;
 		    break;
 		  }
 	    }
--- a/gas/config/tc-i386-intel.c
+++ b/gas/config/tc-i386-intel.c
@@ -768,10 +768,20 @@  i386_intel_operand (char *operand_string
 	  break;
 
 	case O_ymmword_ptr:
+	  if (vector_size < VSZ256)
+	    {
+	      as_bad (_("256-bit operands disabled"));
+	      return 0;
+	    }
 	  i.types[this_operand].bitfield.ymmword = 1;
 	  break;
 
 	case O_zmmword_ptr:
+	  if (vector_size < VSZ512)
+	    {
+	      as_bad (_("512-bit operands disabled"));
+	      return 0;
+	    }
 	  i.types[this_operand].bitfield.zmmword = 1;
 	  break;
 
--- a/gas/testsuite/gas/i386/avx10.1-256-1.d
+++ b/gas/testsuite/gas/i386/avx10.1-256-1.d
@@ -1,4 +1,4 @@ 
-#as: -I${srcdir}/$subdir
+#as: --divide -I${srcdir}/$subdir
 #objdump: -dw
 #name: AVX10.1/256 (part 1)
 #dump: avx512f_vl.d
--- a/gas/testsuite/gas/i386/avx10.1-256-1.s
+++ b/gas/testsuite/gas/i386/avx10.1-256-1.s
@@ -1,3 +1,3 @@ 
 	.arch generic32
-	.arch .avx10.1
+	.arch .avx10.1/256
 	.include "avx512f_vl.s"
--- a/gas/testsuite/gas/i386/avx10.1-256-2.d
+++ b/gas/testsuite/gas/i386/avx10.1-256-2.d
@@ -1,4 +1,4 @@ 
-#as: -I${srcdir}/$subdir
+#as: --divide -I${srcdir}/$subdir
 #objdump: -dw
 #name: AVX10.1/256 (part 2)
 #dump: avx512bw_vl.d
--- a/gas/testsuite/gas/i386/avx10.1-256-2.s
+++ b/gas/testsuite/gas/i386/avx10.1-256-2.s
@@ -1,3 +1,3 @@ 
 	.arch generic32
-	.arch .avx10.1
+	.arch .avx10.1/256
 	.include "avx512bw_vl.s"
--- a/gas/testsuite/gas/i386/avx10.1-256-3.d
+++ b/gas/testsuite/gas/i386/avx10.1-256-3.d
@@ -1,4 +1,4 @@ 
-#as: -I${srcdir}/$subdir
+#as: --divide -I${srcdir}/$subdir
 #objdump: -dw
 #name: AVX10.1/256 (part 3)
 #dump: avx512dq_vl.d
--- a/gas/testsuite/gas/i386/avx10.1-256-3.s
+++ b/gas/testsuite/gas/i386/avx10.1-256-3.s
@@ -1,3 +1,3 @@ 
 	.arch generic32
-	.arch .avx10.1
+	.arch .avx10.1/256
 	.include "avx512dq_vl.s"
--- a/gas/testsuite/gas/i386/avx10.1-256-4.d
+++ b/gas/testsuite/gas/i386/avx10.1-256-4.d
@@ -1,4 +1,4 @@ 
-#as: -I${srcdir}/$subdir
+#as: --divide -I${srcdir}/$subdir
 #objdump: -dw
 #name: AVX10.1/256 (part 4)
 #dump: avx512_fp16_vl.d
--- a/gas/testsuite/gas/i386/avx10.1-256-4.s
+++ b/gas/testsuite/gas/i386/avx10.1-256-4.s
@@ -1,3 +1,3 @@ 
 	.arch generic32
-	.arch .avx10.1
+	.arch .avx10.1/256
 	.include "avx512_fp16_vl.s"
--- a/gas/testsuite/gas/i386/avx10.1-256-5.d
+++ b/gas/testsuite/gas/i386/avx10.1-256-5.d
@@ -1,4 +1,4 @@ 
-#as: -I${srcdir}/$subdir
+#as: --divide -I${srcdir}/$subdir
 #objdump: -dw
 #name: AVX10.1/256 (part 5)
 
--- a/gas/testsuite/gas/i386/avx10.1-256-5.s
+++ b/gas/testsuite/gas/i386/avx10.1-256-5.s
@@ -1,5 +1,5 @@ 
 	.arch generic32
-	.arch .avx10.1
+	.arch .avx10.1/256
 
 	.include "avx512bitalg_vl.s"
 
--- a/gas/testsuite/gas/i386/avx10.1-256-gfni.d
+++ b/gas/testsuite/gas/i386/avx10.1-256-gfni.d
@@ -1,4 +1,4 @@ 
-#as: -I${srcdir}/$subdir
+#as: --divide -I${srcdir}/$subdir
 #objdump: -dw
 #name: AVX10.1/256 + GFNI
 #dump: avx512vl_gfni.d
--- a/gas/testsuite/gas/i386/avx10.1-256-gfni.s
+++ b/gas/testsuite/gas/i386/avx10.1-256-gfni.s
@@ -1,4 +1,4 @@ 
 	.arch generic32
-	.arch .avx10.1
+	.arch .avx10.1/256
 	.arch .gfni
 	.include "avx512vl_gfni.s"
--- a/gas/testsuite/gas/i386/avx10.1-256-vaes.d
+++ b/gas/testsuite/gas/i386/avx10.1-256-vaes.d
@@ -1,4 +1,4 @@ 
-#as: -I${srcdir}/$subdir
+#as: --divide -I${srcdir}/$subdir
 #objdump: -dw
 #name: AVX10.1/256 + VAES
 #dump: avx512vl_vaes.d
--- a/gas/testsuite/gas/i386/avx10.1-256-vaes.s
+++ b/gas/testsuite/gas/i386/avx10.1-256-vaes.s
@@ -1,4 +1,4 @@ 
 	.arch generic32
-	.arch .avx10.1
+	.arch .avx10.1/256
 	.arch .vaes
 	.include "avx512vl_vaes.s"
--- a/gas/testsuite/gas/i386/avx10.1-256-vpclmulqdq.d
+++ b/gas/testsuite/gas/i386/avx10.1-256-vpclmulqdq.d
@@ -1,4 +1,4 @@ 
-#as: -I${srcdir}/$subdir
+#as: --divide -I${srcdir}/$subdir
 #objdump: -dw
 #name: AVX10.1/256 + VPLCMULQDQ
 #dump: avx512vl_vpclmulqdq.d
--- a/gas/testsuite/gas/i386/avx10.1-256-vpclmulqdq.s
+++ b/gas/testsuite/gas/i386/avx10.1-256-vpclmulqdq.s
@@ -1,4 +1,4 @@ 
 	.arch generic32
-	.arch .avx10.1
+	.arch .avx10.1/256
 	.arch .vpclmulqdq
 	.include "avx512vl_vpclmulqdq.s"
--- /dev/null
+++ b/gas/testsuite/gas/i386/avx10-vsz.l
@@ -0,0 +1,294 @@ 
+.*: Assembler messages:
+.*:7: Warning: default
+.*:46: Error: ambiguous operand size for `vcvtpd2ph'
+.*:47: Error: ambiguous operand size for `vcvtpd2ps'
+.*:48: Error: ambiguous operand size for `vfpclassps'
+.*:7: Warning: \.avx10\.1/256
+.*:12: Error: unsupported instruction `kunpckdq'
+.*:16: Error: unsupported instruction `kaddq'
+.*:17: Error: unsupported instruction `kandq'
+.*:18: Error: unsupported instruction `kandnq'
+.*:19: Error: unsupported instruction `kmovq'
+.*:20: Error: unsupported instruction `kmovq'
+.*:21: Error: unsupported instruction `kmovq'
+.*:22: Error: unsupported instruction `knotq'
+.*:23: Error: unsupported instruction `korq'
+.*:24: Error: unsupported instruction `kortestq'
+.*:25: Error: unsupported instruction `kshiftlq'
+.*:26: Error: unsupported instruction `kshiftrq'
+.*:27: Error: unsupported instruction `ktestq'
+.*:28: Error: unsupported instruction `kxorq'
+.*:29: Error: unsupported instruction `kxnorq'
+.*:32: Error: unsupported instruction `vcvtpd2phz'
+.*:34: Error: unsupported broadcast for `vcvtpd2ph'
+.*:37: Error: unsupported instruction `vcvtpd2ps'
+.*:43: Error: unsupported instruction `vfpclasspsz'
+.*:46: Error: ambiguous operand size for `vcvtpd2ph'
+.*:47: Error: ambiguous operand size for `vcvtpd2ps'
+.*:48: Error: ambiguous operand size for `vfpclassps'
+.*:7: Warning: \.avx10\.1/128
+.*:10: Error: unsupported instruction `kmovd'
+.*:11: Error: unsupported instruction `kmovd'
+.*:12: Error: unsupported instruction `kunpckdq'
+.*:13: Error: unsupported instruction `kunpckwd'
+.*:16: Error: unsupported instruction `kaddd'
+.*:17: Error: unsupported instruction `kandd'
+.*:18: Error: unsupported instruction `kandnd'
+.*:19: Error: unsupported instruction `kmovd'
+.*:20: Error: unsupported instruction `kmovd'
+.*:21: Error: unsupported instruction `kmovd'
+.*:22: Error: unsupported instruction `knotd'
+.*:23: Error: unsupported instruction `kord'
+.*:24: Error: unsupported instruction `kortestd'
+.*:25: Error: unsupported instruction `kshiftld'
+.*:26: Error: unsupported instruction `kshiftrd'
+.*:27: Error: unsupported instruction `ktestd'
+.*:28: Error: unsupported instruction `kxord'
+.*:29: Error: unsupported instruction `kxnord'
+.*:16: Error: unsupported instruction `kaddq'
+.*:17: Error: unsupported instruction `kandq'
+.*:18: Error: unsupported instruction `kandnq'
+.*:19: Error: unsupported instruction `kmovq'
+.*:20: Error: unsupported instruction `kmovq'
+.*:21: Error: unsupported instruction `kmovq'
+.*:22: Error: unsupported instruction `knotq'
+.*:23: Error: unsupported instruction `korq'
+.*:24: Error: unsupported instruction `kortestq'
+.*:25: Error: unsupported instruction `kshiftlq'
+.*:26: Error: unsupported instruction `kshiftrq'
+.*:27: Error: unsupported instruction `ktestq'
+.*:28: Error: unsupported instruction `kxorq'
+.*:29: Error: unsupported instruction `kxnorq'
+.*:32: Error: unsupported instruction `vcvtpd2phz'
+.*:33: Error: unsupported instruction `vcvtpd2phy'
+.*:34: Error: unsupported broadcast for `vcvtpd2ph'
+.*:35: Error: unsupported broadcast for `vcvtpd2ph'
+.*:37: Error: .*
+.*:38: Error: unsupported instruction `vcvtpd2psy'
+.*:39: Error: unsupported instruction `vcvtpd2psy'
+.*:40: Error: unsupported broadcast for `vcvtpd2ps'
+.*:42: Error: unsupported instruction `vfpclasspsy'
+.*:43: Error: unsupported instruction `vfpclasspsz'
+.*:7: Warning: \.avx10\.1
+.*:46: Error: ambiguous operand size for `vcvtpd2ph'
+.*:47: Error: ambiguous operand size for `vcvtpd2ps'
+.*:48: Error: ambiguous operand size for `vfpclassps'
+#...
+[ 	]*[0-9]+[ 	]+>  \.arch generic32
+[ 	]*[0-9]+[ 	]+>  \.arch default
+[ 	]*[0-9]+[ 	]+\?+ C5FB93D1[ 	]+>  kmovd %k1,%edx
+[ 	]*[0-9]+[ 	]+\?+ C5FB92D1[ 	]+>  kmovd %ecx,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC4B D9[ 	]+>  kunpckdq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C5EC4BD9[ 	]+>  kunpckwd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  \.irp sz,d,q
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED4A D9[ 	]+>>  kaddd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED41 D9[ 	]+>>  kandd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED42 D9[ 	]+>>  kandnd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F990 D1[ 	]+>>  kmovd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F991 0A[ 	]+>>  kmovd %k1,\(%edx\)
+[ 	]*[0-9]+[ 	]+\?+ C4E1F990 11[ 	]+>>  kmovd \(%ecx\),%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F944 D1[ 	]+>>  knotd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED45 D9[ 	]+>>  kord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F998 D1[ 	]+>>  kortestd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E37933 DA01[ 	]+>>  kshiftld \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E37931 DA01[ 	]+>>  kshiftrd \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F999 D1[ 	]+>>  ktestd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED47 D9[ 	]+>>  kxord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED46 D9[ 	]+>>  kxnord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC4A D9[ 	]+>>  kaddq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC41 D9[ 	]+>>  kandq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC42 D9[ 	]+>>  kandnq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F890 D1[ 	]+>>  kmovq %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F891 0A[ 	]+>>  kmovq %k1,\(%edx\)
+[ 	]*[0-9]+[ 	]+\?+ C4E1F890 11[ 	]+>>  kmovq \(%ecx\),%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F844 D1[ 	]+>>  knotq %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC45 D9[ 	]+>>  korq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F898 D1[ 	]+>>  kortestq %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E3F933 DA01[ 	]+>>  kshiftlq \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E3F931 DA01[ 	]+>>  kshiftrq \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F899 D1[ 	]+>>  ktestq %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC47 D9[ 	]+>>  kxorq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC46 D9[ 	]+>>  kxnorq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD48 5A00[ 	]+>  vcvtpd2phz \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD28 5A00[ 	]+>  vcvtpd2phy \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD58 5A00[ 	]+>  vcvtpd2ph \(%eax\)\{1to8\},%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD38 5A00[ 	]+>  vcvtpd2ph \(%eax\)\{1to4\},%xmm0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+\?+ 62F1FD48 5A00[ 	]+>  vcvtpd2ps \(%eax\),%ymm0
+[ 	]*[0-9]+[ 	]+\?+ C5FD5A00[ 	]+>  vcvtpd2psy \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F1FD29 5A00[ 	]+>  vcvtpd2psy \(%eax\),%xmm0\{%k1\}
+[ 	]*[0-9]+[ 	]+\?+ 62F1FD38 5A00[ 	]+>  vcvtpd2ps \(%eax\)\{1to4\},%xmm0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+\?+ 62F37D28 660000[ 	]+>  vfpclasspsy \$0,\(%eax\),%k0
+[ 	]*[0-9]+[ 	]+\?+ 62F37D48 660000[ 	]+>  vfpclasspsz \$0,\(%eax\),%k0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  \.intel_syntax noprefix
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ph xmm0,\[eax\]
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ps xmm0,\[eax\]
+[ 	]*[0-9]+[ 	]+>  vfpclassps k0,\[eax\],0
+#...
+[ 	]*[0-9]+[ 	]+>  \.arch generic32
+[ 	]*[0-9]+[ 	]+>  \.arch \.avx10\.1/256
+[ 	]*[0-9]+[ 	]+\?+ C5FB93D1[ 	]+>  kmovd %k1,%edx
+[ 	]*[0-9]+[ 	]+\?+ C5FB92D1[ 	]+>  kmovd %ecx,%k2
+[ 	]*[0-9]+[ 	]+>  kunpckdq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C5EC4BD9[ 	]+>  kunpckwd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  \.irp sz,d,q
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED4A D9[ 	]+>>  kaddd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED41 D9[ 	]+>>  kandd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED42 D9[ 	]+>>  kandnd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F990 D1[ 	]+>>  kmovd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F991 0A[ 	]+>>  kmovd %k1,\(%edx\)
+[ 	]*[0-9]+[ 	]+\?+ C4E1F990 11[ 	]+>>  kmovd \(%ecx\),%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F944 D1[ 	]+>>  knotd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED45 D9[ 	]+>>  kord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F998 D1[ 	]+>>  kortestd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E37933 DA01[ 	]+>>  kshiftld \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E37931 DA01[ 	]+>>  kshiftrd \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F999 D1[ 	]+>>  ktestd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED47 D9[ 	]+>>  kxord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED46 D9[ 	]+>>  kxnord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kaddq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kandq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kandnq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kmovq %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kmovq %k1,\(%edx\)
+[ 	]*[0-9]+[ 	]+>>  kmovq \(%ecx\),%k2
+[ 	]*[0-9]+[ 	]+>>  knotq %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  korq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kortestq %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kshiftlq \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kshiftrq \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  ktestq %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kxorq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kxnorq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  vcvtpd2phz \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD28 5A00[ 	]+>  vcvtpd2phy \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ph \(%eax\)\{1to8\},%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD38 5A00[ 	]+>  vcvtpd2ph \(%eax\)\{1to4\},%xmm0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ps \(%eax\),%ymm0
+[ 	]*[0-9]+[ 	]+\?+ C5FD5A00[ 	]+>  vcvtpd2psy \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F1FD29 5A00[ 	]+>  vcvtpd2psy \(%eax\),%xmm0\{%k1\}
+[ 	]*[0-9]+[ 	]+\?+ 62F1FD38 5A00[ 	]+>  vcvtpd2ps \(%eax\)\{1to4\},%xmm0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+\?+ 62F37D28 660000[ 	]+>  vfpclasspsy \$0,\(%eax\),%k0
+[ 	]*[0-9]+[ 	]+>  vfpclasspsz \$0,\(%eax\),%k0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  \.intel_syntax noprefix
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ph xmm0,\[eax\]
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ps xmm0,\[eax\]
+[ 	]*[0-9]+[ 	]+>  vfpclassps k0,\[eax\],0
+#...
+[ 	]*[0-9]+[ 	]+>  \.arch generic32
+[ 	]*[0-9]+[ 	]+>  \.arch \.avx10\.1/128
+[ 	]*[0-9]+[ 	]+>  kmovd %k1,%edx
+[ 	]*[0-9]+[ 	]+>  kmovd %ecx,%k2
+[ 	]*[0-9]+[ 	]+>  kunpckdq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>  kunpckwd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  \.irp sz,d,q
+[ 	]*[0-9]+[ 	]+>>  kaddd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kandd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kandnd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kmovd %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kmovd %k1,\(%edx\)
+[ 	]*[0-9]+[ 	]+>>  kmovd \(%ecx\),%k2
+[ 	]*[0-9]+[ 	]+>>  knotd %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kortestd %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kshiftld \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kshiftrd \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  ktestd %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kxord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kxnord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kaddq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kandq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kandnq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kmovq %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kmovq %k1,\(%edx\)
+[ 	]*[0-9]+[ 	]+>>  kmovq \(%ecx\),%k2
+[ 	]*[0-9]+[ 	]+>>  knotq %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  korq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kortestq %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kshiftlq \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kshiftrq \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  ktestq %k1,%k2
+[ 	]*[0-9]+[ 	]+>>  kxorq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+>>  kxnorq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  vcvtpd2phz \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+>  vcvtpd2phy \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ph \(%eax\)\{1to8\},%xmm0
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ph \(%eax\)\{1to4\},%xmm0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ps \(%eax\),%ymm0
+[ 	]*[0-9]+[ 	]+>  vcvtpd2psy \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+>  vcvtpd2psy \(%eax\),%xmm0\{%k1\}
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ps \(%eax\)\{1to4\},%xmm0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  vfpclasspsy \$0,\(%eax\),%k0
+[ 	]*[0-9]+[ 	]+>  vfpclasspsz \$0,\(%eax\),%k0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  \.intel_syntax noprefix
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD08 5A00[ 	]+>  vcvtpd2ph xmm0,\[eax\]
+[ 	]*[0-9]+[ 	]+\?+ C5F95A00[ 	]+>  vcvtpd2ps xmm0,\[eax\]
+[ 	]*[0-9]+[ 	]+\?+ 62F37D08 660000[ 	]+>  vfpclassps k0,\[eax\],0
+#...
+[ 	]*[0-9]+[ 	]+\?+ C5FB93D1[ 	]+>  kmovd %k1,%edx
+[ 	]*[0-9]+[ 	]+\?+ C5FB92D1[ 	]+>  kmovd %ecx,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC4B D9[ 	]+>  kunpckdq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C5EC4BD9[ 	]+>  kunpckwd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  \.irp sz,d,q
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED4A D9[ 	]+>>  kaddd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED41 D9[ 	]+>>  kandd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED42 D9[ 	]+>>  kandnd %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F990 D1[ 	]+>>  kmovd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F991 0A[ 	]+>>  kmovd %k1,\(%edx\)
+[ 	]*[0-9]+[ 	]+\?+ C4E1F990 11[ 	]+>>  kmovd \(%ecx\),%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F944 D1[ 	]+>>  knotd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED45 D9[ 	]+>>  kord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F998 D1[ 	]+>>  kortestd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E37933 DA01[ 	]+>>  kshiftld \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E37931 DA01[ 	]+>>  kshiftrd \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F999 D1[ 	]+>>  ktestd %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED47 D9[ 	]+>>  kxord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1ED46 D9[ 	]+>>  kxnord %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC4A D9[ 	]+>>  kaddq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC41 D9[ 	]+>>  kandq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC42 D9[ 	]+>>  kandnq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F890 D1[ 	]+>>  kmovq %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F891 0A[ 	]+>>  kmovq %k1,\(%edx\)
+[ 	]*[0-9]+[ 	]+\?+ C4E1F890 11[ 	]+>>  kmovq \(%ecx\),%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1F844 D1[ 	]+>>  knotq %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC45 D9[ 	]+>>  korq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F898 D1[ 	]+>>  kortestq %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E3F933 DA01[ 	]+>>  kshiftlq \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E3F931 DA01[ 	]+>>  kshiftrq \$1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1F899 D1[ 	]+>>  ktestq %k1,%k2
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC47 D9[ 	]+>>  kxorq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+\?+ C4E1EC46 D9[ 	]+>>  kxnorq %k1,%k2,%k3
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD48 5A00[ 	]+>  vcvtpd2phz \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD28 5A00[ 	]+>  vcvtpd2phy \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD58 5A00[ 	]+>  vcvtpd2ph \(%eax\)\{1to8\},%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F5FD38 5A00[ 	]+>  vcvtpd2ph \(%eax\)\{1to4\},%xmm0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+\?+ 62F1FD48 5A00[ 	]+>  vcvtpd2ps \(%eax\),%ymm0
+[ 	]*[0-9]+[ 	]+\?+ C5FD5A00[ 	]+>  vcvtpd2psy \(%eax\),%xmm0
+[ 	]*[0-9]+[ 	]+\?+ 62F1FD29 5A00[ 	]+>  vcvtpd2psy \(%eax\),%xmm0\{%k1\}
+[ 	]*[0-9]+[ 	]+\?+ 62F1FD38 5A00[ 	]+>  vcvtpd2ps \(%eax\)\{1to4\},%xmm0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+\?+ 62F37D28 660000[ 	]+>  vfpclasspsy \$0,\(%eax\),%k0
+[ 	]*[0-9]+[ 	]+\?+ 62F37D48 660000[ 	]+>  vfpclasspsz \$0,\(%eax\),%k0
+[ 	]*[0-9]+[ 	]+> *
+[ 	]*[0-9]+[ 	]+>  \.intel_syntax noprefix
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ph xmm0,\[eax\]
+[ 	]*[0-9]+[ 	]+>  vcvtpd2ps xmm0,\[eax\]
+[ 	]*[0-9]+[ 	]+>  vfpclassps k0,\[eax\],0
+#pass
--- /dev/null
+++ b/gas/testsuite/gas/i386/avx10-vsz.s
@@ -0,0 +1,52 @@ 
+# Test AVX10 vector size restriction
+	.text
+
+	.irp isa, default, .avx10.1/256, .avx10.1/128, .avx10.1
+
+	.att_syntax prefix
+	.warning "\isa"
+	.arch generic32
+	.arch \isa
+	kmovd		%k1, %edx
+	kmovd		%ecx, %k2
+	kunpckdq	%k1, %k2, %k3
+	kunpckwd	%k1, %k2, %k3
+
+	.irp sz, d, q
+	kadd\sz		%k1, %k2, %k3
+	kand\sz		%k1, %k2, %k3
+	kandn\sz	%k1, %k2, %k3
+	kmov\sz		%k1, %k2
+	kmov\sz		%k1, (%edx)
+	kmov\sz		(%ecx), %k2
+	knot\sz		%k1, %k2
+	kor\sz		%k1, %k2, %k3
+	kortest\sz	%k1, %k2
+	kshiftl\sz	$1, %k2, %k3
+	kshiftr\sz	$1, %k2, %k3
+	ktest\sz	%k1, %k2
+	kxor\sz		%k1, %k2, %k3
+	kxnor\sz	%k1, %k2, %k3
+	.endr
+
+	vcvtpd2phz	(%eax), %xmm0
+	vcvtpd2phy	(%eax), %xmm0
+	vcvtpd2ph	(%eax){1to8}, %xmm0
+	vcvtpd2ph	(%eax){1to4}, %xmm0
+
+	vcvtpd2ps	(%eax), %ymm0
+	vcvtpd2psy	(%eax), %xmm0
+	vcvtpd2psy	(%eax), %xmm0{%k1}
+	vcvtpd2ps	(%eax){1to4}, %xmm0
+
+	vfpclasspsy	$0, (%eax), %k0
+	vfpclasspsz	$0, (%eax), %k0
+
+	.intel_syntax noprefix
+	vcvtpd2ph	xmm0, [eax]
+	vcvtpd2ps	xmm0, [eax]
+	vfpclassps	k0, [eax], 0
+
+	.endr
+
+	.p2align 4
--- a/gas/testsuite/gas/i386/i386.exp
+++ b/gas/testsuite/gas/i386/i386.exp
@@ -238,7 +238,7 @@  if [gas_32_check] then {
     run_list_test "noavx-3" "-al"
     run_dump_test "noavx-4"
     run_list_test "noavx512-1" "-almn"
-    run_list_test "noavx512-2" "-al"
+    run_list_test "noavx512-2" "-almn --divide"
     run_dump_test "noextreg"
     run_dump_test "xmmhi32"
     run_dump_test "xsave"
@@ -579,6 +579,7 @@  if [gas_32_check] then {
     run_dump_test "avx10.1-256-gfni"
     run_dump_test "avx10.1-256-vaes"
     run_dump_test "avx10.1-256-vpclmulqdq"
+    run_list_test "avx10-vsz" "-almn --divide --listing-lhs-width=2"
     run_dump_test "fpu-bad"
     run_dump_test "cet"
     run_dump_test "cet-intel"
--- a/gas/testsuite/gas/i386/noavx512-2.l
+++ b/gas/testsuite/gas/i386/noavx512-2.l
@@ -1,106 +1,195 @@ 
 .*: Assembler messages:
-.*:26: Error: .*unsupported masking.*
-.*:27: Error: .*unsupported masking.*
-.*:29: Error: .*unsupported instruction.*
-.*:30: Error: .*unsupported instruction.*
-.*:32: Error: .*unsupported instruction.*
-.*:33: Error: .*unsupported instruction.*
-.*:36: Error: .*unsupported masking.*
-.*:37: Error: .*unsupported masking.*
-.*:39: Error: .*unsupported instruction.*
-.*:40: Error: .*unsupported instruction.*
-.*:43: Error: .*unsupported instruction.*
-.*:44: Error: .*unsupported instruction.*
-GAS LISTING .*
-#...
-[ 	]*1[ 	]+\# Test \.arch \.noavx512vl
-[ 	]*2[ 	]+\.text
-[ 	]*3[ 	]+\?\?\?\? 62F27D4F 		vpabsb %zmm5, %zmm6\{%k7\}		\# AVX512BW
-[ 	]*3[ 	]+1CF5
-[ 	]*4[ 	]+\?\?\?\? 62F27D0F 		vpabsb %xmm5, %xmm6\{%k7\}		\# AVX512BW \+ AVX512VL
-[ 	]*4[ 	]+1CF5
-[ 	]*5[ 	]+\?\?\?\? 62F27D2F 		vpabsb %ymm5, %ymm6\{%k7\}		\# AVX512BW \+ AVX512VL
-[ 	]*5[ 	]+1CF5
-[ 	]*6[ 	]+\?\?\?\? 62F27D48 		vpconflictd %zmm5, %zmm6		\# AVX412CD
-[ 	]*6[ 	]+C4F5
-[ 	]*7[ 	]+\?\?\?\? 62F27D08 		vpconflictd %xmm5, %xmm6		\# AVX412CD \+ AVX512VL
-[ 	]*7[ 	]+C4F5
-[ 	]*8[ 	]+\?\?\?\? 62F27D28 		vpconflictd %ymm5, %ymm6		\# AVX412CD \+ AVX512VL
-[ 	]*8[ 	]+C4F5
-[ 	]*9[ 	]+\?\?\?\? 62F1FD4F 		vcvtpd2qq \(%ecx\), %zmm6\{%k7\}		\# AVX512DQ
-[ 	]*9[ 	]+7B31
-[ 	]*10[ 	]+\?\?\?\? 62F1FD0F 		vcvtpd2qq \(%ecx\), %xmm6\{%k7\}		\# AVX512DQ \+ AVX512VL
-[ 	]*10[ 	]+7B31
-[ 	]*11[ 	]+\?\?\?\? 62F1FD2F 		vcvtpd2qq \(%ecx\), %ymm6\{%k7\}		\# AVX512DQ \+ AVX512VL
-[ 	]*11[ 	]+7B31
-[ 	]*12[ 	]+\?\?\?\? 62F27D4F 		vexp2ps %zmm5, %zmm6\{%k7\}		\# AVX512ER
-[ 	]*12[ 	]+C8F5
-[ 	]*13[ 	]+\?\?\?\? 62F1D54F 		vaddpd %zmm4, %zmm5, %zmm6\{%k7\}		\# AVX512F
-[ 	]*13[ 	]+58F4
-[ 	]*14[ 	]+\?\?\?\? 62F1D50F 		vaddpd %xmm4, %xmm5, %xmm6\{%k7\}		\# AVX512F \+ AVX512VL
-[ 	]*14[ 	]+58F4
-[ 	]*15[ 	]+\?\?\?\? 62F1D52F 		vaddpd %ymm4, %ymm5, %ymm6\{%k7\}		\# AVX512F \+ AVX512VL
-[ 	]*15[ 	]+58F4
-[ 	]*16[ 	]+\?\?\?\? 62F2D54F 		vpmadd52luq %zmm4, %zmm5, %zmm6\{%k7\}	\# AVX512IFMA
-[ 	]*16[ 	]+B4F4
-[ 	]*17[ 	]+\?\?\?\? 62F2D50F 		vpmadd52luq %xmm4, %xmm5, %xmm6\{%k7\}	\# AVX512IFMA \+ AVX512VL
-[ 	]*17[ 	]+B4F4
-[ 	]*18[ 	]+\?\?\?\? 62F2D52F 		vpmadd52luq %ymm4, %ymm5, %ymm6\{%k7\}	\# AVX512IFMA \+ AVX512VL
-[ 	]*18[ 	]+B4F4
-[ 	]*19[ 	]+\?\?\?\? 62F2FD49 		vgatherpf0dpd 23\(%ebp,%ymm7,8\)\{%k1\}	\# AVX512PF
-[ 	]*19[ 	]+C68CFD17 
-[ 	]*19[ 	]+000000
-[ 	]*20[ 	]+\?\?\?\? 62F2554F 		vpermb %zmm4, %zmm5, %zmm6\{%k7\}		\# AVX512VBMI
-[ 	]*20[ 	]+8DF4
-[ 	]*21[ 	]+\?\?\?\? 62F2550F 		vpermb %xmm4, %xmm5, %xmm6\{%k7\}		\# AVX512VBMI \+ AVX512VL
-[ 	]*21[ 	]+8DF4
-[ 	]*22[ 	]+\?\?\?\? 62F2552F 		vpermb %ymm4, %ymm5, %ymm6\{%k7\}		\# AVX512VBMI \+ AVX512VL
-[ 	]*22[ 	]+8DF4
-[ 	]*23[ 	]+
-[ 	]*24[ 	]+\.arch \.noavx512vl
-[ 	]*25[ 	]+\?\?\?\? 62F27D4F 		vpabsb %zmm5, %zmm6\{%k7\}		\# AVX512BW
-[ 	]*25[ 	]+1CF5
-[ 	]*26[ 	]+vpabsb %xmm5, %xmm6\{%k7\}		\# AVX512BW \+ AVX512VL
-[ 	]*27[ 	]+vpabsb %ymm5, %ymm6\{%k7\}		\# AVX512BW \+ AVX512VL
-[ 	]*28[ 	]+\?\?\?\? 62F27D48 		vpconflictd %zmm5, %zmm6		\# AVX412CD
-[ 	]*28[ 	]+C4F5
-[ 	]*29[ 	]+vpconflictd %xmm5, %xmm6		\# AVX412CD \+ AVX512VL
-[ 	]*30[ 	]+vpconflictd %ymm5, %ymm6		\# AVX412CD \+ AVX512VL
-[ 	]*31[ 	]+\?\?\?\? 62F1FD4F 		vcvtpd2qq \(%ecx\), %zmm6\{%k7\}		\# AVX512DQ
-[ 	]*31[ 	]+7B31
-[ 	]*32[ 	]+vcvtpd2qq \(%ecx\), %xmm6\{%k7\}		\# AVX512DQ \+ AVX512VL
-[ 	]*33[ 	]+vcvtpd2qq \(%ecx\), %ymm6\{%k7\}		\# AVX512DQ \+ AVX512VL
-GAS LISTING .*
-
-
-[ 	]*34[ 	]+\?\?\?\? 62F27D4F 		vexp2ps %zmm5, %zmm6\{%k7\}		\# AVX512ER
-[ 	]*34[ 	]+C8F5
-[ 	]*35[ 	]+\?\?\?\? 62F1D54F 		vaddpd %zmm4, %zmm5, %zmm6\{%k7\}		\# AVX512F
-[ 	]*35[ 	]+58F4
-[ 	]*36[ 	]+vaddpd %xmm4, %xmm5, %xmm6\{%k7\}		\# AVX512F \+ AVX512VL
-[ 	]*37[ 	]+vaddpd %ymm4, %ymm5, %ymm6\{%k7\}		\# AVX512F \+ AVX512VL
-[ 	]*38[ 	]+\?\?\?\? 62F2D54F 		vpmadd52luq %zmm4, %zmm5, %zmm6\{%k7\}	\# AVX512IFMA
-[ 	]*38[ 	]+B4F4
-[ 	]*39[ 	]+vpmadd52luq %xmm4, %xmm5, %xmm6\{%k7\}	\# AVX512IFMA \+ AVX512VL
-[ 	]*40[ 	]+vpmadd52luq %ymm4, %ymm5, %ymm6\{%k7\}	\# AVX512IFMA \+ AVX512VL
-[ 	]*41[ 	]+\?\?\?\? 62F2FD49 		vgatherpf0dpd 23\(%ebp,%ymm7,8\)\{%k1\}	\# AVX512PF
-[ 	]*41[ 	]+C68CFD17 
-[ 	]*41[ 	]+000000
-[ 	]*42[ 	]+\?\?\?\? 62F2554F 		vpermb %zmm4, %zmm5, %zmm6\{%k7\}		\# AVX512VBMI
-[ 	]*42[ 	]+8DF4
-[ 	]*43[ 	]+vpermb %xmm4, %xmm5, %xmm6\{%k7\}		\# AVX512VBMI \+ AVX512VL
-[ 	]*44[ 	]+vpermb %ymm4, %ymm5, %ymm6\{%k7\}		\# AVX512VBMI \+ AVX512VL
-[ 	]*45[ 	]+
-[ 	]*46[ 	]+\?\?\?\? C4E2791C 		vpabsb %xmm5, %xmm6
-[ 	]*46[ 	]+F5
-[ 	]*47[ 	]+\?\?\?\? C4E27D1C 		vpabsb %ymm5, %ymm6
-[ 	]*47[ 	]+F5
-[ 	]*48[ 	]+\?\?\?\? C5D158F4 		vaddpd %xmm4, %xmm5, %xmm6
-[ 	]*49[ 	]+\?\?\?\? C5D558F4 		vaddpd %ymm4, %ymm5, %ymm6
-[ 	]*50[ 	]+\?\?\?\? 660F381C 		pabsb %xmm5, %xmm6
-[ 	]*50[ 	]+F5
-[ 	]*51[ 	]+\?\?\?\? 660F58F4 		addpd %xmm4, %xmm6
-[ 	]*52[ 	]+
+.*:8: Error: .*bad register name `%zmm.*
+.*:11: Error: .*bad register name `%zmm.*
+.*:14: Error: .*bad register name `%zmm.*
+.*:17: Error: .*bad register name `%zmm.*
+.*:18: Error: .*bad register name `%zmm.*
+.*:21: Error: .*bad register name `%zmm.*
+.*:24: Error: .*unsupported instruction.*
+.*:25: Error: .*bad register name `%zmm.*
+.*:8: Error: .*bad register name `%zmm.*
+.*:10: Error: .*bad register name `%ymm.*
+.*:11: Error: .*bad register name `%zmm.*
+.*:13: Error: .*bad register name `%ymm.*
+.*:14: Error: .*bad register name `%zmm.*
+.*:16: Error: .*bad register name `%ymm.*
+.*:17: Error: .*bad register name `%zmm.*
+.*:18: Error: .*bad register name `%zmm.*
+.*:20: Error: .*bad register name `%ymm.*
+.*:21: Error: .*bad register name `%zmm.*
+.*:23: Error: .*bad register name `%ymm.*
+.*:24: Error: .*
+.*:25: Error: .*bad register name `%zmm.*
+.*:27: Error: .*bad register name `%ymm.*
+.*:9: Error: .*unsupported masking.*
+.*:10: Error: .*unsupported masking.*
+.*:12: Error: .*unsupported instruction.*
+.*:13: Error: .*unsupported instruction.*
+.*:15: Error: .*unsupported instruction.*
+.*:16: Error: .*unsupported instruction.*
+.*:19: Error: .*unsupported masking.*
+.*:20: Error: .*unsupported masking.*
+.*:22: Error: .*unsupported instruction.*
+.*:23: Error: .*unsupported instruction.*
+.*:26: Error: .*unsupported instruction.*
+.*:27: Error: .*unsupported instruction.*
+#...
+[ 	]*[0-9]+[ 	]+\# Test \.arch \.noavx512vl
+[ 	]*[0-9]+[ 	]+\.text
+[ 	]*[0-9]+[ 	]*
+[ 	]*[0-9]+[ 	]+\.irp isa, default, .*
+#...
+[ 	]*[0-9]+[ 	]+\.endr
+#...
+[ 	]*[0-9]+[ 	]+>  \.arch default
+[ 	]*[0-9]+[ 	]+>  \.arch default
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D4F 	>  vpabsb %zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+1CF5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D0F 	>  vpabsb %xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+1CF5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D2F 	>  vpabsb %ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+1CF5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D48 	>  vpconflictd %zmm5,%zmm6
+[ 	]*[0-9]+[ 	]+C4F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D08 	>  vpconflictd %xmm5,%xmm6
+[ 	]*[0-9]+[ 	]+C4F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D28 	>  vpconflictd %ymm5,%ymm6
+[ 	]*[0-9]+[ 	]+C4F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1FD4F 	>  vcvtpd2qq \(%ecx\),%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+7B31
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1FD0F 	>  vcvtpd2qq \(%ecx\),%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+7B31
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1FD2F 	>  vcvtpd2qq \(%ecx\),%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+7B31
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D4F 	>  vexp2ps %zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+C8F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1D54F 	>  vaddpd %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+58F4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1D50F 	>  vaddpd %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+58F4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1D52F 	>  vaddpd %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+58F4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2D54F 	>  vpmadd52luq %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+B4F4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2D50F 	>  vpmadd52luq %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+B4F4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2D52F 	>  vpmadd52luq %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+B4F4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2FD49 	>  vgatherpf0dpd 23\(%ebp,%ymm7,8\)\{%k1\}
+[ 	]*[0-9]+[ 	]+C68CFD17 *
+[ 	]*[0-9]+[ 	]+000000
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2554F 	>  vpermb %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+8DF4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2550F 	>  vpermb %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+8DF4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2552F 	>  vpermb %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+8DF4
+#...
+[ 	]*[0-9]+[ 	]+>  \.arch default
+[ 	]*[0-9]+[ 	]+>  \.arch \.avx10\.1/256
+[ 	]*[0-9]+[ 	]+>  vpabsb %zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D0F 	>  vpabsb %xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+1CF5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D2F 	>  vpabsb %ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+1CF5
+[ 	]*[0-9]+[ 	]+>  vpconflictd %zmm5,%zmm6
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D08 	>  vpconflictd %xmm5,%xmm6
+[ 	]*[0-9]+[ 	]+C4F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D28 	>  vpconflictd %ymm5,%ymm6
+[ 	]*[0-9]+[ 	]+C4F5
+[ 	]*[0-9]+[ 	]+>  vcvtpd2qq \(%ecx\),%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1FD0F 	>  vcvtpd2qq \(%ecx\),%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+7B31
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1FD2F 	>  vcvtpd2qq \(%ecx\),%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+7B31
+[ 	]*[0-9]+[ 	]+>  vexp2ps %zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vaddpd %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1D50F 	>  vaddpd %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+58F4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1D52F 	>  vaddpd %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+58F4
+[ 	]*[0-9]+[ 	]+>  vpmadd52luq %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2D50F 	>  vpmadd52luq %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+B4F4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2D52F 	>  vpmadd52luq %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+B4F4
+[ 	]*[0-9]+[ 	]+>  vgatherpf0dpd 23\(%ebp,%ymm7,8\)\{%k1\}
+[ 	]*[0-9]+[ 	]+>  vpermb %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2550F 	>  vpermb %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+8DF4
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2552F 	>  vpermb %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+8DF4
+#...
+[ 	]*[0-9]+[ 	]+>  \.arch default
+[ 	]*[0-9]+[ 	]+>  \.arch \.avx10\.1/128
+[ 	]*[0-9]+[ 	]+>  vpabsb %zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D0F 	>  vpabsb %xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+1CF5
+[ 	]*[0-9]+[ 	]+>  vpabsb %ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vpconflictd %zmm5,%zmm6
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D08 	>  vpconflictd %xmm5,%xmm6
+[ 	]*[0-9]+[ 	]+C4F5
+[ 	]*[0-9]+[ 	]+>  vpconflictd %ymm5,%ymm6
+[ 	]*[0-9]+[ 	]+>  vcvtpd2qq \(%ecx\),%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1FD0F 	>  vcvtpd2qq \(%ecx\),%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+7B31
+[ 	]*[0-9]+[ 	]+>  vcvtpd2qq \(%ecx\),%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vexp2ps %zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vaddpd %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1D50F 	>  vaddpd %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+58F4
+[ 	]*[0-9]+[ 	]+>  vaddpd %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vpmadd52luq %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2D50F 	>  vpmadd52luq %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+B4F4
+[ 	]*[0-9]+[ 	]+>  vpmadd52luq %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vgatherpf0dpd 23\(%ebp,%ymm7,8\)\{%k1\}
+[ 	]*[0-9]+[ 	]+>  vpermb %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2550F 	>  vpermb %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+8DF4
+[ 	]*[0-9]+[ 	]+>  vpermb %ymm4,%ymm5,%ymm6\{%k7\}
+#...
+[ 	]*[0-9]+[ 	]+>  \.arch \.noavx512vl
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D4F 	>  vpabsb %zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+1CF5
+[ 	]*[0-9]+[ 	]+>  vpabsb %xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vpabsb %ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D48 	>  vpconflictd %zmm5,%zmm6
+[ 	]*[0-9]+[ 	]+C4F5
+[ 	]*[0-9]+[ 	]+>  vpconflictd %xmm5,%xmm6
+[ 	]*[0-9]+[ 	]+>  vpconflictd %ymm5,%ymm6
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1FD4F 	>  vcvtpd2qq \(%ecx\),%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+7B31
+[ 	]*[0-9]+[ 	]+>  vcvtpd2qq \(%ecx\),%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vcvtpd2qq \(%ecx\),%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F27D4F 	>  vexp2ps %zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+C8F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F1D54F 	>  vaddpd %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+58F4
+[ 	]*[0-9]+[ 	]+>  vaddpd %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vaddpd %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2D54F 	>  vpmadd52luq %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+B4F4
+[ 	]*[0-9]+[ 	]+>  vpmadd52luq %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vpmadd52luq %ymm4,%ymm5,%ymm6\{%k7\}
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2FD49 	>  vgatherpf0dpd 23\(%ebp,%ymm7,8\)\{%k1\}
+[ 	]*[0-9]+[ 	]+C68CFD17 *
+[ 	]*[0-9]+[ 	]+000000
+[ 	]*[0-9]+[ 	]+\?\?\?\? 62F2554F 	>  vpermb %zmm4,%zmm5,%zmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+8DF4
+[ 	]*[0-9]+[ 	]+>  vpermb %xmm4,%xmm5,%xmm6\{%k7\}
+[ 	]*[0-9]+[ 	]+>  vpermb %ymm4,%ymm5,%ymm6\{%k7\}
+#...
+[ 	]*[0-9]+[ 	]+\?\?\?\? C4E2791C 		vpabsb %xmm5, %xmm6
+[ 	]*[0-9]+[ 	]+F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? C4E27D1C 		vpabsb %ymm5, %ymm6
+[ 	]*[0-9]+[ 	]+F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? C5D158F4 		vaddpd %xmm4, %xmm5, %xmm6
+[ 	]*[0-9]+[ 	]+\?\?\?\? C5D558F4 		vaddpd %ymm4, %ymm5, %ymm6
+[ 	]*[0-9]+[ 	]+\?\?\?\? 660F381C 		pabsb %xmm5, %xmm6
+[ 	]*[0-9]+[ 	]+F5
+[ 	]*[0-9]+[ 	]+\?\?\?\? 660F58F4 		addpd %xmm4, %xmm6
+[ 	]*[0-9]+[ 	]+
 [ 	]*[1-9][0-9]*[ 	]+\.intel_syntax noprefix
 [ 	]*[1-9][0-9]*[ 	]+\?\?\?\? 62F3FD48 		vfpclasspd k0, \[eax], 0
 [ 	]*[1-9][0-9]*[ 	]+660000
--- a/gas/testsuite/gas/i386/noavx512-2.s
+++ b/gas/testsuite/gas/i386/noavx512-2.s
@@ -1,27 +1,10 @@ 
 # Test .arch .noavx512vl
 	.text
-	vpabsb %zmm5, %zmm6{%k7}		# AVX512BW
-	vpabsb %xmm5, %xmm6{%k7}		# AVX512BW + AVX512VL
-	vpabsb %ymm5, %ymm6{%k7}		# AVX512BW + AVX512VL
-	vpconflictd %zmm5, %zmm6		# AVX412CD
-	vpconflictd %xmm5, %xmm6		# AVX412CD + AVX512VL
-	vpconflictd %ymm5, %ymm6		# AVX412CD + AVX512VL
-	vcvtpd2qq (%ecx), %zmm6{%k7}		# AVX512DQ
-	vcvtpd2qq (%ecx), %xmm6{%k7}		# AVX512DQ + AVX512VL
-	vcvtpd2qq (%ecx), %ymm6{%k7}		# AVX512DQ + AVX512VL
-	vexp2ps %zmm5, %zmm6{%k7}		# AVX512ER
-	vaddpd %zmm4, %zmm5, %zmm6{%k7}		# AVX512F
-	vaddpd %xmm4, %xmm5, %xmm6{%k7}		# AVX512F + AVX512VL
-	vaddpd %ymm4, %ymm5, %ymm6{%k7}		# AVX512F + AVX512VL
-	vpmadd52luq %zmm4, %zmm5, %zmm6{%k7}	# AVX512IFMA
-	vpmadd52luq %xmm4, %xmm5, %xmm6{%k7}	# AVX512IFMA + AVX512VL
-	vpmadd52luq %ymm4, %ymm5, %ymm6{%k7}	# AVX512IFMA + AVX512VL
-	vgatherpf0dpd 23(%ebp,%ymm7,8){%k1}	# AVX512PF
-	vpermb %zmm4, %zmm5, %zmm6{%k7}		# AVX512VBMI
-	vpermb %xmm4, %xmm5, %xmm6{%k7}		# AVX512VBMI + AVX512VL
-	vpermb %ymm4, %ymm5, %ymm6{%k7}		# AVX512VBMI + AVX512VL
 
-	.arch .noavx512vl
+	.irp isa, default, .avx10.1/256, .avx10.1/128, .noavx512vl
+
+	.arch default
+	.arch \isa
 	vpabsb %zmm5, %zmm6{%k7}		# AVX512BW
 	vpabsb %xmm5, %xmm6{%k7}		# AVX512BW + AVX512VL
 	vpabsb %ymm5, %ymm6{%k7}		# AVX512BW + AVX512VL
@@ -43,6 +26,8 @@ 
 	vpermb %xmm4, %xmm5, %xmm6{%k7}		# AVX512VBMI + AVX512VL
 	vpermb %ymm4, %ymm5, %ymm6{%k7}		# AVX512VBMI + AVX512VL
 
+	.endr
+
 	vpabsb %xmm5, %xmm6
 	vpabsb %ymm5, %ymm6
 	vaddpd %xmm4, %xmm5, %xmm6
--- a/opcodes/i386-gen.c
+++ b/opcodes/i386-gen.c
@@ -464,6 +464,7 @@  static bitfield opcode_modifiers[] =
   BITFIELD (StaticRounding),
   BITFIELD (SAE),
   BITFIELD (Disp8MemShift),
+  BITFIELD (Vsz),
   BITFIELD (Optimize),
   BITFIELD (ATTMnemonic),
   BITFIELD (ATTSyntax),
--- a/opcodes/i386-opc.h
+++ b/opcodes/i386-opc.h
@@ -716,6 +716,16 @@  enum
 #define DISP8_SHIFT_VL 7
   Disp8MemShift,
 
+  /* insn has vector size restrictions, requiring a minimum of:
+	0: 128 bits.
+	1: 256 bits.
+	2: 512 bits.
+   */
+#define VSZ128                 0 /* Not to be used in templates.  */
+#define VSZ256                 1
+#define VSZ512                 2
+  Vsz,
+
   /* Support encoding optimization.  */
   Optimize,
 
@@ -776,6 +786,7 @@  typedef struct i386_opcode_modifier
   unsigned int staticrounding:1;
   unsigned int sae:1;
   unsigned int disp8memshift:3;
+  unsigned int vsz:3;
   unsigned int optimize:1;
   unsigned int attmnemonic:1;
   unsigned int attsyntax:1;
--- a/opcodes/i386-opc.tbl
+++ b/opcodes/i386-opc.tbl
@@ -131,6 +131,9 @@ 
 #define EVexLIG EVex=EVEXLIG
 #define EVexDYN EVex=EVEXDYN
 
+#define Vsz256 Vsz=VSZ256
+#define Vsz512 Vsz=VSZ512
+
 // The EVEX purpose of StaticRounding appears only together with SAE. Re-use
 // the bit to mark commutative VEX encodings where swapping the source
 // operands may allow to switch from 3-byte to 2-byte VEX encoding.
@@ -987,9 +990,9 @@  pause, 0xf390, i186, NoSuf, {}
     b:0:VexW0:Byte:AVX512DQ:66:AVX512VBMI, +
     w:1:VexW1:Word:AVX512F::AVX512BW>
 
-<dq:opc:vexw:vexw64:elem:cpu64:gpr:kpfx, +
-    d:0:VexW0::Dword::Reg32:66, +
-    q:1:VexW1:VexW1:Qword:x64:Reg64:>
+<dq:opc:vexw:vexw64:elem:cpu64:gpr:kpfx:kvsz, +
+    d:0:VexW0::Dword::Reg32:66:Vsz256, +
+    q:1:VexW1:VexW1:Qword:x64:Reg64::Vsz512>
 
 emms, 0xf77, MMX, NoSuf, {}
 // These really shouldn't allow for Reg64 (movq is the right mnemonic for
@@ -2623,22 +2626,22 @@  vpmovzxwq, 0x6634, AVX512F|AVX512VL, Mod
 
 // AVX512BW instructions.
 
-kadd<dq>, 0x<dq:kpfx>4a, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|NoSuf, { RegMask, RegMask, RegMask }
-kand<dq>, 0x<dq:kpfx>41, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|NoSuf, { RegMask, RegMask, RegMask }
-kandn<dq>, 0x<dq:kpfx>42, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|NoSuf|Optimize, { RegMask, RegMask, RegMask }
-kmov<dq>, 0x<dq:kpfx>90, AVX512BW, Modrm|Vex128|Space0F|VexW1|NoSuf, { RegMask|<dq:elem>|Unspecified|BaseIndex, RegMask }
-kmov<dq>, 0x<dq:kpfx>91, AVX512BW, Modrm|Vex128|Space0F|VexW1|NoSuf, { RegMask, <dq:elem>|Unspecified|BaseIndex }
-kmov<dq>, 0xf292, AVX512BW, D|Modrm|Vex128|Space0F|<dq:vexw64>|NoSuf, { <dq:gpr>, RegMask }
-knot<dq>, 0x<dq:kpfx>44, AVX512BW, Modrm|Vex128|Space0F|VexW1|NoSuf, { RegMask, RegMask }
-kor<dq>, 0x<dq:kpfx>45, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|NoSuf, { RegMask, RegMask, RegMask }
-kortest<dq>, 0x<dq:kpfx>98, AVX512BW, Modrm|Vex128|Space0F|VexW1|NoSuf, { RegMask, RegMask }
-ktest<dq>, 0x<dq:kpfx>99, AVX512BW, Modrm|Vex128|Space0F|VexW1|NoSuf, { RegMask, RegMask }
-kxnor<dq>, 0x<dq:kpfx>46, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|NoSuf, { RegMask, RegMask, RegMask }
-kxor<dq>, 0x<dq:kpfx>47, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|NoSuf|Optimize, { RegMask, RegMask, RegMask }
-kunpckdq, 0x4B, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|NoSuf, { RegMask, RegMask, RegMask }
-kunpckwd, 0x4B, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW0|NoSuf, { RegMask, RegMask, RegMask }
-kshiftl<dq>, 0x6633, AVX512BW, Modrm|Vex128|Space0F3A|<dq:vexw>|NoSuf, { Imm8, RegMask, RegMask }
-kshiftr<dq>, 0x6631, AVX512BW, Modrm|Vex128|Space0F3A|<dq:vexw>|NoSuf, { Imm8, RegMask, RegMask }
+kadd<dq>, 0x<dq:kpfx>4a, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|<dq:kvsz>|NoSuf, { RegMask, RegMask, RegMask }
+kand<dq>, 0x<dq:kpfx>41, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|<dq:kvsz>|NoSuf, { RegMask, RegMask, RegMask }
+kandn<dq>, 0x<dq:kpfx>42, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|<dq:kvsz>|NoSuf|Optimize, { RegMask, RegMask, RegMask }
+kmov<dq>, 0x<dq:kpfx>90, AVX512BW, Modrm|Vex128|Space0F|VexW1|<dq:kvsz>|NoSuf, { RegMask|<dq:elem>|Unspecified|BaseIndex, RegMask }
+kmov<dq>, 0x<dq:kpfx>91, AVX512BW, Modrm|Vex128|Space0F|VexW1|<dq:kvsz>|NoSuf, { RegMask, <dq:elem>|Unspecified|BaseIndex }
+kmov<dq>, 0xf292, AVX512BW, D|Modrm|Vex128|Space0F|<dq:vexw64>|<dq:kvsz>|NoSuf, { <dq:gpr>, RegMask }
+knot<dq>, 0x<dq:kpfx>44, AVX512BW, Modrm|Vex128|Space0F|VexW1|<dq:kvsz>|NoSuf, { RegMask, RegMask }
+kor<dq>, 0x<dq:kpfx>45, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|<dq:kvsz>|NoSuf, { RegMask, RegMask, RegMask }
+kortest<dq>, 0x<dq:kpfx>98, AVX512BW, Modrm|Vex128|Space0F|VexW1|<dq:kvsz>|NoSuf, { RegMask, RegMask }
+ktest<dq>, 0x<dq:kpfx>99, AVX512BW, Modrm|Vex128|Space0F|VexW1|<dq:kvsz>|NoSuf, { RegMask, RegMask }
+kxnor<dq>, 0x<dq:kpfx>46, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|<dq:kvsz>|NoSuf, { RegMask, RegMask, RegMask }
+kxor<dq>, 0x<dq:kpfx>47, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|<dq:kvsz>|NoSuf|Optimize, { RegMask, RegMask, RegMask }
+kunpckdq, 0x4B, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW1|Vsz512|NoSuf, { RegMask, RegMask, RegMask }
+kunpckwd, 0x4B, AVX512BW, Modrm|Vex256|Space0F|VexVVVV|VexW0|Vsz256|NoSuf, { RegMask, RegMask, RegMask }
+kshiftl<dq>, 0x6633, AVX512BW, Modrm|Vex128|Space0F3A|<dq:vexw>|<dq:kvsz>|NoSuf, { Imm8, RegMask, RegMask }
+kshiftr<dq>, 0x6631, AVX512BW, Modrm|Vex128|Space0F3A|<dq:vexw>|<dq:kvsz>|NoSuf, { Imm8, RegMask, RegMask }
 
 vdbpsadbw, 0x6642, AVX512BW, Modrm|Masking|Space0F3A|VexVVVV|VexW0|Disp8ShiftVL|CheckOperandSize|NoSuf, { Imm8|Imm8S, RegXMM|RegYMM|RegZMM|Unspecified|BaseIndex, RegXMM|RegYMM|RegZMM, RegXMM|RegYMM|RegZMM }