[PATCH v1] x86: Improve memset-vec-unaligned-erms.S

H.J. Lu hjl.tools@gmail.com
Mon Jun 7 04:33:44 GMT 2021


On Sun, Jun 6, 2021 at 8:39 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>
>
>
> On Sun, Jun 6, 2021 at 11:21 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>>
>> On Sun, Jun 6, 2021 at 8:06 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>> >
>> >
>> >
>> > On Sun, Jun 6, 2021 at 10:48 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>> >>
>> >> On Sun, Jun 6, 2021 at 7:35 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>> >> >
>> >> >
>> >> >
>> >> > On Thu, May 20, 2021 at 4:45 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>> >> >>
>> >> >> On Thu, May 20, 2021 at 1:03 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>> >> >> >
>> >> >> > On Thu, May 20, 2021 at 3:40 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>> >> >> > >
>> >> >> > > On Thu, May 20, 2021 at 11:45 AM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>> >> >> > > >
>> >> >> > > > No bug. This commit makes a few small improvements to
>> >> >> > > > memset-vec-unaligned-erms.S. The changes are 1) only aligning to 64
>> >> >> > > > instead of 128. Either alignment will perform equally well in a loop
>> >> >> > > > and 128 just increases the odds of having to do an extra iteration
>> >> >> > > > which can be significant overhead for small values. 2) Align some
>> >> >> > > > targets and the loop. 3) Remove an ALU from the alignment process. 4)
>> >> >> > > > Reorder the last 4x VEC so that they are stored after the loop. 5)
>> >> >> > > > Move the condition for leq 8x VEC to before the alignment
>> >> >> > > > process. test-memset and test-wmemset are both passing.
>> >> >> > > >
>> >> >> > > > Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
>> >> >> > > > ---
>> >> >> > > > Tests where run on the following CPUs:
>> >> >> > > >
>> >> >> > > > Skylake: https://ark.intel.com/content/www/us/en/ark/products/149091/intel-core-i7-8565u-processor-8m-cache-up-to-4-60-ghz.html
>> >> >> > > >
>> >> >> > > > Icelake: https://ark.intel.com/content/www/us/en/ark/products/196597/intel-core-i7-1065g7-processor-8m-cache-up-to-3-90-ghz.html
>> >> >> > > >
>> >> >> > > > Tigerlake: https://ark.intel.com/content/www/us/en/ark/products/208921/intel-core-i7-1165g7-processor-12m-cache-up-to-4-70-ghz-with-ipu.html
>> >> >> > > >
>> >> >> > > > All times are the geometric mean of N=50. The unit of time is
>> >> >> > > > seconds.
>> >> >> > > >
>> >> >> > > > "Cur" refers to the current implementation
>> >> >> > > > "New" refers to this patches implementation
>> >> >> > > >
>> >> >> > > > Performance data attached in memset-data.pdf
>> >> >> > > >
>> >> >> > > > Some notes on the numbers:
>> >> >> > > >
>> >> >> > > > I only included numbers for the proper VEC_SIZE for the corresponding
>> >> >> > > > cpu.
>> >> >> > > >
>> >> >> > > > skl -> avx2
>> >> >> > > > icl -> evex
>> >> >> > > > tgl -> evex
>> >> >> > > >
>> >> >> > > > The changes only affect sizes > 2 * VEC_SIZE. The performance
>> >> >> > > > differences in the size <= 2 * VEC_SIZE come from changes in alignment
>> >> >> > > > after linking (i.e ENTRY aligns to 16, but performance can be affected
>> >> >> > > > by alignment % 64 or alignment % 4096) and generally affects
>> >> >> > > > throughput only, not latency (i.e with an lfence to the benchmark loop
>> >> >> > > > the deviations go away). Generally I think they can be ignored (both
>> >> >> > > > positive and negative affects).
>> >> >> > > >
>> >> >> > > > The interesting part of the data is in the medium size range [128,
>> >> >> > > > 1024] where the new implementation has a reasonable speedup. This is
>> >> >> > > > especially pronounced when the more conservative alignment saves a
>> >> >> > > > full loop iteration. The only significant exception is
>> >> >> > > > skylake-avx2-erms case for size = 416, alignment = 416 where the
>> >> >> > > > current implementation is meaningfully faster. I am unsure of the root
>> >> >> > > > cause for this. The skylake-avx2 case only performs a bit worse in
>> >> >> > > > this case which makes me think part of it is code alignment related,
>> >> >> > > > though comparative to the speedup in other size/alignment
>> >> >> > > > configurations it is still a trough.  Despite this, I still think the
>> >> >> > > > numbers are overall an improvement.
>> >> >> > > >
>> >> >> > > > As well due to aligning the loop (and possibly slightly more efficient
>> >> >> > > > DSB behavior with the replacement of addq 4 * VEC_SIZE in the loop
>> >> >> > > > with subq -4 * VEC_SIZE) in the non-erms cases there is often a slight
>> >> >> > > > improvement to the main loop for large sizes.
>> >> >> > > >
>> >> >> > > >  .../multiarch/memset-vec-unaligned-erms.S     | 50 +++++++++++--------
>> >> >> > > >  1 file changed, 28 insertions(+), 22 deletions(-)
>> >> >> > > >
>> >> >> > > > diff --git a/sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S b/sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S
>> >> >> > > > index 08cfa49bd1..ff196844a0 100644
>> >> >> > > > --- a/sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S
>> >> >> > > > +++ b/sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S
>> >> >> > > > @@ -173,17 +173,22 @@ ENTRY (MEMSET_SYMBOL (__memset, unaligned_erms))
>> >> >> > > >         VMOVU   %VEC(0), (%rdi)
>> >> >> > > >         VZEROUPPER_RETURN
>> >> >> > > >
>> >> >> > > > +       .p2align 4
>> >> >> > > >  L(stosb_more_2x_vec):
>> >> >> > > >         cmp     __x86_rep_stosb_threshold(%rip), %RDX_LP
>> >> >> > > >         ja      L(stosb)
>> >> >> > > > +#else
>> >> >> > > > +       .p2align 4
>> >> >> > > >  #endif
>> >> >> > > >  L(more_2x_vec):
>> >> >> > > > -       cmpq  $(VEC_SIZE * 4), %rdx
>> >> >> > > > -       ja      L(loop_start)
>> >> >> > > > +       /* Stores to first 2x VEC before cmp as any path forward will
>> >> >> > > > +          require it.  */
>> >> >> > > >         VMOVU   %VEC(0), (%rdi)
>> >> >> > > >         VMOVU   %VEC(0), VEC_SIZE(%rdi)
>> >> >> > > > -       VMOVU   %VEC(0), -VEC_SIZE(%rdi,%rdx)
>> >> >> > > > +       cmpq    $(VEC_SIZE * 4), %rdx
>> >> >> > > > +       ja      L(loop_start)
>> >> >> > > >         VMOVU   %VEC(0), -(VEC_SIZE * 2)(%rdi,%rdx)
>> >> >> > > > +       VMOVU   %VEC(0), -VEC_SIZE(%rdi,%rdx)
>> >> >> > > >  L(return):
>> >> >> > > >  #if VEC_SIZE > 16
>> >> >> > > >         ZERO_UPPER_VEC_REGISTERS_RETURN
>> >> >> > > > @@ -192,28 +197,29 @@ L(return):
>> >> >> > > >  #endif
>> >> >> > > >
>> >> >> > > >  L(loop_start):
>> >> >> > > > -       leaq    (VEC_SIZE * 4)(%rdi), %rcx
>> >> >> > > > -       VMOVU   %VEC(0), (%rdi)
>> >> >> > > > -       andq    $-(VEC_SIZE * 4), %rcx
>> >> >> > > > -       VMOVU   %VEC(0), -VEC_SIZE(%rdi,%rdx)
>> >> >> > > > -       VMOVU   %VEC(0), VEC_SIZE(%rdi)
>> >> >> > > > -       VMOVU   %VEC(0), -(VEC_SIZE * 2)(%rdi,%rdx)
>> >> >> > > >         VMOVU   %VEC(0), (VEC_SIZE * 2)(%rdi)
>> >> >> > > > -       VMOVU   %VEC(0), -(VEC_SIZE * 3)(%rdi,%rdx)
>> >> >> > > >         VMOVU   %VEC(0), (VEC_SIZE * 3)(%rdi)
>> >> >> > > > -       VMOVU   %VEC(0), -(VEC_SIZE * 4)(%rdi,%rdx)
>> >> >> > > > -       addq    %rdi, %rdx
>> >> >> > > > -       andq    $-(VEC_SIZE * 4), %rdx
>> >> >> > > > -       cmpq    %rdx, %rcx
>> >> >> > > > -       je      L(return)
>> >> >> > > > +       cmpq    $(VEC_SIZE * 8), %rdx
>> >> >> > > > +       jbe     L(loop_end)
>> >> >> > > > +       andq    $-(VEC_SIZE * 2), %rdi
>> >> >> > > > +       subq    $-(VEC_SIZE * 4), %rdi
>> >> >> > > > +       leaq    -(VEC_SIZE * 4)(%rax, %rdx), %rcx
>> >> >
>> >> > If this overflows loop will exit first iteration. Is that an issue?
>> >>
>> >> Please do following:
>> >>
>> >> 1.  Update memset assembly codes with
>> >>         check conditions for underwrite/overwrite.
>> >>         if true then branch to the HLT instruction.
>> >
>> > codes? Just this file or others as well?
>>
>> All string/memory functions should work for all valid inputs.
>>
>> >>
>> >> 2.  Update and run memset test to verify the test coverage for the condition.
>> >
>> > What is the desired result? Segfault?
>>
>> For invalid inputs, anything can happen, including segfault.
>>
>> >>
>> >> 3.  Update memset assembly codes to cover such conditions.
>> >
>> > Memcmp as well? What about wmemset? Currently (and with previous versions as well)
>> > a value like 2^63 would cause similar behavior after the `salq $2, %rdx`
>>
>> Is this condition a valid input?  If not, there is nothing to do.
>
>
> AFAIK and value length [0, SIZE_MAX] for either is a valid input for any string/memory function
> from the perspective of the standard. But I don't know if it has any qualifiers.
>
> As well I don't know what is meant to happen if the machine/OS is unable to perform the necessary operations.
> Normally you would see segfault. The difference from the 3 commits below is essentially just that it won't
> segfault.
>
> That is already case, and has been for a while, certain inputs for many of the wcsmbs will have roughly the
> same behavior from `salq $2, %rdx`. For example wmemset(ptr, 0, 2^62 + 1) will currently set 1 wchar then
> return.

Please construct a testcase which will return normally instead of
segfault if not fixed.  So on
x86-64, the expected behavior should be segfault.

>>
>>
>> >>
>> >>
>> >> Thanks.
>> >>
>> >> > If so the following commits from me have the same bug:
>> >> > https://sourceware.org/git/?p=glibc.git;a=commit;h=6abf27980a947f9b6e514d6b33b83059d39566ae
>> >> > https://sourceware.org/git/?p=glibc.git;a=commit;h=4ad473e97acdc5f6d811755b67c09f2128a644ce
>> >> > https://sourceware.org/git/?p=glibc.git;a=commit;h=16d12015c57701b08d7bbed6ec536641bcafb428
>> >> >>
>> >> >>
>> >> >>
>> >> >> > > > +       .p2align 4
>> >> >> > > >  L(loop):
>> >> >> > > > -       VMOVA   %VEC(0), (%rcx)
>> >> >> > > > -       VMOVA   %VEC(0), VEC_SIZE(%rcx)
>> >> >> > > > -       VMOVA   %VEC(0), (VEC_SIZE * 2)(%rcx)
>> >> >> > > > -       VMOVA   %VEC(0), (VEC_SIZE * 3)(%rcx)
>> >> >> > > > -       addq    $(VEC_SIZE * 4), %rcx
>> >> >> > > > -       cmpq    %rcx, %rdx
>> >> >> > > > -       jne     L(loop)
>> >> >> > > > +       VMOVA   %VEC(0), (%rdi)
>> >> >> > > > +       VMOVA   %VEC(0), VEC_SIZE(%rdi)
>> >> >> > > > +       VMOVA   %VEC(0), (VEC_SIZE * 2)(%rdi)
>> >> >> > > > +       VMOVA   %VEC(0), (VEC_SIZE * 3)(%rdi)
>> >> >> > > > +       subq    $-(VEC_SIZE * 4), %rdi
>> >> >> > > > +       cmpq    %rcx, %rdi
>> >> >> > > > +       jb      L(loop)
>> >> >
>> >> > Issue because %rdi will not be below %rcx here.
>> >> >>
>> >> >> > > > +L(loop_end):
>> >> >> > > > +       /* NB: rax is set as ptr in MEMSET_VDUP_TO_VEC0_AND_SET_RETURN.
>> >> >> > > > +              rdx as length is also unchanged.  */
>> >> >> > > > +       VMOVU   %VEC(0), -(VEC_SIZE * 4)(%rax, %rdx)
>> >> >> > > > +       VMOVU   %VEC(0), -(VEC_SIZE * 3)(%rax, %rdx)
>> >> >> > > > +       VMOVU   %VEC(0), -(VEC_SIZE * 2)(%rax, %rdx)
>> >> >> > > > +       VMOVU   %VEC(0), -VEC_SIZE(%rax, %rdx)
>> >> >> > > >         VZEROUPPER_SHORT_RETURN
>> >> >> > > >
>> >> >> > > >         .p2align 4
>> >> >> > > > --
>> >> >> > > > 2.25.1
>> >> >> > > >
>> >> >> > >
>> >> >> > > LGTM.
>> >> >> >
>> >> >> > Awesome!
>> >> >> >
>> >> >> > For future patches do you prefer performance numbers like this or
>> >> >> > raw text? Or some other alternative?
>> >> >>
>> >> >> The current data format is fine.   Thanks.
>> >> >>
>> >> >> > >
>> >> >> > > Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
>> >> >> > >
>> >> >> > > Thanks.
>> >> >> > >
>> >> >> > > --
>> >> >> > > H.J.
>> >> >>
>> >> >>
>> >> >>
>> >> >> --
>> >> >> H.J.
>> >>
>> >>
>> >>
>> >> --
>> >> H.J.
>>
>>
>>
>> --
>> H.J.



-- 
H.J.


More information about the Libc-alpha mailing list