Skip to content

Commit

Permalink
Merge pull request #14 from DenisYaroshevskiy/feb_23
Browse files Browse the repository at this point in the history 
copy mismatch things
  • Loading branch information
@DenisYaroshevskiy
DenisYaroshevskiy authored Feb 23, 2025
2 parents 1a457d7 + 554e991 commit 7cc7d71
Showing 1 changed file with 118 additions and 24 deletions.
142 changes: 118 additions & 24 deletions dyaroshev_presentations/conference_talks/from_simd_wrappers_1.html
View file
Original file line number Diff line number Diff line change
Expand Up @@ -38,49 +38,143 @@ <h2> This talk </h2>
<ul>
<li class="fragment"> TODO: link to pt1 slides </li>
<li class="fragment"> SIMD ranges are coming </li>
<li class="fragment"> We need to teach you some simd </li>
<li class="fragment"> strlen </li>
<li class="fragment"> inclusive_scan </li>
<li class="fragment"> min_element </li>
<li class="fragment"> copy_if </li>
<li class="fragment"> search </li>
<li class="fragment"> We want to knowledge share </li>
<li class="fragment"> We need to teach you simd first </li>
</ul>
</section>
<section>
<h2> How do you use SIMD today? </h2>
<h2> Let's do some algorithms </h2>
<ul>
<li class="fragment"> I DUNNO WHERE TO PUT THIS SLIDE </li>
<li class="fragment"> libc </li>
<li class="fragment"> mathematics </li>
<li class="fragment"> auto-vectorization </li>
<li class="fragment"> special important libraries (compression/hash_maps) </li>
<li class="fragment">memcmp</li>
<li class="fragment">memchr</li>
<li class="fragment">strlen</li>
<li class="fragment">reduce</li>
<li class="fragment">inclusive_scan</li>
<li class="fragment">min_element</li>
<li class="fragment">copy_if</li>
<li class="fragment">reverse</li>
</ul>
</section>
<section id="advanced_simd_algorithms_in_pictures/memcmp">
<script> imagesSlideShow("advanced_simd_algorithms_in_pictures/memcmp", 5); </script>
</section>
<section id='int64 vs memcmp'>
<script>
addBenchmarkForParameters('int64 vs memcmp',
'Memcmp results, x86',
{
name: 'mismatch',
size: 10000,
algorithm: 'selection',
type: 'char',
time: 'y',
padding: 'min',
group: 'avx2+bmi',
},
['mismatch_scalar',
'std::mismatch', 'use_int64s', 'std::memcmp'],
['std::memcmp', 'std::mismatch']);
</script>
</section>
<section>
<h2> Notes </h2>
<ul>
<li class="fragment"> std::mismatch - Nikolas Klauser </li>
</ul>
</section>
<section>
<h2> eve library </h2>
<ul>
<li class="fragment"> <a href="https://github.com/jfalcou/eve">
github
</a></li>
<li class="fragment"> Joel Falcou, Jean-Thierry Lapresté, Alexis
Aune,
Denis
Yaroshevskiy </li>
<li class="fragment"> eve::algo::mismatch </li>
</ul>
</section>
<section>
<section id='performance of the while loop'>
<section id='eve,mismatch,x86'>
<script>
addBenchmarkForParameters('performance of the while loop',
'Performance of the while loop',
addBenchmarkForParameters('eve,mismatch,x86',
'Memcmp results, x86',
{
name: 'search',
size: 1000,
name: 'mismatch',
size: 10000,
algorithm: 'selection',
type: 'char',
percentage: 100,
time: 'y',
padding: 'min',
group: 'avx2+bmi',
},
['std::strlen', 'std_::find_unguarded'],
['std::strlen']);
['mismatch_scalar',
'std::mismatch', 'use_int64s', 'std::memcmp', 'eve::algo::mismatch'],
['mismatch_scalar']);
</script>
</section>
<section>
What's SIMD?
Single Instruction Multiple Data
Wide registers
<section id='eve,mismatch,apple_m2'>
<script>
addBenchmarkForParameters('eve,mismatch,apple_m2',
'Memcmp results, Apple M2',
{
name: 'mismatch',
size: 10000,
algorithm: 'selection',
type: 'char',
time: 'y',
padding: 'min',
group: 'apple_m2',
},
['mismatch_scalar',
'std::mismatch', 'use_int64s', 'std::memcmp', 'eve::algo::mismatch'],
['mismatch_scalar']);
</script>
</section>
</section>
<section>
<h2>Vector Processor Extensions</h2>
<ul>
<li>x86</li>
<ul>
<li>128 bits: SSE2, SSE3, SSSE3, SSE4, SSE4.1, SSE4.2</li>
<li>256 bits: AVX, AVX2, XOP</li>
<li>512 bits: AVX512 and its myriad of sub-genre</li>
</ul>
<li>ARM</li>
<ul>
<li>128 bits: NEON, ASIMD</li>
<li> SVE (VLS/VLA)</li>
</ul>
<li> RVV </li>
<li>PowerPC</li>
<li>WASM</li>
</ul>
</section>
<section id='eve::mistamtch,arch'>
<script>
addBenchmarkForParameters('eve::mistamtch,arch',
'Marching',
{
name: 'mismatch',
size: 10000,
algorithm: 'selection',
type: 'char',
time: 'y',
padding: 'min',
group: 'selection',
},
['std::memcmp/avx2+bmi', 'use_int64s/avx2+bmi', 'eve::algo::mismatch/avx2+bmi', 'eve::algo::mismatch/sse2'],
offByDefault = ['std::memcmp/avx2+bmi', 'use_int64s/avx2+bmi']);
</script>
</section>
<section>
<h2> What's inside mismatch? </h2>
<ul>
<li class="fragment"> find_if_not + zip </li>
</ul>
</section>
</div>

<script src="../../dist/reveal.js"></script>
Expand Down

0 comments on commit 7cc7d71

Please sign in to comment.