Intel® ISA Extensions
Use hardware-based isolation and memory encryption to provide more code protection in your solutions.
1132 Discussions

How to zero extend packed unsigned 8-bit integers with an AVX intrinsic

Henk-Jan_L_
Beginner
‎08-29-2016 07:13 AM
1,819 Views

I need to zero extend packed unsigned 8-bit integers with an AVX intrinsic. Question: How do you generate a "vpmovzxbd xmm, xmm" instruction with intrinsics. There seems to exist something close: __m256i _mm256_cvtepu8_epi32 (__m128i a) which generates "vpmovzxbd ymm, xmm", but this is AVX2. Regards HJ!

0 Kudos

Link Copied

2 Replies
andysem
New Contributor III
‎08-30-2016 03:55 AM
1,819 Views

AVX does not provide support for 256-bit integer instructions. 128-bit vpmovzxbd can be generated with _mm_cvtepu8_epi32, but it will only extend the lower 4 bytes. If you aim to process all bytes in the vector your best approach is to use unpack intrinsics.

 

0 Kudos
Copy link
Henk-Jan_L_
Beginner
‎08-31-2016 03:20 AM
1,819 Views
Thanks, I didn't realize that _mm_cvtepu8_epi32 would generate different instructions: pmovzxbd and vpmovzxbd depending on the "Enabled Enhanced Instruction Set" compiler parameter.
0 Kudos
Copy link
Reply

Community support is provided Monday to Friday. Other contact methods are available here.

Intel does not verify all solutions, including but not limited to any file transfers that may appear in this community. Accordingly, Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

For more complete information about compiler optimizations, see our Optimization Notice.