MacOs GATK4 Haplotype Caller : AVX instruction set support needed for the accelerated AVX PairHmm

Follow

Marcela Martinez

• April 30, 2023 01:33

I am running Haplotype Caller of GATK4.0.0, into a docker container in Mac (MacBookPro, Apple M2). I got the following warning message:

PairHMM - OpenMP multi-threaded AVX-accelerated native PairHMM implementation is not supported

WARN  PairHMM - ***WARNING: Machine does not have the AVX instruction set support needed for the accelerated AVX PairHmm. Falling back to the MUCH slower LOGLESS_CACHING implementation!

I do not how to solve this since in GATK instructions, there is mentioned that AVX requirements are covered in the image.

May you help me how to solve this in the Mac docker container?

thank you

 

0

• 

  SkyWarrior

  • May 01, 2023 07:23

  Unless GKL and PairHMM requirements are updated with possible aarch64 vector instruction support I don't think you can ever use the aforementioned optimizations. AVX is strictly x86_64v3. 

  0

  Comment actions Permalink
• 

  David Roazen

  • May 12, 2023 18:56

  Hi Marcela Martinez,

  Unfortunately, SkyWarrior is correct: the accelerated version of the PairHMM used by HaplotypeCaller currently only supports x86 chips, and we do not yet have support for M1/M2 chips. HaplotypeCaller should still run, but it will be significantly slower than on an x86 (ie., Intel/AMD) CPU.

  Regards,

  David

  0

  Comment actions Permalink

Please sign in to leave a comment.