support for M1 chip (ARM)

[iqtree]

support for M1 chip (ARM)

[jianshu93]

Any update for this?

Thanks,

Jianshu

[bqminh]

Do you mean the new M1 chip in MacBook? Tried this version: https://github.com/iqtree/iqtree2/releases/tag/v2.2-beta <https://github.com/iqtree/iqtree2/releases/tag/v2.2-beta>. It worked for us. Minh

…

On 17 Jan 2022, at 5:20 am, Jianshu_Zhao ***@***.***> wrote: Any update for this? Thanks, Jianshu — Reply to this email directly, view it on GitHub <#42 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADRTPUY7L7QWJ4YI5XKKS73UWMD4VANCNFSM5HRTKYUQ>. Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you are subscribed to this thread.

[jianshu93]

I have the following error when building from source:

[ 1%] Building CXX object CMakeFiles/kernelavx.dir/tree/phylotreeavx.cpp.o
[ 1%] Building CXX object CMakeFiles/kernelfma.dir/tree/phylokernelfma.cpp.o
[ 1%] Building CXX object CMakeFiles/kernelsse.dir/tree/phylokernelsse.cpp.o
[ 1%] Building CXX object terraphast/CMakeFiles/terraphast.dir/lib/advanced.cpp.o
c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7'
c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7'
c++: error: unrecognized command-line option '-mavx'
c++: error: unrecognized command-line option '-mavx'
c++: error: unrecognized command-line option '-mfma'
make[2]: *** [CMakeFiles/kernelavx.dir/tree/phylotreeavx.cpp.o] Error 1
make[2]: *** [CMakeFiles/kernelfma.dir/tree/phylokernelfma.cpp.o] Error 1
make[1]: *** [CMakeFiles/kernelavx.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....
make[1]: *** [CMakeFiles/kernelfma.dir/all] Error 2
c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7'
c++: error: unrecognized command-line option '-msse3'
make[2]: *** [CMakeFiles/kernelsse.dir/tree/phylokernelsse.cpp.o] Error 1
make[1]: *** [CMakeFiles/kernelsse.dir/all] Error 2
c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7'
c++: error: unrecognized command-line option '-msse3'
make[2]: *** [terraphast/CMakeFiles/terraphast.dir/lib/advanced.cpp.o] Error 1
make[2]: *** Waiting for unfinished jobs....
[ 2%] Building CXX object terraphast/CMakeFiles/terraphast.dir/lib/bigint.cpp.o
[ 2%] Building CXX object terraphast/CMakeFiles/terraphast.dir/lib/bipartitions.cpp.o
[ 2%] Building CXX object terraphast/CMakeFiles/terraphast.dir/lib/bitmatrix.cpp.o
c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7'
c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7'
c++: error: unrecognized command-line option '-msse3'
c++: error: unrecognized command-line option '-msse3'
make[2]: *** [terraphast/CMakeFiles/terraphast.dir/lib/bipartitions.cpp.o] Error 1
make[2]: *** [terraphast/CMakeFiles/terraphast.dir/lib/bigint.cpp.o] Error 1
c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7'
c++: error: unrecognized command-line option '-msse3'
make[2]: *** [terraphast/CMakeFiles/terraphast.dir/lib/bitmatrix.cpp.o] Error 1
make[1]: *** [terraphast/CMakeFiles/terraphast.dir/all] Error 2
make: *** [all] Error 2

Apparently it does not works for ARM64 cpu structure (not through rosetta2).

Any possibilities to support in the future?

Thanks,

Jianshu

[trongnhanuit]

Dear Jianshu_Zhao, I'm working on that but have not yet pushed the latest codes to Github. Currently, I successfully compiled it on my Mac M1 but IQ-TREE crashes when running phylogenetic inference. It requires a little bit more time to fix. I'll notify you once it's done. Cheers, Nhan

…

On Mon, Jan 17, 2022 at 12:20 PM Jianshu_Zhao ***@***.***> wrote: I have the following error when building from source: [ 1%] Building CXX object CMakeFiles/kernelavx.dir/tree/phylotreeavx.cpp.o [ 1%] Building CXX object CMakeFiles/kernelfma.dir/tree/phylokernelfma.cpp.o [ 1%] Building CXX object CMakeFiles/kernelsse.dir/tree/phylokernelsse.cpp.o [ 1%] Building CXX object terraphast/CMakeFiles/terraphast.dir/lib/advanced.cpp.o c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7' c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7' c++: error: unrecognized command-line option '-mavx' c++: error: unrecognized command-line option '-mavx' c++: error: unrecognized command-line option '-mfma' make[2]: *** [CMakeFiles/kernelavx.dir/tree/phylotreeavx.cpp.o] Error 1 make[2]: *** [CMakeFiles/kernelfma.dir/tree/phylokernelfma.cpp.o] Error 1 make[1]: *** [CMakeFiles/kernelavx.dir/all] Error 2 make[1]: *** Waiting for unfinished jobs.... make[1]: *** [CMakeFiles/kernelfma.dir/all] Error 2 c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7' c++: error: unrecognized command-line option '-msse3' make[2]: *** [CMakeFiles/kernelsse.dir/tree/phylokernelsse.cpp.o] Error 1 make[1]: *** [CMakeFiles/kernelsse.dir/all] Error 2 c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7' c++: error: unrecognized command-line option '-msse3' make[2]: *** [terraphast/CMakeFiles/terraphast.dir/lib/advanced.cpp.o] Error 1 make[2]: *** Waiting for unfinished jobs.... [ 2%] Building CXX object terraphast/CMakeFiles/terraphast.dir/lib/bigint.cpp.o [ 2%] Building CXX object terraphast/CMakeFiles/terraphast.dir/lib/bipartitions.cpp.o [ 2%] Building CXX object terraphast/CMakeFiles/terraphast.dir/lib/bitmatrix.cpp.o c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7' c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7' c++: error: unrecognized command-line option '-msse3' c++: error: unrecognized command-line option '-msse3' make[2]: *** [terraphast/CMakeFiles/terraphast.dir/lib/bipartitions.cpp.o] Error 1 make[2]: *** [terraphast/CMakeFiles/terraphast.dir/lib/bigint.cpp.o] Error 1 c++: error: unrecognized command-line option '--target=x86_64-apple-macos10.7' c++: error: unrecognized command-line option '-msse3' make[2]: *** [terraphast/CMakeFiles/terraphast.dir/lib/bitmatrix.cpp.o] Error 1 make[1]: *** [terraphast/CMakeFiles/terraphast.dir/all] Error 2 make: *** [all] Error 2 Apparently it does not works for ARM64 cpu structure (not through rosetta2). Any possibilities to support in the future? Thanks, Jianshu — Reply to this email directly, view it on GitHub <#42 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABZPMLDGDJFIGP6YKDNTXBLUWNVEJANCNFSM5HRTKYUQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- *Nhan Ly-Trong (*MSc.) University of Information Technology - Vietnam National University-HoChiMinh City. *Phone:* (+84) 383.205.103 *Skype:* trongnhanuit

[jianshu93]

many thanks！

[trongnhanuit]

Dear Jianshu_Zhao, I've successfully built IQ-TREE, run a few simple tests, and pushed my code to Github. We can now run IQ-TREE on Mac M1 thanks to joshlvmh <https://github.com/joshlvmh> for publishing IQ-TREE : ARM NEON Implementation <https://github.com/joshlvmh/iqtree_arm_neon>. I updated my codes based on his work. However, this version on Mac M1 is a little bit slower than on an Intel chipset. Thus, further optimization is needed to utilize the power of the M1 chipset. Please pull the latest codes and feel free to let me know if you found any bugs/issues. Thank you! Cheers, Nhan

…

On Mon, Jan 17, 2022 at 12:47 PM Jianshu_Zhao ***@***.***> wrote: many thanks！ — Reply to this email directly, view it on GitHub <#42 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABZPMLDTSLZU5M2CVEICPJ3UWNYK3ANCNFSM5HRTKYUQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you commented.Message ID: ***@***.***>

-- *Nhan Ly-Trong (*MSc.) University of Information Technology - Vietnam National University-HoChiMinh City. *Phone:* (+84) 383.205.103 *Skype:* trongnhanuit

[jianshu93]

Hello Nhan,

I have the follow error of nt sequences:

IQ-TREE multicore version 2.2.0-beta COVID-edition for Mac OS X 64-bit built Jan 18 2022 built Jan 18 2022
Developed by Bui Quang Minh, James Barbetti, Nguyen Lam Tung,
Olga Chernomor, Heiko Schmidt, Dominik Schrempf, Michael Woodhams, Ly Trong Nhan.

Host: Jianshus-MBP-2.attlocal.net (SSE4.2, 16 GB RAM)
Command: iqtree2 -s 16S_sequences_muscle5.fasta -m MFP -T AUTO
Seed: 596148 (Using SPRNG - Scalable Parallel Random Number Generator)
Time: Tue Jan 18 10:49:08 2022
OMP: Info #271: omp_set_nested routine deprecated, please use omp_set_max_active_levels instead.
Kernel: SSE2 - auto-detect threads (10 CPU cores detected)

Reading alignment file 16S_sequences_muscle5.fasta ... Fasta format detected
Reading fasta file: done in 0.000364065 secs using 77.18% CPU
Alignment most likely contains DNA/RNA sequences
Constructing alignment: done in 0.00183916 secs using 315.3% CPU
Alignment has 10 sequences with 1538 columns, 121 distinct patterns
89 parsimony-informative, 66 singleton sites, 1383 constant sites
Gap/Ambiguity Composition p-value
Analyzing sequences: done in 6.79493e-05 secs using 331.1% CPU
1 3 0.20% passed 97.78%
2 8 0.20% passed 99.79%
3 4 0.20% passed 95.80%
4 10 0.20% passed 99.98%
5 7 0.20% passed 99.96%
6 5 0.59% passed 99.88%
7 9 0.00% passed 98.71%
8 6 0.20% passed 99.98%
9 1 0.07% passed 98.17%
10 2 0.20% passed 99.99%
**** TOTAL 0.20% 0 sequences failed composition chi2 test (p-value<5%; df=3)
Checking for duplicate sequences: done in 0.000381947 secs using 181.4% CPU

Create initial parsimony tree by phylogenetic likelihood library (PLL)... 0.001 seconds
Measuring multi-threading efficiency up to 10 CPU cores
Rate parameters: A-C: 1.00000 A-G: 1.00000 A-T: 1.00000 C-G: 1.00000 C-T: 1.00000 G-T: 1.00000
Base frequencies: A: 0.250 C: 0.228 G: 0.318 T: 0.204
Proportion of invariable sites: 0.450
Gamma shape alpha: 1.000
ERROR: Numerical underflow (lh-derivative). Run again with the safe likelihood kernel via -safe option

16S_sequences_muscle5.fasta.zip

Any idea?

Thanks,

Jianshu

[trongnhanuit]

Hi Jianshu, It's a known issue reported at #48 <#48> but it has not yet been fixed. You can temporarily rerun the inference (adding *-redo* option to your command) or try with *-safe *option. Thanks. Cheers, Nhan

…

On Wed, Jan 19, 2022 at 2:52 AM Jianshu_Zhao ***@***.***> wrote: Hello Nhan, I have the follow error of nt sequences: IQ-TREE multicore version 2.2.0-beta COVID-edition for Mac OS X 64-bit built Jan 18 2022 built Jan 18 2022 Developed by Bui Quang Minh, James Barbetti, Nguyen Lam Tung, Olga Chernomor, Heiko Schmidt, Dominik Schrempf, Michael Woodhams, Ly Trong Nhan. Host: Jianshus-MBP-2.attlocal.net (SSE4.2, 16 GB RAM) Command: iqtree2 -s 16S_sequences_muscle5.fasta -m MFP -T AUTO Seed: 596148 (Using SPRNG - Scalable Parallel Random Number Generator) Time: Tue Jan 18 10:49:08 2022 OMP: Info #271: omp_set_nested routine deprecated, please use omp_set_max_active_levels instead. Kernel: SSE2 - auto-detect threads (10 CPU cores detected) Reading alignment file 16S_sequences_muscle5.fasta ... Fasta format detected Reading fasta file: done in 0.000364065 secs using 77.18% CPU Alignment most likely contains DNA/RNA sequences Constructing alignment: done in 0.00183916 secs using 315.3% CPU Alignment has 10 sequences with 1538 columns, 121 distinct patterns 89 parsimony-informative, 66 singleton sites, 1383 constant sites Gap/Ambiguity Composition p-value Analyzing sequences: done in 6.79493e-05 secs using 331.1% CPU 1 3 0.20% passed 97.78% 2 8 0.20% passed 99.79% 3 4 0.20% passed 95.80% 4 10 0.20% passed 99.98% 5 7 0.20% passed 99.96% 6 5 0.59% passed 99.88% 7 9 0.00% passed 98.71% 8 6 0.20% passed 99.98% 9 1 0.07% passed 98.17% 10 2 0.20% passed 99.99% **** TOTAL 0.20% 0 sequences failed composition chi2 test (p-value<5%; df=3) Checking for duplicate sequences: done in 0.000381947 secs using 181.4% CPU Create initial parsimony tree by phylogenetic likelihood library (PLL)... 0.001 seconds Measuring multi-threading efficiency up to 10 CPU cores Rate parameters: A-C: 1.00000 A-G: 1.00000 A-T: 1.00000 C-G: 1.00000 C-T: 1.00000 G-T: 1.00000 Base frequencies: A: 0.250 C: 0.228 G: 0.318 T: 0.204 Proportion of invariable sites: 0.450 Gamma shape alpha: 1.000 ERROR: Numerical underflow (lh-derivative). Run again with the safe likelihood kernel via -safe option 16S_sequences_muscle5.fasta.zip <https://github.com/iqtree/iqtree2/files/7890437/16S_sequences_muscle5.fasta.zip> Any idea? Thanks, Jianshu — Reply to this email directly, view it on GitHub <#42 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABZPMLC4Y3OAIWCDJXSOLDLUWWEC3ANCNFSM5HRTKYUQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you commented.Message ID: ***@***.***>

-- *Nhan Ly-Trong (*MSc.) University of Information Technology - Vietnam National University-HoChiMinh City. *Phone:* (+84) 383.205.103 *Skype:* trongnhanuit

[jianshu93]

Many thanks！ the dataset sequences I am using are very consertive though.

Jianshu

[trongnhanuit]

I close this issue. Further optimization for the M1 chip is welcome!

[jianshu93]

Hello Nhan,

Interestingly, the same input file 16S_sequences_muscle5.fasta.zip (I attached them above), I do not have the error using the non-arm version but the newest Linux version version 2.2-beta. Is this related to the ARM specific Numerical operations? It is not the bug you mentioned I think.

Thanks,

Jianshu