Different results for Mac vs Linux/Windows

[jasondbiggs]

Using the binaries downloaded from this repo I get on linux (same answer on windows):

$ ./surge-linux-v1.0 -u C7H12O2N2
C7N2O2H12  H=12 C=7 N=2 O=2  nv=11 edges=10-13 DBE=3 maxd=4 maxc=4
>Z generated 24901 -> 7811169 -> 22834059 in 1.76 sec

but on my mac I get

% ./surge-macos-v1.0 -u C7H12O2N2
C7N2O2H12  H=12 C=7 N=2 O=2  nv=11 edges=10-13 DBE=3 maxd=4 maxc=4
>Z generated 22674 -> 7378042 -> 22261950 in 1.08 sec

I haven't built surge locally, were these binaries built from the same source code? If so, do you know why there is this platform dependence?

[jasondbiggs]

For what it's worth, after building from source on my mac I now get the same result as linux

% ./surge -u C7H12O2N2
C7N2O2H12  H=12 C=7 N=2 O=2  nv=11 edges=10-13 DBE=3 maxd=4 maxc=4
>Z generated 24901 -> 7811169 -> 22834059 in 1.03 sec

[steinbeck]

Thanks, Jason. Still strange! Are you using an M1 by any chance? Maybe it was an emulation issue. The mac binary is for intel.
I have an M1 and built surge natively and get the same results as you on Windows.

[jasondbiggs]

I am using an intel mac. If I use an M1 mac then I am able to use the executable to call for --help but not for generating isomers:

% ./surge-macos-v1.0 -u C7H12O2N2
zsh: illegal hardware instruction  ./surge-macos-v1.0 -u C7H12O2N2

[steinbeck]

Hi Jason, my sincere apologies. This issue complete slipped my attention after my last reply due to local workload. Are you still interested or did you solve the problem by compiling it yourself? I can send you an M1 binary.

[jasondbiggs]

I was able to build the binaries for all the platforms we support, so this is settled for me.

But to be clear, I still see the issue if I download the "surge-macos-v1.0" located here. I am using an Intel mac, so it should be compatible with that binary I think. We recompiled for both M1 and Intel macs, and both agree on the 22834059 isomer count, but the binary on the release page here gives 22261950