ethz-asl / ethzasl_ptam

Modified version of Parallel Tracking and Mapping (PTAM)
http://wiki.ros.org/ethzasl_ptam
235 stars 184 forks source link

Minor fixes #24

Closed paralin closed 11 years ago

paralin commented 11 years ago

Hi,

I had a couple of minor errors in my syntax for the ARM conditionals. The conditionals will work on /bin/bash but not /bin/sh - I have fixed the issue.

Christian

paralin commented 11 years ago

I have added a commit that improves performance across the board using some compiler flags to optimize the code. Generally -o3 is already set - I added it for a couple places where it is not, and added some math optimizations that should help to accelerate the binaries on any architecture. I have tested it on ARMv7l and i386.

Best, Christian

paralin commented 11 years ago

On the ODROID I saw a nearly 10fps increase (from 29.9 steady to 39/40 fps) with these options. It has sped everything up a bit for me - if there's a better way to add these arguments please let me know.

If you think the gvars optimization parameter is unnecessary I can always remove it.

I think the most important thing here is that first commit - fixed a couple of syntax errors :)

Best, Christian

simonlynen commented 11 years ago

Thank you for the contributions.

cfo commented 11 years ago

is it save to use the -ffinite-math-only flag? if there is an isnan() call somewhere in the code it might not be considered [1].

best regards, christian

[1] http://www.velocityreviews.com/forums/t737937-p2-isnan-or-isntnan.html

stephanweiss commented 11 years ago

the bundle adjustment can throw NANs. I removed the flag.

Thanks Stephan


From: Christian Forster [notifications@github.com] Sent: Monday, August 12, 2013 8:31 AM To: ethz-asl/ethzasl_ptam Subject: Re: [ethzasl_ptam] Minor fixes (#24)

is it save to use the -ffinite-math-only flag? if there is an isnan() call somewhere in the code it might not be considered [1].

best regards, christian

[1] http://www.velocityreviews.com/forums/t737937-p2-isnan-or-isntnan.html

— Reply to this email directly or view it on GitHubhttps://github.com/ethz-asl/ethzasl_ptam/pull/24#issuecomment-22501818.