Giter Site home page Giter Site logo

Comments (17)

neworderofjamie avatar neworderofjamie commented on June 15, 2024 1

from brian2genn.

neworderofjamie avatar neworderofjamie commented on June 15, 2024

Hah good idea - sorry for confusing the issue!

Unless you have any strong objections my plan is, in the short term, to add support for independantly setting double-precision time in GeNN and, in the longer term, I'll remove the transcendental function name mangling along with a lot of essentially legacy stuff meaning that you'll have much more control over precision.

from brian2genn.

tnowotny avatar tnowotny commented on June 15, 2024

I think this is a good plan. After phase one (introducing the independent precision control for t) we can adjust Brian2genn and finally get to benchmarking.
Remains the question about a version info functionality for GeNN ...

from brian2genn.

neworderofjamie avatar neworderofjamie commented on June 15, 2024

Excellent! I'll implement that ASAP so as not to hold you up. Regarding the versioning, if you want a hack you can read the PROJECT_NUMBER from doxygen/genn-doxygen.conf 😄 however we really should do it properly - would a #define work for you @mstimberg?

from brian2genn.

mstimberg avatar mstimberg commented on June 15, 2024

genn-doxygen.conf could be a stop-gap solution :) a #define wouldn't be the ideal solution for us, since we need the version before compiling anything...

from brian2genn.

neworderofjamie avatar neworderofjamie commented on June 15, 2024

I'll add a version.txt ASAP as well - I fear that any cunning attempt at reading a version.txt into the doxygen will break any stop gap hack involving parsing it so probably best not to do that...

from brian2genn.

tnowotny avatar tnowotny commented on June 15, 2024

I assume we will then have to do the logic in a away that has GeNN version < XXX if the version.txt does not exist and otherwise its content ... don't think it is possible to introduce retrospectively into earlier versions ...

from brian2genn.

neworderofjamie avatar neworderofjamie commented on June 15, 2024

@mstimberg - does genn-team/genn#206 work for you?

from brian2genn.

mstimberg avatar mstimberg commented on June 15, 2024

@neworderofjamie : yes looking good to me
@tnowotny : yes, not having it for older versions is perfectly fine

from brian2genn.

tnowotny avatar tnowotny commented on June 15, 2024

@mstimberg , once we have merged genn-team/genn#207 we should be ready to enable float32 support for brian2genn. Just needs inserting the setTimePrecision command if the GeNN used is new enough to have it.

from brian2genn.

mstimberg avatar mstimberg commented on June 15, 2024

Great, I'll take care of this tomorrow.

from brian2genn.

mstimberg avatar mstimberg commented on June 15, 2024

@tnowotny , @neworderofjamie Which version should we require for this feature? Current version.txt still says 3.1.1, but that release did not yet contain the feature. Should we require 3.1.2 or will the next release be 3.2?

from brian2genn.

denisalevi avatar denisalevi commented on June 15, 2024

When do you expect single precision to be available in the brian2genn? I guess it depends on the GeNN 3.2 release?
We will start running final benchmarks on Friday. If single precision is available by then, we would be happy to include it.

from brian2genn.

mstimberg avatar mstimberg commented on June 15, 2024

I just opened PR #72 which removes the need to wait for GeNN 3.2 -- it just raises a warning for now, we can change this into an error as soon as GeNN 3.2 is released.

from brian2genn.

tnowotny avatar tnowotny commented on June 15, 2024

being lazy I will ask and not inspect the code: What happens of one tries to use the single precision brian2genn with an older GeNN version before the setTimePrecision function was introduced? I suspect it crashes during compilation?

from brian2genn.

mstimberg avatar mstimberg commented on June 15, 2024

Yes, unfortunately it will raise a warning and then crash... Maybe it would be cleaner to say something about the GeNN version in the error message you get for the failed compilation?

from brian2genn.

tnowotny avatar tnowotny commented on June 15, 2024

on the other hand it could have crashed during compilation for many different reasons ... on second thought, I think it is fine with warning + crash as you have it for now. Once GeNN 3.2. is released we change it to an error and everything is clean.

from brian2genn.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.