Dragan Djuric
Dragan Djuric
The issue: When calling matrix multiplication in a tight loop (as is common in neural network training), the performance is hugely degraded - order(s) of magnitude. The trick is that...
Those functions are part of ATLAS's "extension" of blas. If you already have them implemented internally (code search indicates so) and if it's not a hassle, maybe it would be...
[Please support my work on Patreon](https://patreon.com/draganrocks). I'll intive you to a dedicated Discord discussion server. Can't afford to donate? Ask for a free invite.
[Please support my work on Patreon](https://patreon.com/draganrocks). I'll intive you to a dedicated [Discord](https://discordapp.com) discussion server. Can't afford to donate? Ask for a free invite.
[Please support my work on Patreon](https://patreon.com/draganrocks). I'll intive you to a dedicated Discord discussion server. Can't afford to donate? Ask for a free invite.
[Please support my work on Patreon](https://patreon.com/draganrocks). I'll intive you to a dedicated Discord discussion server. Can't afford to donate? Ask for a free invite.
[Please support my work on Patreon](https://patreon.com/draganrocks). I'll intive you to a dedicated Discord discussion server. Can't afford to donate? Ask for a free invite.
Hi Marco, It seems that [MlOpen](https://github.com/ROCmSoftwarePlatform/MIOpen) has been supporting OpenCL for some time now, and there appear to be libraries that use it more or less successfully. I know that...
Scenarios: 1. some of the `enqueueReleaseX` methods is called twice on the same objects 2. an operation that needs valid context is called after the context has been released Result:...
[RNN] The purpose of hidden state vectors (and possible bug in handling them when src_iter is NULL)
Hi, For the sake of simplicity, I'll assume that we are talking about basic RNN networks that should learn simple things such as a sequence of numbers. So, a basic...