# Revision history [back]

You can compile with CUDA support, however you won't be able to run the resulting binary unless your card has compute capability >= 5.3.

To confirm this you could try compiling with -DCUDA_ARCH_PTX=5.2, -DCUDA_ARCH_BIN=5.3, and seeing what the happens when you run the dnn accuracy tests.

Does anyone know what features the DNN module uses from 5.3 that were not available in 5.2?

You can compile with CUDA support, however you won't be able to run the resulting binary unless your card has compute capability >= 5.3.

To confirm this you could try compiling with -DCUDA_ARCH_PTX=5.2, -DCUDA_ARCH_BIN=5.3, and seeing what the happens when you run the dnn accuracy tests.

Does anyone know what features It looks like it relies on half precision (I thought this was CC 6.0 and above), see this post by the DNN module uses from 5.3 that were not available in 5.2?author.

You can compile with CUDA support, however you won't be able to run the resulting binary unless your card has compute capability >= 5.3.

To confirm this you could try compiling with -DCUDA_ARCH_PTX=5.2, -DCUDA_ARCH_BIN=5.3, and seeing what the happens when you run the dnn accuracy tests.

It looks like it relies on half precision (I thought this was CC 6.0 and above), see this post by the author.

You can compile with CUDA support, support by including -DCUDA_ARCH_BIN=X with X>=5.3, however you won't be able to run the resulting binary unless your card has compute capability >= 5.3.X.

It looks like it relies on half precision (I thought this was CC 6.0 and above), see this post by the author.

You can compile with CUDA support by including -DCUDA_ARCH_BIN=X with X>=5.3, however you won't be able to run the resulting DNN CUDA binary unless your card has compute capability X.

It looks like it relies on half precision (I thought this was CC 6.0 and above), see this post by the author.