Hello everyone!
You uses Ubuntu 14.04 > arm32 > computecpp-ce-1.1.3-ubuntu.14.04-arm32.tar.gz in your guide for arm platform, but I can’t find it in https://developer.codeplay.com/computecppce/latest/download.
Is this version already unavailable?
Hello everyone!
You uses Ubuntu 14.04 > arm32 > computecpp-ce-1.1.3-ubuntu.14.04-arm32.tar.gz in your guide for arm platform, but I can’t find it in https://developer.codeplay.com/computecppce/latest/download.
Is this version already unavailable?
Hi there,
It looks like the guide is wrong but you could try using the last published 14.04 -> arm32 release.
I would like to ask what hardware you are targetting though?
We stopped producing 14.04 and therefore 32 bit Arm builds a while ago so it would be interesting to understand what you are trying to achieve.
Thanks,
Rod.
Let me correct my response. We effectively stopped supporting arm32 a while ago. If you do want to use arm32 I’d recommend that you contact us directly since we would need to provide you with specific instructions.
Thank you for the fast answer.
I would like to cross-compile Tensorflow with OpenCL supporting for the ODROID XU4 board (CPU - armv7), which has ARM Mali-T628 GPU.
We’ve just taken a look at the specifications of that board. The drivers for the GPU are quite old and would not support SPIR-V which is required for ComputeCpp and therefore the SYCL branch of TensorFlow to work. We are in the process of testing the latest drivers for the HiKey 960 board and I would be happy to let you know when we have confirmed this is working? You would need to use the HiKey 960 board though.
I was interested to check the work on ARM Mali-T628. I will consider to switch to HiKey 960 board.
Thank you for your advices.
Just as an update to this thread, the latest ComputeCpp release (v1.3.0) can be used with the public Arm drivers, see the instructions page for more information.