Wednesday, 13 December 2017

Hmmm bit too early for the UpCore

The UpCore has given me a few issues, not least of which was a reluctance to actually boot. But a request for help on the Up forums eventually yielded a solution, turn it on and off again...basically.

Removal of the battery for 30 seconds or so, and reconnect, saw it finally boot up.

But oh boy its slow. It also had no GL or GLES drivers, or GDB on board, though these were easy enough to install, and getting a buildable config was fairly painless.

However, despite that ease it has been incredibly reluctant to build the test map demo. The build refused to build with a -j4 settings despite being a quad core, hyper threading capable machine. It runs out of virtual memory, no idea what causes that.

But annoyingly, but after trying -j2 and getting a hang I went to -j1 and finally we get a build...a very slllllooooow build

But sadly not to completion, it eventually chokes on the modelmanger.cpp file, which as far as I can see isn't anything out of the ordinary. After sitting waiting for it to move on it gives an unspecified error. Cancelling the build on the PC side does not do anything to cancel on the target side, which becomes crippled as its building.

I used the same config but with a different target address for the Up^2 and it built and ran just fine, though with the known issues with shaders. As the Core is built on the same principles as the Up^2 its all a bit suspect.

As this is so new, I suspect the issues is more to do with a very early OS and certain features of the CPU not being fully activated (like multicore).

I'll put it away for a while and come back to it in a few weeks to see if there is an improvement.

No comments:

Post a Comment