I received my mini cluster Kickstarter reward a few weeks again and have only finished checking all the Parallella boards.
I had been test with two of the boards, trying the different SD images available (the standard Linaro one, the Linaro-nano, Debian 7) and have found one of the boards to fail the matmul-16 test constantly. The error I get is 'ERROR: C_epiphany is different from C_host !!!" all of the other three boards pass this test no problems. I have been trying different things, putting heatsinks on both the Zyng & Epiphany16 chips, different fans etc, all without change. running the ztemp.sh file from on one of the other threads shows that on this board the Zyng Temp is 51.4 C / 124.5 F. The other boards have been working ok with higher temps than this, and no heatsink on the Zyng or Epiphany (but with a fan blowning across the boards)
Does anybody have an Idea how I could workout what is going wrong on the Epiphany chip? it will run the other example programs with no problems, but I understand that they are simple programs that don't work the chip very much. Has anybody come up with a program to run test's on a single core, as I would guess that one core is generating an incorrect result.
Has anybody else had a problem like this and worked out a fix?
I'll keep working on the problem, as I would like find out what is happening and maybe workout a work around, this is all good experience to workout ways to handle cores that my be generating invalid data.