maesfahani / gpuocelot

Automatically exported from code.google.com/p/gpuocelot
0 stars 0 forks source link

Incorrect results in unrolled loops #20

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago
What steps will reproduce the problem?
1. Run the mri-fhd parboil benchmark with and without loop unrolling in the
inner loop.

What is the expected output? What do you see instead?
1. With unrolling, the kernel produces incorrect outputs starting with the
first ld.const compared to the same kernel without unrolling.

What version of the product are you using? On what operating system?
Ubuntu 9.04, r107

Please provide any additional information below.

This bug is also causing incorrect results in the mandelbrot 2.2 sdk
example without manual rolling of loop bodies.  We need a simple test case
to reproduce this before we can start diagnosing the problem in detail.  It
is not obvious from examining the dataflow traces from either example.

Original issue reported on code.google.com by mailsud...@gmail.com on 15 Aug 2009 at 6:57

GoogleCodeExporter commented 8 years ago

Original comment by gregory....@gatech.edu on 15 Aug 2009 at 6:57

GoogleCodeExporter commented 8 years ago
Accepted bug, started writing a test case.

Original comment by gregory....@gatech.edu on 15 Aug 2009 at 6:58

GoogleCodeExporter commented 8 years ago
Turns out that this was not an issue with Ocelot.  See the description in commit
r108, there was a bug in parboil that made the correct operation of mri-fhd 
dependent
on the initial value of an uninitialized parameter that should have been read 
from
the input data file but is actually not present in the file...

Original comment by gregory....@gatech.edu on 16 Aug 2009 at 1:15

GoogleCodeExporter commented 8 years ago
r108 adds a fix to the version of mri-fhd in ocelot.

Original comment by gregory....@gatech.edu on 16 Aug 2009 at 1:15