Eric,
The same driver (pocl-0.8-rc7, which I fetched today from pocl-0.8
repository today) is all fine with Ubuntu/x64 build. All tests under
examples directory are clean, so as most of "make check" targets.
I was hoping the Ubuntu-ARM build is as good as the Ubuntu/x64 build.
Unfortunately, this is not the case.
Since the failure cases (example1, example2, example2a, scalarwave,
trig) are all relatively small programs, I would rather think the
overall infrastructure is fine, and there is need for some ARM-specific
tuning. That is why I was asking in my 1st post.
I can serve as a Ubuntu/ARM tester and report errors/status once there
is a major code drop into the master branch.
Or, I can start digging into the existing issues, and potentially
contribute patches.
Please let me know when the right time arrives.
Thank you
Chuck
On 7/17/2013 2:27 PM, Erik Schnetter wrote:
Chuck
We have been stabilizing the source tree and correcting errors for
about two months... I hope that pocl-0.8 can be released soon.
Unfortunately, there were a host of build problems that needed to be
addressed first. Most of these problems were either connected to
creating a release tarball, or to build problems (e.g. configuring). I
should think that, if you can build pocl-0.8 with vecmathlib on your
system, then you should be fine to go.
We have not yet folded some of the release-related changes back into
the main development tree of pocl.
I am worried about your failures; I would have hoped that ARM works
fine with the release branch. I wonder -- is this something that
should hold up the release until this is corrected?
-erik
On Wed, Jul 17, 2013 at 3:44 PM, Chuck Zhao <[email protected]
<mailto:[email protected]>> wrote:
Is this a schedule for POCL-0.8 release?
(since LLVM/CLANG-3.3 has been out for slightly over a month now).
I did a test build on my PandaBoard (ARM/Ubuntu 12.04 +
LLVM/CLANG-3.3 +
gcc/g++-4.5.3, pocl-0.8-rc7), the build is fine, but "make check"
seems
to have some issues with the basic tests.
I also verified the same driver on my X64/Ubuntu-11.10 build.
"make check" seems to be sufficiently clean.
Please let me know when pocl-0.8's code base is "stable/mature"
enough,
so that I can dive in to debug some ARM-specific issues.
Thank you
Chuck
## ----------------------- ##
## pocl 0.8rc7 test suite. ##
## ----------------------- ##
OpenCL specification tests
1: example1: dot product FAILED
(testsuite.at:39 <http://testsuite.at:39>)
2: example2: matrix transpose FAILED
(testsuite.at:44 <http://testsuite.at:44>)
3: example2a: matrix transpose (automatic locals) FAILED
(testsuite.at:50 <http://testsuite.at:50>)
Kernel runtime library
4: Kernel functions as_type ok
5: Kernel functions convert_type expected failure
(testsuite.at:76 <http://testsuite.at:76>)
6: Kernel functions min and max when the operands are of different
sign ok
7: A saturating conversion from long to uint ok
8: Kernel functions abs bitselect clz max min popcount ok
9: Kernel functions fabs signbit isfinite isinf isnan isnormal
copysign ok
10: Kernel functions abs abs_diff add_sat hadd mad_hi mad_sat mul_hi
rhadd sub_sat ok
11: Kernel functions << >> rotate expected failure
(testsuite.at:155 <http://testsuite.at:155>)
12: Trigonometric functions FAILED
(testsuite.at:162 <http://testsuite.at:162>)
Full applications
13: Scalar wave equation expected failure
(testsuite.at:198 <http://testsuite.at:198>)
Workgroup creation tests
14: Unconditional barriers (full replication) ok
15: Unconditional barriers (loops) ok
16: Unbarriered for loops (full replication) ok
17: Unbarriered for loops (loops) ok
18: Barriered for loops (full replication) ok
19: Barriered for loops (loops) ok
20: Conditional barrier (full replication) ok
21: Conditional barrier (loops) ok
22: b-loop with none of the WIs reaching the barrier (full
replication) ok
23: b-loop with none of the WIs reaching the barrier (loops) ok
24: Forcing horizontal parallelization to some outer loops (repl) ok
25: Forcing horizontal parallelization to some outer loops
(loops) ok
26: a loop with two paths to the latch (full replication) ok
27: a loop with two paths to the latch (loops) ok
28: a b-loop with two latches (full replication) ok
29: a b-loop with two latches (loops) ok
30: workgroup_sizes: work-items get wrong ids (full replication) ok
31: workgroup_sizes: work-items get wrong ids (loop) ok
------------------------------------------------------------------------------
See everything from the browser to the database with AppDynamics
Get end-to-end visibility with application monitoring from AppDynamics
Isolate bottlenecks and diagnose root cause in seconds.
Start your free trial of AppDynamics Pro today!
http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
_______________________________________________
pocl-devel mailing list
[email protected]
<mailto:[email protected]>
https://lists.sourceforge.net/lists/listinfo/pocl-devel
--
Erik Schnetter <[email protected]
<mailto:[email protected]>>
http://www.perimeterinstitute.ca/personal/eschnetter/
AIM: eschnett247, Skype: eschnett, Google Talk: [email protected]
<mailto:[email protected]>
------------------------------------------------------------------------------
See everything from the browser to the database with AppDynamics
Get end-to-end visibility with application monitoring from AppDynamics
Isolate bottlenecks and diagnose root cause in seconds.
Start your free trial of AppDynamics Pro today!
http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
_______________________________________________
pocl-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pocl-devel
------------------------------------------------------------------------------
See everything from the browser to the database with AppDynamics
Get end-to-end visibility with application monitoring from AppDynamics
Isolate bottlenecks and diagnose root cause in seconds.
Start your free trial of AppDynamics Pro today!
http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
_______________________________________________
pocl-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pocl-devel
