Age | Commit message (Collapse) | Author |
|
toghether.
Summary:
LNT's benchmarks produce their output into files which are then compared in
various ways.
The benchmarks output numerical scores, which can then be compared by a custom
program in the test-suite tree which does a floating-point comparison with a
given tolerance (fpcmp).
The test-suite system allows allows users to do a hash of the program output
for a subsequently speedier comparison. This is used for benchmarks that
generate a significant amount of output (currently defined as >100k), The
problem is that some benchmarks that chose this feature also chose a
floating-point comparison of these hashes, which at best is nonsense, and at
worse causes an infinite loop.
The fpcmp program is not resilient when it's input is weird like this. For a
particular case of an observed mafft failure, it caused fpcmp to go into an
infinite loop.
Reviewers: kristof.beyls, ddunbar, rengolin
Subscribers: rengolin, llvm-commits
Differential Revision: http://reviews.llvm.org/D13283
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@248898 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
Now also with special casing to work around difference of strip command line syntax between linux and Darwin platforms.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@246035 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
It break LNT on Darwin because strip does not support --remove-section.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@243481 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@243400 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@219968 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
The default value of XCORE_TARGET_NEEDS_MEMORY is 32 (MB).
It should be noted that the XCore emulator's speed is affected by the size of memory.
Also the amount of host memory required is several times XCore-image memory size.
Thus running several large xcore-images in parallel may result in thrashing!
The default of 32 will require ~300MB of host memory.
As Makefile.config.in is included multiple times (viz: include $(LEVEL)/Makefile.config),
the addition of XCORE_TARGET_FLAGS to X_TARGET_FLAGS is done in Makefile.test
RUNTIMELIMIT is set to a default of 1 hour for the XCore emulator.
Test that take around or more than 1 hour to run have their RUNTIMELIMIT extended.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@208667 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@201382 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
To allow the RUNUNDER script to be run with parameters, the
the definition of RUNSAFELY needs to quote the interpolation
of RUNUNDER.
Patch by Doug Gilmore
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@192485 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
Patch by Doug Gilmore of Imagination technologies.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@187691 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@181499 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
Because several of the reference outputs require hashed output, running
with reference outputs but without hashed program output yields anomalous
failures.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@181260 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
- Use naming convention such as ".reference_output.big-endian" or, for type
specific outputs, something like ".reference_output.big-endian.small".
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@180647 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@178560 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@177405 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
build programs with profiling instrumentation, run the program to generate
profile data and then recompile while loading the profiling data.
Patch by Alastair Murray with minor modifications from Manman to make it work
with SPEC.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@167270 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
- Exit codes already get compared for tests, and only a few tests actually
bother to set this.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@166905 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
- We don't need multiple ways to run sandboxed tools.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@166904 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@166845 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
remove llvm-ld.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@155145 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
Signed off by Eric Christopher.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@154856 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@152996 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
TimedExec.sh.
- Also eliminate NO_WATCHDOG option, and TimedExec.sh script itself.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@152995 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@152992 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
just a hack, LNT does this better.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@152989 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@152988 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@152987 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
use of reference output because there is no accompanying time info
for how long the native run took. Make such tools happy by creating
fake timing info for such tests.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@151258 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
This is not the right fix. TimedExec.sh needs to be replaced by adding
a timeout option to the timeit tool. But, this has been sitting in my
copy of the test-suite for months and I would like to be able to tell
others to use the option. Disabling TimedExec.sh is the only way to get
meaningful run times for short benchmarks.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@151237 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
causing every test in the nightly testsuite to fail. But llvm-ld is only
using libc to work out which symbols to internalize, not for linking with
it. How can libc effect which symbols are internalized? I can see two
possibilities: either libc makes use of a function defined in the bitcode
(which it clearly doesn't) or a weak function occurs both in the bitcode
and in libc (in which case using the bitcode's copy should be OK). Thus
there doesn't seem to any point in looking at libc and comparable system
libraries (eg: libstdc++) when internalizing, and this is confirmed by my
testing: not passing -lc or $(LIBS) when invoking llvm-ld doesn't cause
any failures. So a simple fix for the Debian issue would be simply not
to pass them to llvm-ld. But then the question arises: why use llvm-ld
at all rather than "opt -std-link-opts"? This patch goes all the way
and drops llvm-ld in favour of "opt -std-link-opts". It also removes
some apparently pointless Makefile rules, which I suspect exist because
llvm-ld produces two output files, X (a script) and X.bc (the bitcode).
With this llvm-ld is not used by the nightly testsuite at all. It works
for me, but since I don't know the nightly testsuite infrastructure well
I may have broken something, so please review.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@143460 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
used to override various environment variables for test execution. Useful for
tweaking things like dyld library paths, for example.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@130578 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@130143 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
I use this as follows
make -j8 build TEST=simple ...
make TEST=simple ...
This is the only reasonable way I've come up with for manual
benchmarking. I'm checking in since others have asked for the feature.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@129530 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@129529 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@128423 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
Stay within 80 cols.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@126329 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
- timeit is just a simple replacement for /usr/bin/time which prints a whopping
four digits of precision. That's right, FOUR!
- fpcmp is just stubbed out at the moment. Yes, I will be reimplementing fpcmp
in C.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@125147 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@124923 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@124606 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@124493 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
llc-beta.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@120983 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
just "exit 0".
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@120805 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@110639 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@106936 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
outputs are missing.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@106558 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@106077 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
unfortunately hit the fpcmp infinite loop when testing with Clang.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@106035 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
nightly test target. Allows using other test targets for ARM more
effectively.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@105334 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
with USE_REFERENCE_OUTPUT).
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@105236 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
Also, add USE_REFERENCE_OUTPUT and HASH_PROGRAM_OUTPUT support for SPEC.
Also, add an UPDATE_REFERENCE_OUTPUTS make variable, which copies the out-nat
outputs to the reference_output files.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@105210 91177308-0d34-0410-b5e6-96231b3b80d8
|
|
- What a wonderful language for a test harness.
git-svn-id: https://llvm.org/svn/llvm-project/test-suite/trunk@105196 91177308-0d34-0410-b5e6-96231b3b80d8
|