Conversation
…ayfire#3304) Co-authored-by: Gallagher Donovan Pryor <gallagher@arrayfire.com> Co-authored-by: Umar Arshad <umar@arrayfire.com>
The way we were formatting the backend ID was incorrect and failed when we had more than 3 backends. With the new oneAPI backend, this mechanism was failing and causing errors.
* histogram ported to oneapi. 50/62 tests pass. see below
Batch tests fail with value errors:
Histogram/1.40Bins0min100max_Batch, where TypeParam = float
Histogram/3.40Bins0min100max_Batch, where TypeParam = int
Histogram/4.40Bins0min100max_Batch, where TypeParam = unsigned int
Histogram/5.40Bins0min100max_Batch, where TypeParam = char
Histogram/6.40Bins0min100max_Batch, where TypeParam = unsigned char
Histogram/7.40Bins0min100max_Batch, where TypeParam = short
Histogram/8.40Bins0min100max_Batch, where TypeParam = unsigned short
Histogram/9.40Bins0min100max_Batch, where TypeParam = long long
Histogram/10.40Bins0min100max_Batch, where TypeParam = unsigned long long
Tests fail because reductions do not function for test harness:
Histogram.SNIPPET_hist_nominmax
Histogram.SNIPPET_histequal
GFOR, LargeBins expected to fail (getMaxMemorySize not OneAPI supported):
histogram.GFOR
histogram.LargeBins
IndexedArray expected to fail without JIT support:
LargeBins
Authored-by: Gallagher Donovan Pryor <gallagher@arrayfire.com>
The install target was copying the forge library installed on the system. This is not expected because the install command only copies the artifacts generated by the project and not libraries installed on the system. We do want system libraries to be installed when AF_INSTALL_STANDALONE is enabled. This commit addresses both of these issues.
umar456
force-pushed
the
github_actions_compiler
branch
4 times, most recently
from
36cfb57 to
e5bac9d
Compare
umar456
force-pushed
the
github_actions_compiler
branch
6 times, most recently
from
f948936 to
829ffa0
Compare
The random number generator for b8 was producing incorrect results on clang 14 due to loop unrolling. This commit addresses the underlying issue caused by ineffective indexing on the b8 RNG and updates one ireduce test to use the ASSERT_VEC_ARRAYS_EQ function
Previously the b8 RNG was only using the lest significant bits for the RNG this is probably okay but it made the CPU indexing difficult. This commit ensures that the LSB of each of the 4 integers are used instead of only the first integer
* support 64bit hamming distance on CUDA * CPU support for 64 bit __popc in hamming distance * adds hammingMatcher tests for uintll type Co-authored-by: syurkevi <stefan@arrayfire.com>
The arguments provided to OpenCL uses the C++ standard library function std::to_string(). This function uses the locale to render it's argument to a string. It is a problem when arrayfire is used in a software initialised with non "C" locale. For instance, on a French computer, to_string(1.0) will output the string "1,0000000". This string is provided to OpenCL kernels, generating a syntax error. The most portable way to fix this problem is to use a local ostringstream imbued withe "C" locale. An Other way would be to use C++17 to_chars function, as it only renders it argument with "C" locale, without impact from the application or system locale. The patch is pretty simple, it changes the toString() function to use the stringstream in src/backend/common/TemplateArg.cpp and changed the to_string calls to this toString function in types.cpp.
umar456
force-pushed
the
github_actions_compiler
branch
8 times, most recently
from
2f3b914 to
384e52b
Compare
umar456
force-pushed
the
github_actions_compiler
branch
from
384e52b to
d17c212
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Fixes: # ...
Changes to Users
Checklist