Графік комітів

2004 Коміти

Автор SHA1 Повідомлення Дата
Chris Kitching 0c09bdf523 Remove unused field 2017-10-27 20:12:32 +01:00
Chris Kitching c6707ef33c Remove CUDA_EXCLUDES
An artefact from a now-defunct hack to avoid corrupting programs
2017-10-27 20:12:32 +01:00
Chris Kitching 2f376c9b25 Make unsupported actually be a bool... 2017-10-27 20:12:31 +01:00
Evgeny Mankov 9151a355c6 Merge pull request #234 from ChrisKitching/warningSpam
[HIPIFY] Do not process __fetch_builtin_* in cudaCall()
2017-10-27 21:30:42 +03:00
Evgeny Mankov a865ebfe10 Merge pull request #235 from ChrisKitching/preprocessorEnhancements
[HIPIFY] Handle unconditional preprocessor directives far better
2017-10-27 21:21:20 +03:00
Ben Sander f288f24e95 Merge pull request #198 from AlexVlx/feature_support_globals_for_module_api
Feature support globals for module api
2017-10-27 01:53:34 +02:00
Ben Sander e97f675397 Merge pull request #218 from ChrisKitching/nodiscard
Add [[nodiscard]] attribute to hipError_t in C++17 mode
2017-10-26 22:48:54 +02:00
Ben Sander 772fe865fc Merge pull request #223 from bensander/2x_bidir
Use 2X for bidir memory bandwidth calc
2017-10-26 21:49:06 +02:00
Chris Kitching 094b2b9b05 Greatly enhance handling of macros in kernel launches
All but the most contrived use of macros is now properly handled -
have a look at the new testcases this commit adds. You can have
macros in kernel calls, macros spanning chunks of your arguments,
the call, call parameters, or callee can all be macros or
partially macros.
2017-10-26 17:28:46 +01:00
Chris Kitching eff86d975b Simplify how kernel launch expressions get translated
It seems like there was a lot of machinery here that is no longer
needed now we have hipLaunchKernelGGL (which doesn't require us
to insert an extra argument into kernel functions). We no longer
need to waste cycles scanning the AST for callees.

We can literally just do "Take the callee expression, and dump
it into the first argument of hipLaunchKernelGGL()".
2017-10-26 17:28:30 +01:00
Chris Kitching fd911e1839 Deduplicate preprocessor code
There's three functions here that all do the same thing...

There was also logic that looks for numeric literals and works
backwards to find the macro name from which they are expanded.
I previously introduced code that rewrites macro references at
expand-time in the `MacroExpands` callback, so that code is no
longer doing anything useful.
2017-10-26 17:28:30 +01:00
Chris Kitching d1e26b2e7e Rewrite _all_ CUDA macro identifiers in the preprocessor
Calls to macros that were themselves CUDA API calls were often
being missed - this applies the identifier transform to macro
names at the callsites, too.
2017-10-26 17:27:56 +01:00
Chris Kitching 4a794ed8c0 Don't special-case source locations for calls in macros
The source location for a call that's inside a macro body will,
by default, point into the macro definition itself. The original
logic was causing macro invocations to be overwritten, as I
explain here:
https://github.com/ROCm-Developer-Tools/HIP/issues/207#issuecomment-337521851

The existing PPCallbacks code is correctly rewriting macro
definitions, so the practical effect of this change is that AST
rewrites on code that's expanded from macros are no-ops.

It might be a performance optimisation to put a short-circiut at
the top of the AST callbacks to abort when faced with code that
was expanded from macros.

It might yet prove wise to do absolutely everything at lex-time...
2017-10-26 17:26:37 +01:00
Chris Kitching 35a892bc77 Prefer early-return to deep nesting
A chain of 7 closing braces is never a great sign :D

In the process it became apparant that the unsupported flag
was being silently ignored, causing users to be left with cuda
API calls in their programs with no warning given. This has been
rectified for consistency.
2017-10-26 17:26:37 +01:00
Chris Kitching 2a5acac80e Do not process __fetch_builtin_* in cudaCall()
Fixes #205
2017-10-26 17:23:55 +01:00
Evgeny Mankov 3bacb69e20 Merge pull request #213 from ChrisKitching/simplify
[HIPIFY] Simplify (and accelerate) hipification of CUDA type identifiers
2017-10-26 19:17:01 +03:00
Ben Sander c0ef6d7b62 Merge pull request #232 from kknox/update-cuda-version
Update container to newer cuda driver and sdk 9.0
2017-10-25 23:45:24 +02:00
Kent Knox 160d92af20 Update container to newer cuda driver and sdk 9.0 2017-10-25 16:14:32 -05:00
Chris Kitching 59071b895e Don't use now-defunct cmake variable in lit test config 2017-10-24 20:52:51 +01:00
Chris Kitching 778d6827f9 Refactor cudaCall to prefer early return to deep nesting
Sorry for the invasive refactor, but this was making reasoning
about this function more difficult.
2017-10-24 20:38:49 +01:00
Chris Kitching 695a1eb059 Split the giant lookup table into 3 smaller ones
Instead of having a single, enormous LUT for all CUDA names, let's
have separate ones for different types of entity. We often know
that we're looking at a typename, or a function name, or a macro
name - so we can be more efficient (and resilient to name
collisions) by having smaller lookup tables for each of those
classes of entity).

Here we start that off by having three LUTs:
- Header names
- Type names
- Everything else

Future work could usefully split "everything else" into:
- enum values
- macro names
- function names
- everything else

It's worth noting that the "needs new matcher" todos I delete here
were actually resolved with the previous commit. It no longer
naively searches for things that start with "cu*" - it will find
exactly those things that are present in our lookup tables.
2017-10-24 20:38:49 +01:00
Chris Kitching a1d8340314 One matcher for type expressions to rule them all
Previously, there were different AST matchers for each
language construct that contains a type reference, and custom
logic to perform the transformation within each of those
structures.

Since the transformation in all such cases was only replacing
CUDA types with hip ones, we can instead use an AST matcher
that finds and updates the type references directly.
This simplifies the program considerably, and it won't fail
when it finds a language feature (or complicated type expression)
that nobody wrote custom logic for yet.
2017-10-24 20:38:49 +01:00
Chris Kitching 182d346356 Make control flow less insane
`while(false)` is certainly a bold choice.
2017-10-24 20:38:49 +01:00
Chris Kitching be34a017ef Use a cmake glob for collecting hipify sources
Should make breaking this monstrosity into multiple files a bit
easier...
2017-10-24 20:38:48 +01:00
Chris Kitching 4ab29113a1 Move giant lookup table into another translation unit
Also, rewrote it as a constant variable instead of a function
that imperatively fills a map. It's shorter, faster the compile,
and (depending on how badly the compiler screws it up) maybe
faster to run.

And, of course, it starts breaking up that giant .cpp file.
2017-10-24 20:38:25 +01:00
Evgeny Mankov fea0b6ef2f Merge pull request #230 from emankov/master
[HIPIFY][fix] cmake: NO_DEFAULT_PATH is strongly needed in find_package for LLVM
2017-10-24 20:44:56 +03:00
Evgeny Mankov 15c3030d95 Merge pull request #224 from ChrisKitching/tests
[HIPIFY] Make the automated tests more useful
2017-10-24 20:12:36 +03:00
Evgeny Mankov 3f4d435d31 Merge branch 'master' into tests 2017-10-24 20:03:51 +03:00
Evgeny Mankov 061fe7a192 Merge pull request #227 from ChrisKitching/clang-silly
Tweak some version numbers in clang version compatibility checks
2017-10-24 17:16:07 +03:00
Evgeny Mankov 07a6d3ec06 [HIPIFY][fix] cmake: NO_DEFAULT_PATH is strongly needed in find_package for LLVM
Otherwise LLVM will be searched in system folders.
2017-10-24 16:35:10 +03:00
Rahul Garg d638cd4fc1 Example showing globals use with module APIs 2017-10-24 18:12:25 +05:30
Maneesh Gupta 5b6c5ca55e Merge pull request #225 from emankov/master
[HIPIFY] cmake: fix standalone build
2017-10-24 17:06:51 +05:30
Evgeny Mankov 5a9d4e3435 [HIPIFY][fix] cmake: do not build hipify-clang if not asked
+ warn "hipify-clang will not be built" if HIPIFY_CLANG_LLVM_DIR is not specified.
+ fix typo in previous commit .
2017-10-24 14:16:05 +03:00
emankov 062127c5ea [HIPIFY] cmake: simplify build 2017-10-24 10:51:11 +03:00
Chris Kitching 0207a29f78 Tweak some version numbers in clang version compatibility checks
Apparently a couple of those APIs changed in clang 5, not 4.

Drat.
2017-10-24 01:45:23 +01:00
Evgeny Mankov ffd22ec756 [HIPIFY] cmake: fix standalone build 2017-10-23 21:16:13 +03:00
Rahul Garg 08c37f3296 Use 2X for bidir p2p memory bandwidth calc 2017-10-23 21:57:20 +05:30
Chris Kitching ac42fa53ce Add concurentKernels.cu to the testsuite 2017-10-23 13:39:37 +01:00
Chris Kitching 71aa7e267a Add the CUDA samples include dir to the path for tests
Means we get to easily steal CUDA examples for tests
2017-10-23 13:39:37 +01:00
Chris Kitching 33f88fe7a1 Add cudaRegister.cu lit test 2017-10-23 13:39:37 +01:00
Chris Kitching 2faf2800a0 Add square.cu to lit testsuite 2017-10-23 13:39:37 +01:00
Chris Kitching 80120aed22 Introduce a test runner script to simplify invocation
... And to use a standard, highly amusing trick for making
coloured output work.
2017-10-23 13:39:37 +01:00
Chris Kitching 711f495041 Adapt lit test for the hipLaunchKernelGGL changes from before... 2017-10-23 13:39:37 +01:00
Chris Kitching 66cc45afc5 Migrate lit test to using FileCheck, so failures are readable
It seems the test is already broken, but look how awesome the
error message is now:

/home/chris/HIP/tests/hipify-clang/axpy.cu:31:12: error: expected string not found in input
 // CHECK: hipLaunchKernel(HIP_KERNEL_NAME(axpy), dim3(1), dim3(kDataLen), 0, 0, a, device_x, device_y);
           ^
<stdin>:31:2: note: scanning from here
 //
 ^
<stdin>:33:2: note: possible intended match here
 hipLaunchKernelGGL(axpy, dim3(1), dim3(kDataLen), 0, 0, a, device_x, device_y);
 ^
2017-10-23 13:39:37 +01:00
Chris Kitching 640afd4186 Look for FileCheck for running lit tests, too
Use of grep in `lit` RUN lines is deprecated:
https://llvm.org/docs/TestingGuide.html#writing-new-regression-tests

Using grep leads to really unhelpful failure output (it literally
just says "the test failed"). FileCheck is much more helpful, and
distributed with LLVM on most distros anyway, so this extra
dependency shouldn't prove problematic.
2017-10-23 13:39:36 +01:00
Chris Kitching efa814e381 Propagate the CUDA toolkit directory into the lit tests
Allows the tests to actually run... :D
2017-10-23 13:39:36 +01:00
Chris Kitching 76ccda4205 Remove dependency on nonexistent lit target
This just... isn't a target. Maybe you were trying to do a file
dependency? But that's handled implicitly by the find_program
call anyway, so all this is doing is generating a cmake warning
2017-10-23 13:39:36 +01:00
Maneesh Gupta b792f9f507 Make elfio headers private
Change-Id: I3ba174bb46e84a75380207d93a0da6fe3703689e
2017-10-23 10:24:36 +05:30
Ben Sander 0d1ad06458 Use 2X for bidir memory bandwidth calc 2017-10-21 07:47:32 -05:00
Evgeny Mankov d1ba399eb1 Merge pull request #219 from ChrisKitching/newClang
[HIPIFY] Migrate hipify-clang to a newer version of clang
2017-10-21 11:59:46 +03:00