diff options
author | Andrew Stubbs <ams@codesourcery.com> | 2021-12-03 17:46:41 +0000 |
---|---|---|
committer | Andrew Stubbs <ams@codesourcery.com> | 2023-12-06 16:48:57 +0000 |
commit | 30486fab717a90dc7516722c24ef9c5ea246c350 (patch) | |
tree | 473b0117e6eb85aa3d358ef4bf9ccdc3cff5ec6b /libgomp/plugin | |
parent | 458e7c937924bbcef80eb006af0b61420dbfc1c1 (diff) |
libgomp, nvptx: low-latency memory allocator
This patch adds support for allocating low-latency ".shared" memory on
NVPTX GPU device, via the omp_low_lat_mem_space and omp_alloc. The memory
can be allocated, reallocated, and freed using a basic but fast algorithm,
is thread safe and the size of the low-latency heap can be configured using
the GOMP_NVPTX_LOWLAT_POOL environment variable.
The use of the PTX dynamic_smem_size feature means that low-latency allocator
will not work with the PTX 3.1 multilib.
For now, the omp_low_lat_mem_alloc allocator also works, but that will change
when I implement the access traits.
libgomp/ChangeLog:
* allocator.c (MEMSPACE_ALLOC): New macro.
(MEMSPACE_CALLOC): New macro.
(MEMSPACE_REALLOC): New macro.
(MEMSPACE_FREE): New macro.
(predefined_alloc_mapping): New array. Add _Static_assert to match.
(ARRAY_SIZE): New macro.
(omp_aligned_alloc): Use MEMSPACE_ALLOC.
Implement fall-backs for predefined allocators. Simplify existing
fall-backs.
(omp_free): Use MEMSPACE_FREE.
(omp_calloc): Use MEMSPACE_CALLOC. Implement fall-backs for
predefined allocators. Simplify existing fall-backs.
(omp_realloc): Use MEMSPACE_REALLOC, MEMSPACE_ALLOC, and MEMSPACE_FREE.
Implement fall-backs for predefined allocators. Simplify existing
fall-backs.
* config/nvptx/team.c (__nvptx_lowlat_pool): New asm variable.
(__nvptx_lowlat_init): New prototype.
(gomp_nvptx_main): Call __nvptx_lowlat_init.
* libgomp.texi: Update memory space table.
* plugin/plugin-nvptx.c (lowlat_pool_size): New variable.
(GOMP_OFFLOAD_init_device): Read the GOMP_NVPTX_LOWLAT_POOL envvar.
(GOMP_OFFLOAD_run): Apply lowlat_pool_size.
* basic-allocator.c: New file.
* config/nvptx/allocator.c: New file.
* testsuite/libgomp.c/omp_alloc-1.c: New test.
* testsuite/libgomp.c/omp_alloc-2.c: New test.
* testsuite/libgomp.c/omp_alloc-3.c: New test.
* testsuite/libgomp.c/omp_alloc-4.c: New test.
* testsuite/libgomp.c/omp_alloc-5.c: New test.
* testsuite/libgomp.c/omp_alloc-6.c: New test.
Co-authored-by: Kwok Cheung Yeung <kcy@codesourcery.com>
Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
Diffstat (limited to 'libgomp/plugin')
-rw-r--r-- | libgomp/plugin/plugin-nvptx.c | 23 |
1 files changed, 22 insertions, 1 deletions
diff --git a/libgomp/plugin/plugin-nvptx.c b/libgomp/plugin/plugin-nvptx.c index 0548e7e09e5..d4a254ed4f0 100644 --- a/libgomp/plugin/plugin-nvptx.c +++ b/libgomp/plugin/plugin-nvptx.c @@ -341,6 +341,11 @@ struct ptx_device static struct ptx_device **ptx_devices; +/* OpenMP kernels reserve a small amount of ".shared" space for use by + omp_alloc. The size is configured using GOMP_NVPTX_LOWLAT_POOL, but the + default is set here. */ +static unsigned lowlat_pool_size = 8 * 1024; + static inline struct nvptx_thread * nvptx_thread (void) { @@ -1219,6 +1224,22 @@ GOMP_OFFLOAD_init_device (int n) instantiated_devices++; } + const char *var_name = "GOMP_NVPTX_LOWLAT_POOL"; + const char *env_var = secure_getenv (var_name); + notify_var (var_name, env_var); + + if (env_var != NULL) + { + char *endptr; + unsigned long val = strtoul (env_var, &endptr, 10); + if (endptr == NULL || *endptr != '\0' + || errno == ERANGE || errno == EINVAL + || val > UINT_MAX) + GOMP_PLUGIN_error ("Error parsing %s", var_name); + else + lowlat_pool_size = val; + } + pthread_mutex_unlock (&ptx_dev_lock); return dev != NULL; @@ -2178,7 +2199,7 @@ GOMP_OFFLOAD_run (int ord, void *tgt_fn, void *tgt_vars, void **args) " [(teams: %u), 1, 1] [(lanes: 32), (threads: %u), 1]\n", __FUNCTION__, fn_name, teams, threads); r = CUDA_CALL_NOCHECK (cuLaunchKernel, function, teams, 1, 1, - 32, threads, 1, 0, NULL, NULL, config); + 32, threads, 1, lowlat_pool_size, NULL, NULL, config); if (r != CUDA_SUCCESS) GOMP_PLUGIN_fatal ("cuLaunchKernel error: %s", cuda_error (r)); if (reverse_offload) |