mirror of
https://github.com/NVIDIA/nccl-tests.git
synced 2026-01-13 18:37:16 +08:00
1021260ca9
Build option DSO=1 generates libverifiable.so which can be used to reduce the combined binary size. Build option NAME_SUFFIX can be used to a add suffix to all generated binaries. e.g. NAME_SUFFIX=_mpi Added new make target: clean_intermediates
65 lines
2.4 KiB
C++
65 lines
2.4 KiB
C++
#ifndef _d41d8cd98f00b204e9800998ecf8427e
|
|
#define _d41d8cd98f00b204e9800998ecf8427e
|
|
|
|
#include <cuda_runtime.h>
|
|
|
|
#include <stdint.h>
|
|
|
|
/* Routines for launching kernels that verify reduction results. A significant
|
|
* feature of these routines is they carefully craft floating point input
|
|
* to produce exactly predictable output.
|
|
*
|
|
* int elt_ty: actually just a ncclDataType_t
|
|
*
|
|
* int red_op: mostly just a ncclRedOp_t. Since PreMulSum ops are dynamically
|
|
* created, these are encoded as the value ncclNumOps and their scalar is
|
|
* assumed to be `ncclVerifiablePremulScalar(rank_me)`
|
|
*
|
|
* uint64_t seed: arbitrary 64-bits to use in seeding the random values
|
|
*
|
|
* intptr_t elt_ix0: index of first element pointed to by elts when generating
|
|
* random values. This makes it possible to generate subsequences independently
|
|
* as well as in aggregate.
|
|
*
|
|
* int rank_n: Number of contributions into the reduction. Non-reduction
|
|
* collectives like broadcast, gather, etc will always set this to one.
|
|
*
|
|
* int rank_me: Index of this contribution
|
|
*/
|
|
|
|
// Use this as the local scalar for PreMulSum ops
|
|
template<typename T>
|
|
__host__ __device__ T ncclVerifiablePremulScalar(int rank_me) {
|
|
return T(rank_me%2 == 0 ? 1.0f : 2.0f);
|
|
}
|
|
|
|
// Enqueue kernel to generate data which is to be reduced.
|
|
cudaError_t ncclVerifiablePrepareInput(
|
|
void *elts, intptr_t elt_n, int elt_ty, int red_op, int rank_n, int rank_me,
|
|
uint64_t seed, intptr_t elt_ix0, cudaStream_t stream
|
|
);
|
|
|
|
// Enqueue kernel to generate expected results of reduction.
|
|
cudaError_t ncclVerifiablePrepareExpected(
|
|
void *elts, intptr_t elt_n, int elt_ty, int red_op, int rank_n,
|
|
uint64_t seed, intptr_t elt_ix0, cudaStream_t stream
|
|
);
|
|
|
|
// Enqueue kernel to verify reduced data matches expectation. The number of
|
|
// failed elements is written to bad_elt_n which must be in cudaHost memory.
|
|
// If `expected == nullptr` then the expected results are generated on-the-fly
|
|
// which can be costly. Thus if you plan to run the same reduction multiple
|
|
// times it is advantageous to precompute the expected values with
|
|
// ncclVerifiablePrepareExpected and pass them as `expected` here.
|
|
cudaError_t ncclVerifiableVerify(
|
|
void const *results, void const *expected, intptr_t elt_n, int elt_ty,
|
|
int red_op, int rank_n, uint64_t seed, intptr_t elt_ix0,
|
|
int64_t *bad_elt_n, cudaStream_t stream
|
|
);
|
|
|
|
#ifdef NCCL_VERIFIABLE_SELF_TEST
|
|
void ncclVerifiableLaunchSelfTest();
|
|
#endif
|
|
|
|
#endif
|