Add new benchmarks #899
Add new benchmarks #899
Conversation
lplewa
commented
Nov 14, 2024
lplewa
commented
- L0 benchmark will be added in separate PR (unless merge of this one will take too long ;))
- Old benchmark will be removed in separate PRs too
- Please focus more on bechmark framework - particular benchmark will be also changed in separate PRs - current set is more a demo of framework - to select good benchmark scenarios there is litte bit more research to be done.
lplewa
force-pushed
the
benchmark
branch
3 times, most recently
from
dbb4976
to
57edb2d
Compare
benchmark/benchmark.cpp
Outdated
| }; | ||
|
|
||
| #if (defined UMF_BUILD_LIBUMF_POOL_DISJOINT) | ||
|
|
benchmark/benchmark.cpp
Outdated
| return &disjoint_memory_pool_params; | ||
| } | ||
| }; | ||
|
|
benchmark/benchmark.cpp
Outdated
| } | ||
| }; | ||
|
|
||
| #if (defined UMF_BUILD_LIBUMF_POOL_DISJOINT) |
There was a problem hiding this comment.
if single def check, #ifdef should suffice
benchmark/CMakeLists.txt
Outdated
| # Under the Apache License v2.0 with LLVM Exceptions. See LICENSE.TXT. | ||
| # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception | ||
|
|
||
| # In MSVC builds, there is no way to determine the actual build type during the | ||
| # CMake configuration step. Therefore, this message is printed in all MSVC | ||
| # builds. | ||
|
|
||
| set(CMAKE_CXX_STANDARD 17) |
There was a problem hiding this comment.
if you want to enforce cxx standard setting, pls use OPTIONS_REQUIRING_CXX variable in top-level CMake
benchmark/CMakeLists.txt
Outdated
| @@ -1,10 +1,22 @@ | |||
| # Copyright (C) 2023 Intel Corporation | |||
| # Copyright (C) 2023-2024 Intel Corporation | |||
| # Under the Apache License v2.0 with LLVM Exceptions. See LICENSE.TXT. | |||
| # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception | |||
|
|
|||
| # In MSVC builds, there is no way to determine the actual build type during the | |||
| # CMake configuration step. Therefore, this message is printed in all MSVC | |||
| # builds. | |||
There was a problem hiding this comment.
heh the comment here is for the if down below, pls move it down
| # Benchmark passes if it prints "PASSED" in the output, because ubench of | ||
| # scalable pool fails if the confidence interval exceeds maximum permitted | ||
| # 2.5%. | ||
| set_tests_properties( |
There was a problem hiding this comment.
hmm, as long as we're still using ubench the line should be unchanged
There was a problem hiding this comment.
Well - i whold have to do extra work to enable it only for ubench, when we will remove it few days later
| @@ -117,7 +123,12 @@ endif() | |||
| add_umf_benchmark( | |||
| NAME ubench | |||
| SRCS ubench.c | |||
| LIBS ${LIBS_OPTIONAL} | |||
| LIBDIRS ${LIB_DIRS}) | |||
There was a problem hiding this comment.
optional libs are no longer needed...?
| } | ||
| }; | ||
|
|
||
| ALLOC_BENCHMARK_TEMPLATE_DEFINE(alloc_benchmark, stdmalloc, fix_alloc_size); |
There was a problem hiding this comment.
you could add some comments here; I'm assuming these are benchmarking scenarios...?
There was a problem hiding this comment.
Yep - i added comment to indicate it.
I have plan to make it little bit better - those scenarios are more a demo then actual final benchmarks.
benchmark/benchmark.hpp
Outdated
| size_t size; | ||
| }; | ||
|
|
||
| #define ALLOC_BENCHMARK_TEMPLATE_DEFINE(A, B, ...) \ |
There was a problem hiding this comment.
what's A and B? perhaps a link to docs, here, in the comment?
lplewa
force-pushed
the
benchmark
branch
2 times, most recently
from
a16b405
to
fa7183a
Compare
