NVIDIA/nvbench
1
0
Fork 0
mirror of https://github.com/NVIDIA/nvbench.git synced 2026-04-29 19:21:14 +00:00
Code Issues Packages Projects Releases Wiki Activity

Don't override m_check_throttling if throttling threshold is non-positive

Browse Source 
measure_cold class now directly inherits m_check_throttling from state.
This ensures that when `--jsonbin` is specified frequency data corresponding
to timing data are available to write out.
This commit is contained in:
Oleksandr Pavlyk
2026-02-20 16:34:53 -06:00
parent 731e0c2c30
commit 998ab125ce
2 changed files with 10 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
nvbench/detail/measure_cold.cu
View File

@@ -44,7 +44,7 @@ measure_cold_base::measure_cold_base(state &exec_state)
exec_state.get_stopping_criterion())} exec_state.get_stopping_criterion())}
, m_disable_blocking_kernel{exec_state.get_disable_blocking_kernel()} , m_disable_blocking_kernel{exec_state.get_disable_blocking_kernel()}
, m_run_once{exec_state.get_run_once()} , m_run_once{exec_state.get_run_once()}
, m_check_throttling(!exec_state.get_run_once() && exec_state.get_throttle_threshold() > 0.f) , m_check_throttling(!exec_state.get_run_once())
, m_min_samples{exec_state.get_min_samples()} , m_min_samples{exec_state.get_min_samples()}
, m_skip_time{exec_state.get_skip_time()} , m_skip_time{exec_state.get_skip_time()}
, m_timeout{exec_state.get_timeout()} , m_timeout{exec_state.get_timeout()}

11
nvbench/detail/measure_cold.cuh
View File

@@ -140,6 +140,13 @@ struct measure_cold_base::kernel_launch_timer
, m_check_throttling{measure.m_check_throttling} , m_check_throttling{measure.m_check_throttling}
{} {}
explicit kernel_launch_timer(measure_cold_base &measure, bool disable_blocking_kernel)
: m_measure{measure}
, m_disable_blocking_kernel{disable_blocking_kernel}
, m_run_once{measure.m_run_once}
, m_check_throttling{measure.m_check_throttling}
{}
explicit kernel_launch_timer(measure_cold_base &measure, explicit kernel_launch_timer(measure_cold_base &measure,
bool disable_blocking_kernel, bool disable_blocking_kernel,
bool run_once, bool run_once,
@@ -234,7 +241,7 @@ private:
// disable use of blocking kernel for warm-up run // disable use of blocking kernel for warm-up run
// see https://github.com/NVIDIA/nvbench/issues/240 // see https://github.com/NVIDIA/nvbench/issues/240
constexpr bool disable_blocking_kernel = true; constexpr bool disable_blocking_kernel = true;
kernel_launch_timer timer(*this, disable_blocking_kernel, m_run_once, m_check_throttling); kernel_launch_timer timer(*this, disable_blocking_kernel);
this->launch_kernel(timer); this->launch_kernel(timer);
this->check_skip_time(m_cuda_timer.get_duration()); this->check_skip_time(m_cuda_timer.get_duration());
@@ -245,7 +252,7 @@ private:
// do not use blocking kernel if benchmark is only run once, e.g., when profiling // do not use blocking kernel if benchmark is only run once, e.g., when profiling
// ref: https://github.com/NVIDIA/nvbench/issue/242 // ref: https://github.com/NVIDIA/nvbench/issue/242
const bool disable_blocking_kernel = m_run_once || m_disable_blocking_kernel; const bool disable_blocking_kernel = m_run_once || m_disable_blocking_kernel;
kernel_launch_timer timer(*this, disable_blocking_kernel, m_run_once, m_check_throttling); kernel_launch_timer timer(*this, disable_blocking_kernel);
do do
{ {
this->launch_kernel(timer); this->launch_kernel(timer);