1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
// Copyright 2021-2022 Clemens Lutz
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
//! Rust-ified wrapper of NVIDIA Tools Extension (NVTX) library.
use crate::error::{ErrorKind, Result};
use nvtx_sys::{
nvtxMarkA, nvtxRangeEnd, nvtxRangeId_t, nvtxRangePop, nvtxRangePushA, nvtxRangeStartA,
};
use serde::Serialize;
use std::ffi::CStr;
use std::fmt;
const NVTX_NO_PUSH_POP_TRACKING: i32 = -2;
/// A range denoting a time span.
///
/// A range denotes an arbitrary time span in a process. Each range can contain
/// a text message or specify additional attributes.
///
/// A unique (opaque) correlation ID is created for each range. The correlation
/// ID is annotated to the nvprof profiler output.
///
/// The official documentation is available in [Nvidia's Profiler User
/// Guide](https://docs.nvidia.com/cuda/profiler-users-guide/index.html#nvtx).
///
/// # Thread-safety
///
/// A range may be ended by a different thread than it is started by. Thus, a
/// range is `Send` but not `Sync`, `Copy` or `Clone`. However, a pop must occur
/// on the same thread as the push.
///
/// # Examples
///
/// A range can be started and stopped at arbitrary points.
///
/// ```
/// # use numa_gpu::runtime::nvtx::Range;
/// # use std::ffi::CStr;
/// #
/// let message = unsafe { CStr::from_bytes_with_nul_unchecked(b"Hello World!\0") };
/// let range = Range::new(message);
/// // ...
/// let range_id = range.end();
/// println!("Range {} ended", range_id);
/// ```
///
/// Alternatively, a ranges can be nested by pushing and popping them.
///
/// ```no_run
/// # use numa_gpu::runtime::nvtx::Range;
/// # use std::ffi::CStr;
/// #
/// # // Note: push and pop return error when NVTX is not initialized. However,
/// # // linking nvtxInitialize gives an error.
/// let message = unsafe { CStr::from_bytes_with_nul_unchecked(b"Hello World!\0") };
/// Range::push(&message).unwrap();
/// // ...
/// Range::pop().unwrap();
/// ```
#[derive(Debug)]
pub struct Range {
id: nvtxRangeId_t,
}
/// A range identifier.
///
/// Each range ID is associated with a range, and can be matched to the output
/// of a profiler. For example, nvprof lists the range ID in its output.
#[derive(Clone, Copy, Debug, PartialEq, Serialize)]
pub struct RangeId(nvtxRangeId_t);
impl Range {
/// Start a process range.
///
/// The `message` is associated to this range event.
pub fn new(message: &CStr) -> Self {
let id = unsafe { nvtxRangeStartA(message.as_ptr()) };
Self { id }
}
/// End a process range.
pub fn end(self) -> RangeId {
unsafe {
nvtxRangeEnd(self.id);
}
RangeId(self.id)
}
/// Get the ID of a range.
pub fn id(&self) -> RangeId {
RangeId(self.id)
}
/// Start a new nested process range.
///
/// The `message` is associated to this range event. The push returns the
/// zero-based depth of the range being started.
pub fn push(message: &CStr) -> Result<u16> {
let depth = unsafe { nvtxRangePushA(message.as_ptr()) };
Self::push_pop_error(depth)
}
/// Stop a nested range.
///
/// The pop returns the zero-based depth of the range being ended. A runtime
/// error is returned if no range was pushed before the pop.
pub fn pop() -> Result<u16> {
let depth = unsafe { nvtxRangePop() };
Self::push_pop_error(depth)
}
fn push_pop_error(depth: i32) -> Result<u16> {
match depth {
d if d >= 0 => Ok(depth as u16),
NVTX_NO_PUSH_POP_TRACKING => {
Err(ErrorKind::RuntimeError("No NVTX push-pop tracking.".to_string()).into())
}
_ => Err(ErrorKind::RuntimeError(
"A NVTX range must be pushed before it can be popped.".to_string(),
)
.into()),
}
}
}
impl fmt::Display for RangeId {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
self.0.fmt(f)
}
}
/// Set a marker.
///
/// A marker describes an instanteneous event.
///
/// # Example
///
/// ```
/// # use numa_gpu::runtime::nvtx;
/// # use std::ffi::CStr;
/// #
/// let message = unsafe { CStr::from_bytes_with_nul_unchecked(b"Hello World!\0") };
/// nvtx::mark(&message);
/// ```
pub fn mark(message: &CStr) {
unsafe { nvtxMarkA(message.as_ptr()) }
}