File: config.cpp

package info (click to toggle)
pytorch-cuda 2.6.0%2Bdfsg-7
  • links: PTS, VCS
  • area: contrib
  • in suites: forky, sid, trixie
  • size: 161,620 kB
  • sloc: python: 1,278,832; cpp: 900,322; ansic: 82,710; asm: 7,754; java: 3,363; sh: 2,811; javascript: 2,443; makefile: 597; ruby: 195; xml: 84; objc: 68
file content (90 lines) | stat: -rw-r--r-- 2,320 bytes parent folder | download | duplicates (3)
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
#include <torch/csrc/lazy/core/config.h>

C10_DEFINE_bool(torch_lazy_ir_debug, false, "Enable lazy tensor IR debugging")

C10_DEFINE_bool(
    torch_lazy_param_aliasing,
    true,
    "Enable parameter aliasing support")

C10_DEFINE_bool(
    torch_lazy_handle_special_scalars,
    false,
    "Handle special scalars 0 and 1 differently")

C10_DEFINE_bool(
    torch_lazy_all_numbers_special_scalars,
    false,
    "Handle all numbers as special scalars")

C10_DEFINE_bool(
    torch_lazy_reuse_ir,
    false,
    "Reuse IR nodes from previous tracing when possible")

C10_DEFINE_bool(
    torch_lazy_use_thread_pool,
    false,
    "Use thread pool to schedule backend execution")

C10_DEFINE_bool(
    torch_lazy_enable_device_data_cache,
    true,
    "Enable or disable device data cache (turns cache on or off), does not change cache state")

C10_DEFINE_int(
    torch_lazy_compilation_cache_size,
    1024,
    "Size of the compilation cache")

C10_DEFINE_int(
    torch_lazy_device_data_cache_size,
    128,
    "Size of the DeviceData cache")

C10_DEFINE_int(
    torch_lazy_io_thread_pool_size,
    // TODO: measure which default value
    // will give better performance,
    // std::thread::hardware_concurrency()?
    1,
    "Size of the execution thread pool")

C10_DEFINE_int(torch_lazy_metrics_samples, 1024, "Max metrics sample size")

C10_DEFINE_int(
    torch_lazy_trim_graph_check_frequency,
    5000,
    "How often to check for whether a graph needs to be split")

C10_DEFINE_int(
    torch_lazy_trim_graph_size,
    100000,
    "The threshold (in terms of the number of nodes) for splitting a graph")

C10_DEFINE_string(
    torch_lazy_metrics_percentiles,
    "0.01:0.05:0.1:0.2:0.5:0.8:0.9:0.95:0.99",
    "Metrics percentiles to be collected, using : as the delimiter")

C10_DEFINE_int(
    torch_lazy_shape_cache_size,
    4096,
    "Set the size for the shape cache used for shape inference")

namespace torch::lazy {
std::string& getLTCForceFallback() {
  static std::string config;
  static bool _ignore = [&]() {
    char* envptr = std::getenv("LTC_FORCE_FALLBACK");
    if (envptr) {
      config = std::string(envptr);
    }
    return true;
  }();
  (void)_ignore; // avoid unused variables warning
  return config;
}

// NOLINTEND(misc-use-internal-linkage)
} // namespace torch::lazy