public inbox for gdb-cvs@sourceware.org
help / color / mirror / Atom feed
* [binutils-gdb] Add batching parameter to parallel_for_each
@ 2022-04-12 15:40 Tom Tromey
0 siblings, 0 replies; only message in thread
From: Tom Tromey @ 2022-04-12 15:40 UTC (permalink / raw)
To: gdb-cvs
https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=82d734f7a3b6f08813a9ad6272aa026779c88975
commit 82d734f7a3b6f08813a9ad6272aa026779c88975
Author: Tom Tromey <tom@tromey.com>
Date: Sun May 23 09:04:27 2021 -0600
Add batching parameter to parallel_for_each
parallel_for_each currently requires each thread to process at least
10 elements. However, when indexing, it's fine for a thread to handle
just a single CU. This patch parameterizes this, and updates the one
user.
Diff:
---
gdb/minsyms.c | 4 ++--
gdb/unittests/parallel-for-selftests.c | 2 +-
gdbsupport/parallel-for.h | 18 ++++++++++++------
3 files changed, 15 insertions(+), 9 deletions(-)
diff --git a/gdb/minsyms.c b/gdb/minsyms.c
index f1c28907fa6..cbd0ad22392 100644
--- a/gdb/minsyms.c
+++ b/gdb/minsyms.c
@@ -1460,8 +1460,8 @@ minimal_symbol_reader::install ()
std::vector<computed_hash_values> hash_values (mcount);
msymbols = m_objfile->per_bfd->msymbols.get ();
- gdb::parallel_for_each
- (&msymbols[0], &msymbols[mcount],
+ /* Arbitrarily require at least 10 elements in a thread. */
+ gdb::parallel_for_each (10, &msymbols[0], &msymbols[mcount],
[&] (minimal_symbol *start, minimal_symbol *end)
{
for (minimal_symbol *msym = start; msym < end; ++msym)
diff --git a/gdb/unittests/parallel-for-selftests.c b/gdb/unittests/parallel-for-selftests.c
index adfc82f82bd..e0c07e55e03 100644
--- a/gdb/unittests/parallel-for-selftests.c
+++ b/gdb/unittests/parallel-for-selftests.c
@@ -52,7 +52,7 @@ test (int n_threads)
#define NUMBER 10000
std::atomic<int> counter (0);
- gdb::parallel_for_each (0, NUMBER,
+ gdb::parallel_for_each (1, 0, NUMBER,
[&] (int start, int end)
{
counter += end - start;
diff --git a/gdbsupport/parallel-for.h b/gdbsupport/parallel-for.h
index 915814e485e..811ffd61bda 100644
--- a/gdbsupport/parallel-for.h
+++ b/gdbsupport/parallel-for.h
@@ -32,11 +32,16 @@ namespace gdb
This approach was chosen over having the callback work on single
items because it makes it simple for the caller to do
- once-per-subrange initialization and destruction. */
+ once-per-subrange initialization and destruction.
+
+ The parameter N says how batching ought to be done -- there will be
+ at least N elements processed per thread. Setting N to 0 is not
+ allowed. */
template<class RandomIt, class RangeFunction>
void
-parallel_for_each (RandomIt first, RandomIt last, RangeFunction callback)
+parallel_for_each (unsigned n, RandomIt first, RandomIt last,
+ RangeFunction callback)
{
/* So we can use a local array below. */
const size_t local_max = 16;
@@ -48,10 +53,11 @@ parallel_for_each (RandomIt first, RandomIt last, RangeFunction callback)
size_t n_elements = last - first;
if (n_threads > 1)
{
- /* Arbitrarily require that there should be at least 10 elements
- in a thread. */
- if (n_elements / n_threads < 10)
- n_threads = std::max (n_elements / 10, (size_t) 1);
+ /* Require that there should be at least N elements in a
+ thread. */
+ gdb_assert (n > 0);
+ if (n_elements / n_threads < n)
+ n_threads = std::max (n_elements / n, (size_t) 1);
size_t elts_per_thread = n_elements / n_threads;
n_actual_threads = n_threads - 1;
for (int i = 0; i < n_actual_threads; ++i)
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2022-04-12 15:40 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-04-12 15:40 [binutils-gdb] Add batching parameter to parallel_for_each Tom Tromey
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).