FazBrowse GitHub Viewer | Trending |
URL:
| Home
Tools: [Download Repo ZIP]   [Original HTTPS Page]

GH-90997: Improve inline cache performance for MSVC by brandtbucher · Pull Request #96781 · python/cpython · GitHub

  1. python
  2. cpython

Repository navigation

GH-90997: Improve inline cache performance for MSVC #96781

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
brandtbucher merged 3 commits into from
Sep 15, 2022
Conversation 6 Commits 3 Checks 0 Files changed
Merged

GH-90997: Improve inline cache performance for MSVC #96781

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
eaa91ed
Use memcpy for cache reads/writes
brandtbucher Sep 12, 2022
ae8dfea
blurb add
brandtbucher Sep 12, 2022
5985a4a
Update comment
brandtbucher Sep 14, 2022
File filter

Filter by extension

Filter by extension .h  (1) .rst  (1) All 2 file types selected
Viewed files
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Unified
Split
Hide whitespace
Diff view
Unified
Split
Hide whitespace
93 changes: 19 additions & 74 deletions Include/internal/pycore_code.h
Show comments
View file
Open in desktop
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode characters
Show hidden characters
Original file line number Diff line number Diff line change
Expand Up @@ -285,110 +285,55 @@ PyAPI_FUNC(PyObject*) _Py_GetSpecializationStats(void);
#define EVAL_CALL_STAT_INC_IF_FUNCTION(name, callable) ((void)0)
#endif // !Py_STATS

// Cache values are only valid in memory, so use native endianness.
#ifdef WORDS_BIGENDIAN
// Utility functions for reading/writing 32/64-bit values in the inline caches.
// Great care should be taken to ensure that these functions remain correct and
// performant! They should compile to just "move" instructions on all supported
// compilers and platforms.

// We use memcpy to let the C compiler handle unaligned accesses and endianness
// issues for us. It also seems to produce better code than manual copying for
// most compilers (see https://blog.regehr.org/archives/959 for more info).

static inline void
write_u32(uint16_t *p, uint32_t val)
{
p[0] = (uint16_t)(val >> 16);
p[1] = (uint16_t)(val >> 0);
memcpy(p, &val, sizeof(val));
}

static inline void
write_u64(uint16_t *p, uint64_t val)
{
p[0] = (uint16_t)(val >> 48);
p[1] = (uint16_t)(val >> 32);
p[2] = (uint16_t)(val >> 16);
p[3] = (uint16_t)(val >> 0);
}

static inline uint32_t
read_u32(uint16_t *p)
{
uint32_t val = 0;
val |= (uint32_t)p[0] << 16;
val |= (uint32_t)p[1] << 0;
return val;
}

static inline uint64_t
read_u64(uint16_t *p)
{
uint64_t val = 0;
val |= (uint64_t)p[0] << 48;
val |= (uint64_t)p[1] << 32;
val |= (uint64_t)p[2] << 16;
val |= (uint64_t)p[3] << 0;
return val;
}

#else

static inline void
write_u32(uint16_t *p, uint32_t val)
{
p[0] = (uint16_t)(val >> 0);
p[1] = (uint16_t)(val >> 16);
memcpy(p, &val, sizeof(val));
}

static inline void
write_u64(uint16_t *p, uint64_t val)
write_obj(uint16_t *p, PyObject *val)
{
p[0] = (uint16_t)(val >> 0);
p[1] = (uint16_t)(val >> 16);
p[2] = (uint16_t)(val >> 32);
p[3] = (uint16_t)(val >> 48);
memcpy(p, &val, sizeof(val));
}

static inline uint32_t
read_u32(uint16_t *p)
{
uint32_t val = 0;
val |= (uint32_t)p[0] << 0;
val |= (uint32_t)p[1] << 16;
uint32_t val;
memcpy(&val, p, sizeof(val));
return val;
}

static inline uint64_t
read_u64(uint16_t *p)
{
uint64_t val = 0;
val |= (uint64_t)p[0] << 0;
val |= (uint64_t)p[1] << 16;
val |= (uint64_t)p[2] << 32;
val |= (uint64_t)p[3] << 48;
uint64_t val;
memcpy(&val, p, sizeof(val));
return val;
}

#endif

static inline void
write_obj(uint16_t *p, PyObject *obj)
{
uintptr_t val = (uintptr_t)obj;
#if SIZEOF_VOID_P == 8
write_u64(p, val);
#elif SIZEOF_VOID_P == 4
write_u32(p, val);
#else
#error "SIZEOF_VOID_P must be 4 or 8"
#endif
}

static inline PyObject *
read_obj(uint16_t *p)
{
uintptr_t val;
#if SIZEOF_VOID_P == 8
val = read_u64(p);
#elif SIZEOF_VOID_P == 4
val = read_u32(p);
#else
#error "SIZEOF_VOID_P must be 4 or 8"
#endif
return (PyObject *)val;
PyObject *val;
memcpy(&val, p, sizeof(val));
return val;
}

/* See Objects/exception_handling_notes.txt for details.
Expand Down
2 changes: 2 additions & 0 deletions Misc/NEWS.d/next/Core and Builtins/2022-09-12-15-15-04.gh-issue-90997.sZO8c9.rst
Show comments
View file
Open in desktop
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode characters
Show hidden characters
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
Improve the performance of reading and writing inline bytecode caches on
some platforms.
Back | FazBrowse Home | New Git URL