Skip to content

Improve total L2 size calculation logic on osx-arm64 #75881

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

Closed
neon-sunset wants to merge 2 commits into from
Closed

Improve total L2 size calculation logic on osx-arm64 #75881

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
4dc3c58
Calculate total L2 size of all performance cores on osx-arm64
neon-sunset Sep 20, 2022
3781b7e
Fix formatting
neon-sunset Sep 20, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 29 additions & 8 deletions src/coreclr/gc/unix/gcenv.unix.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -965,17 +965,38 @@ static size_t GetLogicalProcessorCacheSizeFromOS()
#if HAVE_SYSCTLBYNAME
if (cacheSize == 0)
{
bool success = false;
int64_t cacheSizeFromSysctl = 0;
size_t sz = sizeof(cacheSizeFromSysctl);
const bool success = false
// macOS: Since macOS 12.0, Apple added ".perflevelX." to determinate cache sizes for efficiency
// and performance cores separately. "perflevel0" stands for "performance"
|| sysctlbyname("hw.perflevel0.l3cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.perflevel0.l2cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
// macOS: Since macOS 12.0, Apple added ".perflevelX." to determinate cache sizes for efficiency
// and performance cores separately. "perflevel0" stands for "performance"
if (sysctlbyname("hw.perflevel0.l3cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0)
{
success = true;
}
else if (sysctlbyname("hw.perflevel0.l2cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0)
{
int64_t cpusMax;
int64_t cpusPerL2;
size_t szCpusMax = sizeof(cpusMax);
size_t szCpusPerL2 = sizeof(cpusPerL2);
// macOS: The reported L2 cache size is actually not total but rather per CPU "cluster".
// Therefore, we need to account for CPU count per "cluster" to calculate total size
if (sysctlbyname("hw.perflevel0.physicalcpu_max", &cpusMax, &szCpusMax, nullptr, 0) == 0 &&
sysctlbyname("hw.perflevel0.cpusperl2", &cpusPerL2, &szCpusPerL2, nullptr, 0) == 0)
{
cacheSizeFromSysctl *= (cpusMax / cpusPerL2);
}
success = true;
}
else
{
// macOS: these report cache sizes for efficiency cores only:
|| sysctlbyname("hw.l3cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.l2cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.l1dcachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0;
success = sysctlbyname("hw.l3cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.l2cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.l1dcachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0;
}

if (success)
{
assert(cacheSizeFromSysctl > 0);
Expand Down
37 changes: 29 additions & 8 deletions src/coreclr/pal/src/misc/sysinfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -633,17 +633,38 @@ PAL_GetLogicalProcessorCacheSizeFromOS()
#if HAVE_SYSCTLBYNAME
if (cacheSize == 0)
{
bool success = false;
int64_t cacheSizeFromSysctl = 0;
size_t sz = sizeof(cacheSizeFromSysctl);
const bool success = false
// macOS: Since macOS 12.0, Apple added ".perflevelX." to determinate cache sizes for efficiency
// and performance cores separately. "perflevel0" stands for "performance"
|| sysctlbyname("hw.perflevel0.l3cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.perflevel0.l2cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
// macOS: Since macOS 12.0, Apple added ".perflevelX." to determinate cache sizes for efficiency
// and performance cores separately. "perflevel0" stands for "performance"
if (sysctlbyname("hw.perflevel0.l3cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0)
{
success = true;
}
else if (sysctlbyname("hw.perflevel0.l2cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0)
{
int64_t cpusMax;
int64_t cpusPerL2;
size_t szCpusMax = sizeof(cpusMax);
size_t szCpusPerL2 = sizeof(cpusPerL2);
// macOS: The reported L2 cache size is actually not total but rather per CPU "cluster".
// Therefore, we need to account for CPU count per "cluster" to calculate total size
if (sysctlbyname("hw.perflevel0.physicalcpu_max", &cpusMax, &szCpusMax, nullptr, 0) == 0 &&
sysctlbyname("hw.perflevel0.cpusperl2", &cpusPerL2, &szCpusPerL2, nullptr, 0) == 0)
{
cacheSizeFromSysctl *= (cpusMax / cpusPerL2);
}
success = true;
}
else
{
// macOS: these report cache sizes for efficiency cores only:
|| sysctlbyname("hw.l3cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.l2cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.l1dcachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0;
success = sysctlbyname("hw.l3cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.l2cachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0
|| sysctlbyname("hw.l1dcachesize", &cacheSizeFromSysctl, &sz, nullptr, 0) == 0;
}

if (success)
{
_ASSERTE(cacheSizeFromSysctl > 0);
Expand Down