mirror of
https://github.com/boykopovar/AnyPS5.git
synced 2026-10-23 09:03:22 -05:00
Dean Galvin
604826c138
CpuTopology::Get and Pin did nothing off Windows, so on a hybrid CPU the queue workers, the draw front end and the presenter ran wherever the scheduler put them. On a Linux kernel that reports the same capacity for every CPU, those threads spent part of each frame on the efficiency cores. Get now reads the process affinity with sched_getaffinity and the core classes from /sys/devices/cpu_core/cpus and /sys/devices/cpu_atom/cpus, and Pin restricts the calling thread with pthread_setaffinity_np. A CPU is hybrid only when both lists exist and are disjoint, so non-hybrid CPUs are never pinned. On Linux the workers are pinned to the performance cores of a hybrid CPU by default; APS5_NO_WORKER_AFFINITY=1 leaves them free. Windows keeps the APS5_WORKER_AFFINITY=1 opt-in, and APS5_WORKER_AFFINITY_MASK=<hex> still picks the cores on any platform. PPSA21564 on a hybrid CPU, pinned against unpinned (2 runs each): title 46.7 to 49.0 fps, menu 45.4 to 47.7, intro level 37.2 to 41.3 (space) and 31.7 to 37.6 (ships). Hub descent 6.56 to 7.92 fps (3 runs each).
50 lines
2.1 KiB
C++
50 lines
2.1 KiB
C++
#include "prx/libc/include/CpuTopology.hpp"
|
|
#include <cstdio>
|
|
#include <cstdlib>
|
|
#include <filesystem>
|
|
#include <fstream>
|
|
|
|
static void Require(bool value, const char* what) {
|
|
if (value) return;
|
|
std::fprintf(stderr, "cpu topology test failed: %s\n", what);
|
|
std::exit(1);
|
|
}
|
|
|
|
static std::uint64_t Parse(const char* text) {
|
|
const auto path = std::filesystem::temp_directory_path() / "anyps5_cpu_topology_test_cpus";
|
|
{
|
|
std::ofstream file(path, std::ios::trunc);
|
|
file << text;
|
|
}
|
|
const auto mask = CpuTopology::MaskFromCpuList(path.c_str());
|
|
std::filesystem::remove(path);
|
|
return mask;
|
|
}
|
|
|
|
int main() {
|
|
Require(Parse("0-15\n") == 0xffffull, "range");
|
|
Require(Parse("16-31\n") == 0xffff0000ull, "high range");
|
|
Require(Parse("0,2,4-5\n") == 0x35ull, "list of cpus and ranges");
|
|
Require(Parse("3") == 0x8ull, "single cpu without a newline");
|
|
Require(Parse("62-70\n") == 0xc000000000000000ull, "cpus past 63 are left out");
|
|
Require(Parse("") == 0, "empty list");
|
|
Require(Parse("4-2\n") == 0, "reversed range");
|
|
Require(Parse("0-x\n") == 0, "malformed range");
|
|
Require(Parse("0;1\n") == 0, "malformed separator");
|
|
Require(CpuTopology::MaskFromCpuList("/nonexistent/anyps5/cpus") == 0, "missing file");
|
|
|
|
const auto& layout = CpuTopology::Get();
|
|
Require(layout.process != 0, "process affinity read");
|
|
Require(!layout.hybrid || (layout.performant & layout.efficient) == 0, "hybrid classes overlap");
|
|
const std::uint64_t first = layout.process & (~layout.process + 1);
|
|
Require(CpuTopology::Pin(nullptr, first) == first, "pin to one cpu of the process");
|
|
cpu_set_t set;
|
|
CPU_ZERO(&set);
|
|
Require(pthread_getaffinity_np(pthread_self(), sizeof(set), &set) == 0, "read thread affinity");
|
|
Require(CPU_COUNT(&set) == 1 && CPU_ISSET(static_cast<unsigned>(__builtin_ctzll(first)), &set), "thread runs on the pinned cpu only");
|
|
Require(CpuTopology::Pin(nullptr, ~layout.process) == 0, "mask outside the process is not applied");
|
|
int dummy = 0;
|
|
Require(CpuTopology::Pin(&dummy, first) == 0, "other threads are not pinned by handle");
|
|
return 0;
|
|
}
|