Page MenuHomeWildfire Games

AMD Ryzen Fix: Linux Edition
Needs ReviewPublic

Authored by Stan on Sat, Nov 7, 12:28 AM.
This revision needs review, but there are no reviewers specified.

Details

Reviewers
None
Trac Tickets
#5760
Summary

As reported in D2726 there is currently an issue with random slowdowns and speedups with AMD Ryzen CPUs. The former patch took care of windows and this one is for linux.

According to StackOverflow

CLOCK_REALTIME represents the machine's best-guess as to the current wall-clock, time-of-day time. As Ignacio and MarkR say, this means that CLOCK_REALTIME can jump forwards and backwards as the system time-of-day clock is changed, including by NTP.

CLOCK_MONOTONIC represents the absolute elapsed wall-clock time since some arbitrary, fixed point in the past. It isn't affected by changes in the system time-of-day clock.

If you want to compute the elapsed time between two events observed on the one machine without an intervening reboot, CLOCK_MONOTONIC is the best option.

Note that on Linux, CLOCK_MONOTONIC does not measure time spent in suspend, although by the POSIX definition it should. You can use the Linux-specific CLOCK_BOOTTIME for a monotonic clock that keeps running during suspend.

Since we only care about elapsed time, MONOTONIC should have been the initial choice. If supported, one might also want to use MONOTONIC_RAW

According to janwas:

yeah monotonic is better than realtime. there's also monotonic_raw which skips NTP freq adjustments, not sure that's better for long-running sessions.
Test Plan

Compile with the patch on a Ryzen CPU, and check there is no slowdown or speedups. You might have to start a lot of games to experience the bug.
Compile with a non Ryzen CPU and notice there is no change.

Event Timeline

Stan created this revision.Sat, Nov 7, 12:28 AM
Stan edited the summary of this revision. (Show Details)Sat, Nov 7, 12:30 AM
Stan edited the summary of this revision. (Show Details)
Stan edited the summary of this revision. (Show Details)

Successful build - Chance fights ever on the side of the prudent.

builderr-debug-macos.txt
../../../source/simulation2/scripting/JSInterface_Simulation.cpp:155:4: warning: suggest braces around initialization of subobject [-Wmissing-braces]
                        CFixedVector2D(-halfSize.X, -halfSize.Y),
                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated.
../../../source/third_party/fmt/format.cpp:145:7: warning: '_POSIX_C_SOURCE' is not defined, evaluates to 0 [-Wundef]
#if ((_POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600) && !_GNU_SOURCE) || (defined(__ANDROID__) && __ANDROID__)
      ^
../../../source/third_party/fmt/format.cpp:145:37: warning: '_XOPEN_SOURCE' is not defined, evaluates to 0 [-Wundef]
#if ((_POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600) && !_GNU_SOURCE) || (defined(__ANDROID__) && __ANDROID__)
                                    ^
2 warnings generated.
In file included from ../../../source/graphics/tests/test_Camera.cpp:17:
/Users/wfg/Jenkins/workspace/macos-differential/source/graphics/tests/test_Camera.h:168:4: warning: suggest braces around initialization of subobject [-Wmissing-braces]
                        CVector3D(-101.0f, -101.0f, 101.0f),
                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated.
In file included from ../../../source/simulation2/tests/test_SerializeTemplates.cpp:17:
/Users/wfg/Jenkins/workspace/macos-differential/source/simulation2/tests/test_SerializeTemplates.h:39:4: warning: suggest braces around initialization of subobject [-Wmissing-braces]
                        3, 0, 1, 4, 1, 5
                        ^~~~~~~~~~~~~~~~
                        {               }
1 warning generated.
builderr-release-macos.txt
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libnetwork.a(precompiled.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libtinygettext.a(precompiled.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libtinygettext.a(tinygettext.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/liblobby.a(precompiled.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libglooxwrapper.a(precompiled.o) has no symbols
../../../source/simulation2/scripting/JSInterface_Simulation.cpp:155:4: warning: suggest braces around initialization of subobject [-Wmissing-braces]
                        CFixedVector2D(-halfSize.X, -halfSize.Y),
                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated.
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libsimulation2.a(precompiled.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libscriptinterface.a(precompiled.o) has no symbols
../../../source/third_party/fmt/format.cpp:145:7: warning: '_POSIX_C_SOURCE' is not defined, evaluates to 0 [-Wundef]
#if ((_POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600) && !_GNU_SOURCE) || (defined(__ANDROID__) && __ANDROID__)
      ^
../../../source/third_party/fmt/format.cpp:145:37: warning: '_XOPEN_SOURCE' is not defined, evaluates to 0 [-Wundef]
#if ((_POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600) && !_GNU_SOURCE) || (defined(__ANDROID__) && __ANDROID__)
                                    ^
2 warnings generated.
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libengine.a(precompiled.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libgraphics.a(precompiled.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libatlas.a(precompiled.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/libgui.a(precompiled.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/liblowlevel.a(dbghelp.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/liblowlevel.a(file_stats.o) has no symbols
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ranlib: file: ../../../binaries/system/liblowlevel.a(vfs_path.o) has no symbols
In file included from ../../../source/graphics/tests/test_Camera.cpp:17:
/Users/wfg/Jenkins/workspace/macos-differential/source/graphics/tests/test_Camera.h:168:4: warning: suggest braces around initialization of subobject [-Wmissing-braces]
                        CVector3D(-101.0f, -101.0f, 101.0f),
                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated.
In file included from ../../../source/simulation2/tests/test_SerializeTemplates.cpp:17:
/Users/wfg/Jenkins/workspace/macos-differential/source/simulation2/tests/test_SerializeTemplates.h:39:4: warning: suggest braces around initialization of subobject [-Wmissing-braces]
                        3, 0, 1, 4, 1, 5
                        ^~~~~~~~~~~~~~~~
                        {               }
1 warning generated.

Link to build: https://jenkins.wildfiregames.com/job/macos-differential/1768/display/redirect

Stan requested review of this revision.Sat, Nov 7, 1:12 AM

Can you point to specifically the linux issues? That thread has got big. I find it a little odd, at face value, that AMD would have somehow managed to mess up clock_gettime, which seems like a basic... And "ryzen clock_gettime linux" shows nothing too obvious".

Stan added a comment.Sat, Nov 7, 1:53 PM

@OptimusShepard reported that the slowdowns and the speedups happen as well on Arch Linux albeit much rarely.

It would seem the diff here fixes the issue and is technically correct.