<html> <head> <base href="https://bugs.freedesktop.org/"> </head> <body> <div> <a class="bz_bug_link bz_status_NEW " title="NEW - amdgpu [RX Vega 64] system freeze while gaming" href="https://bugs.freedesktop.org/show_bug.cgi?id=109955#c128">Comment # 128</a> on <a class="bz_bug_link bz_status_NEW " title="NEW - amdgpu [RX Vega 64] system freeze while gaming" href="https://bugs.freedesktop.org/show_bug.cgi?id=109955">bug 109955</a> from <a class="email" href="mailto:haro41@gmx.de" title="haro41@gmx.de">haro41@gmx.de</a> <pre>Created <a href="attachment.cgi?id=145901" name="attach_145901" title="proposed fix for crashes, caused by frequent mclk level 0/1 switches">attachment 145901</a> <a href="attachment.cgi?id=145901&action=edit" title="proposed fix for crashes, caused by frequent mclk level 0/1 switches">[details]</a> <a href='page.cgi?id=splinter.html&bug=109955&attachment=145901'>[review]</a> proposed fix for crashes, caused by frequent mclk level 0/1 switches At least one of the causes for crashes, are more frequently, if vsync is enabled. In this case, memory clock levels are switched usually more frequently. By experiments i found, that especially the transient betweeen level 1 and level 0 is critical. The fact, that disabling memory level 0, helps as a workaround, confirms: this approach points in the right direction. Result of further experiments: By sending a 'PPSMC_MSG_SetUclkDownHyst' message to smc (enabling a hysterese feature ?), the crashes can be avoided, even with enabled mclk level 0 and vsync activated.</pre> </div> <hr> You are receiving this mail because: <ul> <li>You are the assignee for the bug.</li> </ul> </body> </html>