Kernel na Linux 6.0, wanda ake sa ran za a fito da shi ranar Litinin mai zuwa, ya haɗa da canjin da ke magance matsalolin aiki tare da tsarin da ke gudana akan na'urori na AMD Zen. An gano tushen faɗuwar aikin lambar da aka ƙara shekaru 20 da suka gabata don aiki a kusa da matsalar kayan aiki a wasu kwakwalwan kwamfuta. Matsalar hardware ta daɗe da gyarawa kuma baya bayyana a cikin kwakwalwan kwamfuta na yanzu, amma an manta da tsohuwar hanyar magance matsalar kuma ta zama tushen lalacewar aiki akan tsarin da ya danganci AMD CPUs na zamani. Sabbin tsarin da ke kan Intel CPUs tsohon tsarin aiki bai shafe su ba, tunda suna samun damar ACPI ta amfani da direban intel_idle daban, kuma ba babban direban processor_idle ba.
An ƙara wani aiki a cikin kwaya a cikin Maris 2002 don toshe bayyanar kwaro a cikin kwakwalwan kwamfuta da ke da alaƙa da rashin saita yanayin rashin aiki yadda yakamata saboda jinkirin sarrafa siginar STPCLK#. Don yin aiki a kusa da matsalar, aiwatar da ACPI ya ƙara ƙarin umarnin WAIT, wanda ke rage jinkirin mai sarrafawa ta yadda chipset ya sami lokacin shiga cikin rashin aiki. Lokacin da aka yi amfani da bayanin IBS (Sampling-Based Sampling) umarni akan na'urori na AMD Zen3, an gano cewa mai sarrafa na'ura yana ciyar da lokaci mai yawa don aiwatar da stubs, wanda ke haifar da fassarar kuskuren yanayin yanayin mai sarrafawa da saita yanayin bacci mai zurfi (C- Jiha) ta hanyar CPUidle.
Ana nuna wannan ɗabi'a a cikin rage yawan aiki a ƙarƙashin nauyin aiki wanda akai-akai tsakanin jahohi marasa aiki da masu aiki. Misali, lokacin amfani da facin da ke hana hanyar wucewa, matsakaicin gwajin tbench yana ƙaruwa daga 32191 MB/s zuwa 33805 MB/s.
source: budenet.ru