A cikin gine-ginen dandamali na x86, abubuwa biyu sun bayyana waɗanda suka dace da juna. Dangane da sigar ɗaya, muna buƙatar matsawa zuwa haɗa kwamfuta da sarrafa albarkatun cikin guntu ɗaya. Hanya ta biyu tana haɓaka rarraba nauyi: na'ura mai sarrafawa tana sanye take da babbar motar bas wacce ke samar da yanayin yanayin yanayi mai iya daidaitawa. Yana samar da tushen tsarin Intel C620 tsarin dabaru na topology don manyan dandamali.
Bambanci mai mahimmanci daga Intel C610 chipset na baya shine fadada tashar sadarwa tsakanin mai sarrafawa da abubuwan da aka haɗa a cikin guntu PCH ta hanyar amfani da hanyoyin haɗin PCIe tare da bas din DMI na gargajiya.
Bari mu dubi sabbin hanyoyin gadar kudu ta Intel Lewisburg: wadanne hanyoyin juyin halitta da juyin juya hali ne suka fadada ikonsa wajen sadarwa da na'urori?
Canje-canjen juyin halitta a cikin sadarwar CPU-PCH
A matsayin wani ɓangare na tsarin juyin halitta, babban tashar sadarwa tsakanin CPU da gadar kudu, wacce ita ce bas ɗin DMI (Direct Media Interface), ta sami tallafi don yanayin PCIe x4 Gen3 tare da aikin 8.0 GT/S. A baya can, a cikin Intel C610 PCH, ana yin sadarwa tsakanin mai sarrafawa da tsarin dabaru a yanayin PCIe x4 Gen 2 a bandwidth 5.0 GT/S.
Kwatanta ayyukan dabaru na tsarin Intel C610 da C620
Lura cewa wannan tsarin yana da ra'ayin mazan jiya fiye da ginanniyar tashoshin PCIe na mai sarrafawa, yawanci ana amfani da su don haɗa GPUs da fayafai na NVMe, inda PCIe 3.0 aka yi amfani da shi na dogon lokaci kuma an shirya canji zuwa PCI Express Gen4.
Canje-canje na juyin juya hali a cikin sadarwar CPU-PCH
Canje-canjen juyin juya hali sun haɗa da ƙarin sabbin tashoshin sadarwa na PCIe CPU-PCH, wanda ake kira Ƙarin Uplinks. A zahiri, waɗannan tashoshin jiragen ruwa na PCI Express guda biyu ne waɗanda ke aiki a cikin PCIe x8 Gen3 da PCIe x16 Gen3 yanayin, duka 8.0 GT/S.
Don hulɗa tsakanin CPU da Intel C620 PCH, ana amfani da bas 3: DMI da tashoshin PCI Express guda biyu.
Me ya sa ya zama dole a sake fasalin yanayin sadarwa na yanzu tare da Intel C620? Da fari dai, har zuwa 4x 10GbE masu kula da hanyar sadarwa tare da aikin RDMA ana iya haɗa su cikin PCH. Abu na biyu, sabon ƙarni na Intel QuickAssist Technology (QAT) coprocessors, waɗanda ke ba da tallafin kayan aiki don matsawa da ɓoyewa, suna da alhakin ɓoye zirga-zirgar hanyar sadarwa da musayar tare da tsarin ajiya. Kuma a ƙarshe, "injin ƙididdiga" - , wanda zai kasance ga OEMs kawai.
Scalability da sassauci
Muhimmin kadarorin shine ikon zaɓin zaɓi ba kawai PCH dangane topology ba, har ma da fifikon albarkatun ciki na guntu don samun damar hanyoyin sadarwa mai sauri tare da na'ura mai sarrafawa (masu sarrafawa). Bugu da ƙari, a cikin EPO na musamman (EndPoint Only Mode), ana gudanar da haɗin PCH a cikin matsayi na na'urar PCI Express na yau da kullum wanda ke dauke da albarkatun 10 GbE da Intel QAT. A lokaci guda, mu'amalar DMI na al'ada, da kuma yawan tsarin tsarin Legacy, wanda aka nuna da baki a cikin zane, an kashe su.
Gine-ginen ciki na guntuwar Intel C620 PCH
A cikin ka'idar, wannan yana ba da damar yin amfani da guntu na Intel C620 PCH fiye da ɗaya a cikin tsarin, ƙaddamar da ayyukan 10 GbE da Intel QAT don saduwa da buƙatun aiki. A lokaci guda, Ayyukan Legacy waɗanda ake buƙata kawai a cikin kwafi ɗaya kawai za'a iya kunna su akan ɗaya daga cikin kwakwalwan PCH da aka shigar.
Don haka, magana ta ƙarshe a cikin ƙira za ta kasance ta mai haɓaka dandamali, tana aiki akan duka abubuwan fasaha da tallace-tallace daidai da matsayi na kowane takamaiman samfurin.
source: www.habr.com