Ekwakhiweni kwamapulatifomu e-x86, kuye kwavela izitayela ezimbili ezihambisanayo. Ngokwenguqulo eyodwa, sidinga ukuya ekuhlanganiseni izinsiza zekhompuyutha nokulawula zibe chip eyodwa. Indlela yesibili ikhuthaza ukusatshalaliswa kwezibopho: iphrosesa ifakwe ibhasi elisebenza kahle elakha i-peripheral scalable ecosystem. Yakha isisekelo se-Intel C620 system logic topology yamapulatifomu asezingeni eliphezulu.
Umehluko oyisisekelo ovela ku-chipset ye-Intel C610 yangaphambilini ukunwetshwa kwesiteshi sokuxhumana phakathi kwephrosesa nama-peripherals afakwe ku-PCH chip ngokusebenzisa izixhumanisi ze-PCIe kanye nebhasi lendabuko le-DMI.
Ake sibhekisise ukusungulwa kwebhuloho eliseningizimu ye-Intel Lewisburg: yiziphi izindlela zokuziphendukela kwemvelo ezikhulise amandla ayo ekuxhumaneni namaphrosesa?
Izinguquko zokuziphendukela kwemvelo ekuxhumaneni kwe-CPU-PCH
Njengengxenye yendlela yokuziphendukela kwemvelo, umzila omkhulu wokuxhumana phakathi kwe-CPU nebhuloho eliseningizimu, okuyibhasi le-DMI (Direct Media Interface), uthole ukusekelwa kwemodi ye-PCIe x4 Gen3 esebenza ngo-8.0 GT/S. Ngaphambilini, ku-Intel C610 PCH, ukuxhumana phakathi kwephrosesa nokucabanga kwesistimu kwenziwa ngemodi ye-PCIe x4 Gen 2 kumkhawulokudonsa ongu-5.0 GT/S.
Ukuqhathaniswa kokusebenza kwe-logic yesistimu ye-Intel C610 ne-C620
Qaphela ukuthi lesi simiso esingaphansi sinokonga kakhulu kunamachweba we-PCIe akhelwe ngaphakathi ephrosesa, ngokuvamile asetshenziselwa ukuxhuma ama-GPU namadrayivu e-NVMe, lapho i-PCIe 3.0 isetshenziswe isikhathi eside futhi kuhlelwe ukushintshela ku-PCI Express Gen4.
Izinguquko zenguquko ekuxhumaneni kwe-CPU-PCH
Izinguquko eziguquguqukayo zifaka phakathi ukufakwa kweziteshi ezintsha zokuxhumana ze-PCIe CPU-PCH, ezibizwa ngokuthi Ama-Uplink Engeziwe. Ngokomzimba, lawa amachweba amabili e-PCI Express asebenza ku-PCIe x8 Gen3 kanye ne-PCIe x16 Gen3 izindlela, zombili i-8.0 GT/S.
Ngokusebenzelana phakathi kwe-CPU ne-Intel C620 PCH, kusetshenziswa amabhasi ama-3: i-DMI nezimbobo ezimbili ze-PCI Express
Kungani bekudingeka ukuthi kubuyekezwe i-topology yokuxhumana ekhona ne-Intel C620? Okokuqala, kufika kubalawuli benethiwekhi be-4x 10GbE abanokusebenza kwe-RDMA bangahlanganiswa ku-PCH. Okwesibili, isizukulwane esisha nesisheshayo se-Intel QuickAssist Technology (QAT) coprocessors, esihlinzeka ngosekelo lwehadiwe lokucindezelwa nokubethela, sinesibopho sokubethela ithrafikhi yenethiwekhi nokuhwebelana ngohlelo olungaphansi lwesitoreji. Futhi ekugcineni, "injini emisha" - , ezotholakala kuphela kuma-OEM.
Ukunwebeka nokuvumelana nezimo
Impahla ebalulekile yikhono lokukhetha ngokukhetha hhayi kuphela i-topology yokuxhuma kwe-PCH, kodwa futhi izinto eziza kuqala zezinsiza zangaphakathi ze-chip ekufinyeleleni iziteshi zokuxhumana ezinesivinini esiphezulu ne-central processor (amaprosesa). Ngaphezu kwalokho, ku-EPO ekhethekile (Imodi Ye-EndPoint Kuphela), uxhumano lwe-PCH lwenziwa esimweni sedivayisi evamile ye-PCI Express equkethe izinsiza ezingu-10 GbE kanye ne-Intel QAT. Ngesikhathi esifanayo, i-interface ye-DMI yakudala, kanye nenani lezinhlelo ezingaphansi ze-Legacy, eziboniswe ngokumnyama kumdwebo, zikhutshaziwe.
Izakhiwo zangaphakathi ze-Intel C620 PCH chip
Ngokombono, lokhu kwenza kube nokwenzeka ukusebenzisa i-chip ye-Intel C620 PCH engaphezu kweyodwa ohlelweni, ukukala u-10 GbE kanye nokusebenza kwe-Intel QAT ukuze kuhlangatshezwane nezidingo zokusebenza. Ngesikhathi esifanayo, imisebenzi ye-Legacy edingekayo kuphela ekhophini eyodwa inganikwa amandla kuphela kweyodwa yama-chips e-PCH afakiwe.
Ngakho-ke, izwi lokugcina ekwakhiweni lizoba ngelikanjiniyela weplathifomu, esebenza ngesisekelo sazo zombili izici zobuchwepheshe nezokuthengisa ngokuhambisana nokuma komkhiqizo othile ngamunye.
Source: www.habr.com