Umhlinzeki wamafu we-Amazon Web Services (AWS) umemezele ukwethulwa kwemodeli entsha yokusetshenziswa, i-EC2 Capacity Blocks ye-ML, eyenzelwe amabhizinisi afuna ukugodla ukufinyelela kuma-accelerator ekhompyutha ukuze aphathe umthwalo wesikhashana we-AI. I-Amazon's EC2 Capacity Blocks for ML solution ivumela amakhasimende ukuthi agcine ukufinyelela "emakhulwini" ama-accelerator e-NVIDIA H100 ku-EC2 UltraClusters, eklanyelwe imithwalo yokusebenza yomshini esebenza kahle kakhulu. Amaklayenti avele acacise usayizi weqoqo elifunekayo kanye nedethi yokuqala neyokuphela yokufinyelela. Lokhu kukhulisa ukubikezelwa kokutholakala kwensiza ye-AI kuyilapho kususa isidingo sokukhokhela ukufinyelela kumthamo lapho ingasetshenziswa. I-AWS nayo iyazuza ngoba le ndlela isebenzisa kangcono izinsiza ezikhona.
Source: 3dnews.ru