Google e senotse pokello ea eona e ncha ea Maglev JIT, e tla fetisetsoa butle-butle ho basebelisi ba Chrome 114 ka la 5 Phuptjane. JIT compiler e etselitsoe ho hlahisa khoutu ea mochini e sebetsang ka potlako bakeng sa khoutu ea JavaScript e sebelisoang haholo. Ho nolofaletsa Maglev ho felletse ka ntlafatso ea 7.5% sethaleng sa Jetstream le ntlafatso ea 5% ho tekanyo ea Speedometer.
Ntle le moo, matla a akaretsang a kholo ea ts'ebetso ea Chrome a boleloa:
- Tekong ea Speedometer, e shebaneng le ho lekola karabelo ea sebatli ha u bala liwebsaete le ho lekanya lebelo la ts'ebetso ea lilaebrari tse tsebahalang tsa JavaScript, lintlha tsa Chrome li ntlafetse ho tloha ho lintlha tse 330 ho isa ho tse 491. Ntle le phetoho ea Maglev, tlhahlobo e ile ea boela ea nahanela lintlafatso tse ling tse entsoeng liphatlalatsong selemong se fetileng (ho tloha ha ho lokolloa 101), joalo ka optimizations bakeng sa mehala ea ts'ebetso enjeneng ea JavaScript.
- Sebokeng sa Jetstream, se lekang ts'ebetso ka lits'ebetso tsa webo tse tsoetseng pele tsa JavaScript le WebAssembly, Maglev e fihletse lintlha tse 330 (ntlafatso ea 7.5%).
- Ka har'a benchmark ea MotionMark, e lekang bokhoni ba lits'oants'o tsa sebatli sa ho fana ka tlhaiso-leseling ka litefiso tse phahameng tsa foreimi, ts'ebetso e ntlafetse ka makhetlo a mararo ho tloha selemong se fetileng. Ho tloha qalong ea selemo, bahlahisi ba khothalelitse ho feta 20 optimizations ho potlakisa ts'ebetso ea litšoantšo ho Chrome, halofo ea eona e se e kenyelelitsoe ho codebase e tsitsitseng ea tokollo. Mohlala, ts'ebetso ea Canvas e ntlafalitsoe, lintlafatso tse ipapisitseng le profiling ea khoutu li kenyellelitsoe, kemiso ea mesebetsi ea GPU e ntlafalitsoe, ts'ebetso ea ho kopanya e ntlafalitsoe, algorithm e ncha e matla ea anti-aliasing, MSAA (Multisample Anti-Aliasing), e kentsoe ts'ebetsong, 'me rasterization ea canvas ea 2D e fetiselitsoe ho arola lits'ebetso tse tšoanang.
Source: opennet.ru
