Hatsune Miku: Project Diva Extend

[Xele_]

I've made some profiling to see what's missing for a good framerate. Main bottleneck is the MIPS. It'll need to support 10+Millions of instructions per frame.

For one random frame, profiling for the function searching the instruction :

Code:

MIPSGetInstruction: 733.220 ms, call : 12366916

Then for each different instructions executed :

Code:

or: 67.384 ms, call : 1061124 jal: 21.445 ms, call : 369042 addiu: 99.921 ms, call : 1678724 lb: 2.591 ms, call : 29268 slti: 1.105 ms, call : 16446 sw: 38.326 ms, call : 527943 beq: 50.673 ms, call : 623778 bne: 54.813 ms, call : 661700 lw: 153.230 ms, call : 2282270 lh: 19.218 ms, call : 213797 jalr: 14.206 ms, call : 215107 addu: 30.604 ms, call : 460790 ori: 3.573 ms, call : 56272 and: 31.867 ms, call : 513482 andi: 2.882 ms, call : 49210 jr: 36.772 ms, call : 584145 slt: 5.722 ms, call : 92138 lhu: 0.731 ms, call : 8930 sb: 5.124 ms, call : 71372 sll: 46.410 ms, call : 767109 lbu: 16.514 ms, call : 172389 sltu: 2.461 ms, call : 39397 lui: 44.170 ms, call : 689434 bnel: 12.947 ms, call : 182817 blez: 0.081 ms, call : 1128 bltz: 0.327 ms, call : 3347 subu: 0.390 ms, call : 5989 beql: 20.897 ms, call : 333665 bgez: 9.227 ms, call : 155823 xori: 0.182 ms, call : 3449 srl: 11.459 ms, call : 172059 xor: 10.413 ms, call : 153687 sltiu: 11.229 ms, call : 163645 sh: 0.158 ms, call : 1953 syscall: 1.928 ms, call : 2442 sllv: 0.116 ms, call : 988 bgtz: 0.001 ms, call : 10 nor: 0.004 ms, call : 93 sra: 0.072 ms, call : 1035 mult: 0.067 ms, call : 345 mflo: 0.019 ms, call : 349 mtc1: 0.002 ms, call : 16 swc1: 0.006 ms, call : 78 divu: 0.011 ms, call : 29 mfhi: 0.006 ms, call : 25 mov.s: 0.003 ms, call : 8 lwc1: 0.000 ms, call : 5 srlv: 0.000 ms, call : 1 blezl: 0.005 ms, call : 63

I'll look forward to see how this will be optimized in the future.