CPU Tasks
LoadModel       593.861 milliseconds
RunComplete     1.13909 seconds
Run     1.06578 seconds
Callbacks       279.4 microseconds, 4 calls, 69.85 microseconds average
Spectrogram     12.0744 milliseconds, 3 calls, 4.0248 milliseconds average
Sample  3.0016 milliseconds, 28 calls, 107.2 microseconds average
Encode  614.44 milliseconds
Decode  451.048 milliseconds
DecodeStep      448.036 milliseconds, 28 calls, 16.0013 milliseconds average
    GPU Tasks
LoadModel       452.128 milliseconds
Run     1.0518 seconds
Encode  599.964 milliseconds
EncodeLayer     506.67 milliseconds, 24 calls, 21.1113 milliseconds average
Decode  451.832 milliseconds
DecodeStep      451.828 milliseconds, 28 calls, 16.1367 milliseconds average
DecodeLayer     412.142 milliseconds, 672 calls, 613.307 microseconds average
    Compute Shaders
mulMatTiled     562.478 milliseconds, 529 calls, 1.06329 milliseconds average
mulMatByRowTiled        256.062 milliseconds, 7803 calls, 32.8158 microseconds average
softMaxFixed    27.1687 milliseconds, 696 calls, 39.0355 microseconds average
normFixed       24.1828 milliseconds, 2093 calls, 11.5541 microseconds average
fmaRepeat1      23.3089 milliseconds, 2093 calls, 11.1366 microseconds average
addRepeatEx     22.3395 milliseconds, 2064 calls, 10.8234 microseconds average
softMaxLong     19.7192 milliseconds, 28 calls, 704.257 microseconds average
copyConvert     19.301 milliseconds, 1440 calls, 13.4035 microseconds average
copyTranspose   15.3011 milliseconds, 1392 calls, 10.9922 microseconds average
addRepeatScale  13.6043 milliseconds, 1344 calls, 10.1222 microseconds average
convolutionMain2Fixed   12.1242 milliseconds
addRepeatGelu   11.6172 milliseconds, 698 calls, 16.6436 microseconds average
addRepeat       11.5331 milliseconds, 744 calls, 15.5015 microseconds average
scaleInPlace    9.5743 milliseconds, 696 calls, 13.7562 microseconds average
convolutionMain 7.0349 milliseconds
softMax 5.8329 milliseconds, 672 calls, 8.67991 microseconds average
diagMaskInf     4.5297 milliseconds, 672 calls, 6.74062 microseconds average
convolutionPrep1        1.5258 milliseconds, 2 calls, 762.9 microseconds average
convolutionPrep2        383 microseconds, 2 calls, 191.5 microseconds average
addRows 194.6 microseconds, 28 calls, 6.95 microseconds average
add     95.2 microseconds
    Memory Usage
Model   877.966 KB RAM, 1.42785 GB VRAM
Context 1.9831 MB RAM, 723.782 MB VRAM
Total   2.84049 MB RAM, 2.13467 GB VRAM