CPU
Test by PassMark.
int32 : 1.73 GOPS
float32 : 10.01 GOPS
GPU
Test by CLPeak.
Platform: QUALCOMM Snapdragon(TM)
Device: QUALCOMM Adreno(TM)
Driver version : OpenCL 3.0 QUALCOMM build: commit #fadc7176c9 changeid #I9c7199d547 Date: 03/23/22 Wed Local Branch: Remote Branch: Compiler E031.38.11.03 (Android)
Compute units : 4
Clock frequency : 1 MHz
Global memory bandwidth (GBPS)
float : 43.89
float2 : 45.29
float4 : 46.14
float8 : 46.89
float16 : 23.90
Single-precision compute (GFLOPS)
float : 1362.53
float2 : 1488.83
float4 : 1375.68
float8 : 1580.62
float16 : 1568.67
Half-precision compute (GFLOPS)
half : 2071.59
half2 : 2538.30
half4 : 2885.05
half8 : 3004.66
half16 : 3042.55
No double precision support! Skipped
Integer compute (GIOPS)
int : 388.15
int2 : 343.14
int4 : 376.12
int8 : 307.17
int16 : 321.75
Integer compute Fast 24bit (GIOPS)
int : 1245.01
int2 : 722.93
int4 : 1106.13
int8 : 1117.72
int16 : 1086.25
Transfer bandwidth (GBPS)
enqueueWriteBuffer : 13.56
enqueueReadBuffer : 13.26
enqueueWriteBuffer non-blocking : 12.69
enqueueReadBuffer non-blocking : 13.20
enqueueMapBuffer(for read) : 4224.00
memcpy from mapped ptr : 12.42
enqueueUnmap(after write) : 2459.33
memcpy to mapped ptr : 13.36
Kernel launch latency : 287.81 us