1Platform: Portable Computing Language 2 Device: pthread-POWER9, altivec supported 3 Driver version : 3.0-rc2 (Linux unknown) 4 Compute units : 160 5 Clock frequency : 3800 MHz 6 7 Global memory bandwidth (GBPS) 8 float : 30.30 9 float2 : 59.39 10 float4 : 63.92 11 float8 : 60.26 12 float16 : 57.05 13 14 Single-precision compute (GFLOPS) 15 float : 73.11 16 float2 : 179.68 17 float4 : 411.74 18 float8 : 739.41 19 float16 : 910.81 20 21 No half precision support! Skipped 22 23 Double-precision compute (GFLOPS) 24 double : 85.08 25 double2 : 151.08 26 double4 : 275.05 27 double8 : 401.79 28 double16 : 456.30 29 30 Integer compute (GIOPS) 31 int : 112.89 32 int2 : 189.39 33 int4 : 440.41 34 int8 : 708.03 35 int16 : 748.61 36 37 Integer compute Fast 24bit (GIOPS) 38 int : 149.56 39 int2 : 226.40 40 int4 : 407.09 41 int8 : 721.65 42 int16 : 755.17 43 44 Transfer bandwidth (GBPS) 45 enqueueWriteBuffer : 5.88 46 enqueueReadBuffer : 5.37 47 enqueueWriteBuffer non-blocking : 5.52 48 enqueueReadBuffer non-blocking : 5.24 49 enqueueMapBuffer(for read) : 901.70 50 memcpy from mapped ptr : 7.58 51 enqueueUnmap(after write) : 734.74 52 memcpy to mapped ptr : 11.31 53 54 Kernel launch latency : 76.72 us 55