| 1 |
|
| 2 |
Platform: AMD Accelerated Parallel Processing |
| 3 |
Device: Capeverde |
| 4 |
Driver version : 1214.3 (VM) (Linux x86) |
| 5 |
Compute units : 8 |
| 6 |
|
| 7 |
Global memory bandwidth (GBPS) |
| 8 |
float : 60.35 |
| 9 |
float2 : 63.30 |
| 10 |
float4 : 63.32 |
| 11 |
float8 : 33.13 |
| 12 |
float16 : 16.93 |
| 13 |
|
| 14 |
Single-precision compute (GFLOPS) |
| 15 |
float : 835.89 |
| 16 |
float2 : 835.08 |
| 17 |
float4 : 832.62 |
| 18 |
float8 : 828.40 |
| 19 |
float16 : 819.06 |
| 20 |
|
| 21 |
Double-precision compute (GFLOPS) |
| 22 |
double : 52.20 |
| 23 |
double2 : 52.43 |
| 24 |
double4 : 52.37 |
| 25 |
double8 : 52.26 |
| 26 |
double16 : 52.03 |
| 27 |
|
| 28 |
Integer compute (GIOPS) |
| 29 |
int : 167.80 |
| 30 |
int2 : 167.78 |
| 31 |
int4 : 167.74 |
| 32 |
int8 : 166.86 |
| 33 |
int16 : 167.55 |
| 34 |
|
| 35 |
Transfer bandwidth (GBPS) |
| 36 |
enqueueWriteBuffer : 2.81 |
| 37 |
enqueueReadBuffer : 2.90 |
| 38 |
enqueueMapBuffer(for read) : 16.34 |
| 39 |
memcpy from mapped ptr : 1.17 |
| 40 |
enqueueUnmap(after write) : 2.81 |
| 41 |
memcpy to mapped ptr : 1.03 |
| 42 |
|
| 43 |
Kernel launch latency : 120.57 us |
| 44 |
|
| 45 |
|