| 1 |
Platform: AMD Accelerated Parallel Processing |
| 2 |
Device: Intel(R) Xeon(R) CPU E5-2670 0 @ 2.60GHz |
| 3 |
Driver version : 1598.5 (sse2,avx) (Linux x64) |
| 4 |
Compute units : 32 |
| 5 |
Clock frequency : 3283 MHz |
| 6 |
|
| 7 |
Global memory bandwidth (GBPS) |
| 8 |
float : 25.27 |
| 9 |
float2 : 26.60 |
| 10 |
float4 : 26.84 |
| 11 |
float8 : 27.01 |
| 12 |
float16 : 44.40 |
| 13 |
|
| 14 |
Single-precision compute (GFLOPS) |
| 15 |
float : 23.95 |
| 16 |
float2 : 47.73 |
| 17 |
float4 : 95.47 |
| 18 |
float8 : 187.04 |
| 19 |
float16 : 350.93 |
| 20 |
|
| 21 |
Double-precision compute (GFLOPS) |
| 22 |
double : 23.96 |
| 23 |
double2 : 47.85 |
| 24 |
double4 : 95.18 |
| 25 |
double8 : 186.15 |
| 26 |
double16 : 308.99 |
| 27 |
|
| 28 |
Integer compute (GIOPS) |
| 29 |
int : 47.79 |
| 30 |
int2 : 21.65 |
| 31 |
int4 : 126.29 |
| 32 |
int8 : 245.77 |
| 33 |
int16 : 350.12 |
| 34 |
|
| 35 |
Transfer bandwidth (GBPS) |
| 36 |
enqueueWriteBuffer : 3.82 |
| 37 |
enqueueReadBuffer : 3.54 |
| 38 |
enqueueMapBuffer(for read) : 18323.24 |
| 39 |
memcpy from mapped ptr : 6.43 |
| 40 |
enqueueUnmap(after write) : 31956.60 |
| 41 |
memcpy to mapped ptr : 6.55 |
| 42 |
|
| 43 |
Kernel launch latency : 9.52 us |