|
FastDeploy
latest
Fast & Easy to Deploy!
|
Option object to configure GraphCore IPU. More...
#include <option.h>
Public Attributes | |
| int | ipu_device_num |
| IPU device id. | |
| int | ipu_micro_batch_size |
| the batch size in the graph, only work when graph has no batch shape info | |
| bool | ipu_enable_pipelining |
| enable pipelining | |
| int | ipu_batches_per_step |
| the number of batches per run in pipelining | |
| bool | ipu_enable_fp16 |
| enable fp16 | |
| int | ipu_replica_num |
| the number of graph replication | |
| float | ipu_available_memory_proportion |
| the available memory proportion for matmul/conv | |
| bool | ipu_enable_half_partial |
| enable fp16 partial for matmul, only work with fp16 | |
Option object to configure GraphCore IPU.
1.8.13