... | @@ -14,15 +14,28 @@ Note that : |
... | @@ -14,15 +14,28 @@ Note that : |
|
|
|
|
|
**One worker is computing a full token sequence (64 tokens / worker)**
|
|
**One worker is computing a full token sequence (64 tokens / worker)**
|
|
|
|
|
|
|
|
Data transfer is done through tasks using high priority and TAMPI non blocking mode.
|
|
|
|
Dt = 1.0 seconds
|
|
|
|
|
|
|
|
|
|
|
|

|
|
|
|
|
|
|
|
|
|
|
|
Data transfer is done all at once through one single blocking MPI instruction.
|
|
|
|
Dt = 1.0 s
|
|
|
|
|
|
|
|

|
|
|
|

|
|
|
|
|
|
## Using 8 workers and 1 server
|
|
## Using 8 workers and 1 server
|
|
|
|
|
|
**One worker is computing half a token sequence (32 tokens / worker)**
|
|
**One worker is computing half a token sequence (32 tokens / worker)**
|
|
|
|
|
|
Data transfer is done through tasks using high priority.
|
|
Data transfer is done through tasks using high priority and TAMPI non blocking mode.
|
|
Dt = 2.5 seconds
|
|
Dt = 2.5 seconds
|
|

|
|

|
|
|
|
|
|
Data transfer is done through tasks using high priority.
|
|
Data transfer is done all at once through one single blocking MPI instruction.
|
|
Dt = 2.5 s
|
|
Dt = 2.5 s
|
|
|
|
|
|

|
|

|
... | | ... | |