Add asynchronous concurrent execution #3687
Conversation
neon60
force-pushed
the
async-doc
branch
2 times, most recently
from
1484d67 to
f81588d
Compare
neon60
requested review from
chrispaquot,
gandryey,
saleelk,
mangupta and
rakesroy
as code owners
neon60
force-pushed
the
async-doc
branch
4 times, most recently
from
fd5af51 to
6a139c6
Compare
matyas-streamhpc
force-pushed
the
async-doc
branch
from
495e166 to
9835194
Compare
neon60
force-pushed
the
async-doc
branch
4 times, most recently
from
ed7e05f to
7cb3237
Compare
neon60
force-pushed
the
async-doc
branch
2 times, most recently
from
9592272 to
e0ab3f3
Compare
| Streams enable the overlap of computation and data transfer, ensuring | ||
| continuous GPU activity. |
There was a problem hiding this comment.
[NIT] Seems vague and out of place in.
There was a problem hiding this comment.
Need to explain, why should we use streams added an extra sentence to avoid the "out of place"
There was a problem hiding this comment.
The whole paragraph seems out of place. You come from section Streams and concurrent execution where you explain this idea. Then you restate it at the beginning of Managing streams. It is not wrong but it does not read well, and feels unnecessary/redundant. As it is a NIT, feel free to resolve this comment if you don't agree with the suggestion.
| Asynchronous memory operations allow data to be transferred between the host | ||
| and device while kernels are being executed on the GPU. Using operations like |
There was a problem hiding this comment.
I think this sentence is a bit misleading. A reader could miss the fact that this operation must be on a different streams to get this behavior.
Asynchronous memory operations, do not block the host while copying this data.
Asynchronous memory operations on multiple streams allow for data to be transferred between the host and device while kernels are executed. (and do not block the host while copying this data)
| One of the primary benefits of asynchronous operations is the ability to | ||
| overlap data transfer with kernel execution, leading to better resource | ||
| utilization and improved performance. |
There was a problem hiding this comment.
[NIT] you could clarify that multiple streams are needed to copy while executing a kernel in parallel.
| another. This technique is especially useful in applications with large data | ||
| sets that need to be processed quickly. | ||
|
|
||
| Concurrent data transfers |
There was a problem hiding this comment.
How does this differ to the Asynchronous memory operations section?
This feels repetitive, and it is not clear on how you wish to distinguish between concurrent and asynchronous within this context
There was a problem hiding this comment.
Removed concurrent data transfers section and added hipMemcpyPeerAsync mention to previous section.
No description provided.