StackOverflow Questions for Tag: cuda-streams

einpoklum

Reputation: 131385

What's the capacity of a CUDA stream (=queue)?

cudacuda-streams

Score: 4

Answers: 2

einpoklum

Reputation: 131385

Are "cuWhateverAsync" calls with stream handle NULL universally equivalent to "cuWhatever" calls?

asynchronouscudacuda-streamscuda-driver

Score: 0

Answers: 2

Anonymous

Reputation: 4632

How to use parallel torch cuda streams without causing oom? (example included)

pythonpython-3.xmachine-learningpytorchcuda-streams

Score: 1

Answers: 0

Lokesh

Reputation: 31

Compute and Data transfer not happening concurrently in cuda Streams on Iteration 2

cudagpunvccnsightcuda-streams

Score: 3

Answers: 0

user23864711

Reputation: 21

Some TensorRT conv layer forward blocked by cudaMemcpyAsync from another thread

cudatensorrtcuda-streams

Score: 1

Answers: 1

hy pan

Reputation: 1

CUDA Kernel Parameters cause "cudaMemcpy an illegal memory access was encountered"

cudacuda-streams

Score: 0

Answers: 2

Photos

Reputation: 41

Why is using multiple CUDA streams not improving performance as expected?

cudacuda-streams

Score: 2

Answers: 0

Photos

Reputation: 41

CUDA Graph Execution Taking Longer Than Original Kernel Launch Loop

cudacuda-streamscuda-graphs

Score: 1

Answers: 1

Photos

Reputation: 41

How to Use CUDA Graphs with Interdependent Streams and Dynamic Parameters?

cudacuda-streamscuda-graphs

Score: 1

Answers: 2

emely_pi

Reputation: 681

How can I profile CUDA code for memory management?

memory-managementcudaprofilingcuda-streams

Score: 1

Answers: 0

Subject_No_i

Reputation: 33

Is it possible to manually set the SMs used for one CUDA stream?

cudanvidiacudnncuda-streams

Score: 0

Answers: 1

einpoklum

Reputation: 131385

Are CUDA streams device-associated? And how do I get a stream's device?

cudamulti-gpucuda-streams

Score: 6

Answers: 2

Guido Sales Calvano

Reputation: 33

Can multiple cuda kernels execute in parallel on the same SM?

cudacuda-streams

Score: 1

Answers: 1

einpoklum

Reputation: 131385

What are the semantics of CUDA kernel launch priorities?

cudacuda-streamscuda-driver

Score: 0

Answers: 1

einpoklum

Reputation: 131385

What does the "synchronization policy" mean when launching a kernel?

cudasynchronizationcuda-streamscuda-driver

Score: 0

Answers: 1

einpoklum

Reputation: 131385

What are the new unique-id's for CUDA streams and contexts useful for?

cudauniqueidentifiercuda-streamscuda-contextcuda-driver

Score: 2

Answers: 1

Aitar

Reputation: 23

Why am I unable to establish a pipeline when using multiple concurrent streams in CUDA programming?

c++cudapipelinecuda-streams

Score: 2

Answers: 1

prghelps

Reputation: 3

What are the possible mistakes leading to 'fatal error: cudacheck.h: No such file or directory' in CUDA C++?

c++cudacuda-streams

Score: -3

Answers: 1

einpoklum

Reputation: 131385

Does a CUDA stream "become active" after execution of a scheduled host function concludes?

callbackcudacuda-streamscuda-driver

Score: 1

Answers: 0

VIArchitect

Reputation: 83

Can we overlap compute operation with memory operation without pinned memory on CPU?

pytorchcudacuda-streams

Score: 4

Answers: 1

PreviousPage 1Next

StackOverflow Questions for Tag: cuda-streams

What&#39;s the capacity of a CUDA stream (=queue)?

Are &quot;cuWhateverAsync&quot; calls with stream handle NULL universally equivalent to &quot;cuWhatever&quot; calls?

How to use parallel torch cuda streams without causing oom? (example included)

Compute and Data transfer not happening concurrently in cuda Streams on Iteration 2

Some TensorRT conv layer forward blocked by cudaMemcpyAsync from another thread

CUDA Kernel Parameters cause &quot;cudaMemcpy an illegal memory access was encountered&quot;

Why is using multiple CUDA streams not improving performance as expected?

CUDA Graph Execution Taking Longer Than Original Kernel Launch Loop

How to Use CUDA Graphs with Interdependent Streams and Dynamic Parameters?

How can I profile CUDA code for memory management?

Is it possible to manually set the SMs used for one CUDA stream?

Are CUDA streams device-associated? And how do I get a stream&#39;s device?

Can multiple cuda kernels execute in parallel on the same SM?

What are the semantics of CUDA kernel launch priorities?

What does the &quot;synchronization policy&quot; mean when launching a kernel?

What are the new unique-id&#39;s for CUDA streams and contexts useful for?

Why am I unable to establish a pipeline when using multiple concurrent streams in CUDA programming?

What are the possible mistakes leading to &#39;fatal error: cudacheck.h: No such file or directory&#39; in CUDA C++?

Does a CUDA stream &quot;become active&quot; after execution of a scheduled host function concludes?

Can we overlap compute operation with memory operation without pinned memory on CPU?

What's the capacity of a CUDA stream (=queue)?

Are "cuWhateverAsync" calls with stream handle NULL universally equivalent to "cuWhatever" calls?

CUDA Kernel Parameters cause "cudaMemcpy an illegal memory access was encountered"

Are CUDA streams device-associated? And how do I get a stream's device?

What does the "synchronization policy" mean when launching a kernel?

What are the new unique-id's for CUDA streams and contexts useful for?

What are the possible mistakes leading to 'fatal error: cudacheck.h: No such file or directory' in CUDA C++?

Does a CUDA stream "become active" after execution of a scheduled host function concludes?