StackOverflow Questions for Tag: gpu-warp

ctrlaltdel

Reputation: 5

Warp Reduce primitives to threads sharing same value

parallel-processingcudagputhread-synchronizationgpu-warp

Score: 0

Answers: 1

xwt1

Reputation: 15

Does __shfl_sync in CUDA always operate on registers, or does it involve shared memory or global memory in some certain situations?

cudashufflegpu-warp

Score: 0

Answers: 1

Paul

Reputation: 515

Understanting thread utilization in the CUDA reduction examples

cudanvidiagpu-warp

Score: 2

Answers: 1

Serge Rogatch

Reputation: 15090

Pre 8.x equivalent of __reduce_max_sync() in CUDA

c++parallel-processingcudagpu-warpcompute-capability

Score: 1

Answers: 2

Johan

Reputation: 76724

How to partition data in a warp based on a predicate so all keep items are consecutive

cudagpupartitioninggpu-warpstream-compaction

Score: 0

Answers: 1

gonidelis

Reputation: 1063

What is warp shuffling in CUDA and why is it useful?

cudagpugpu-shared-memorygpu-warp

Score: 6

Answers: 1

Regis Portalez

Reputation: 4860

Thread/warp local lock in cuda

cudacritical-sectiongpu-warp

Score: 0

Answers: 1

Armin Rigo

Reputation: 12990

On today's GPUs, can warps be recombined dynamically?

gpugpu-warp

Score: 1

Answers: 0

pem

Reputation: 465

Compute per-warp histogram without shared memory

c++cudahistogramgpu-warp

Score: 2

Answers: 2

SnowSR

Reputation: 3

CUDA __shfl_down_sync does not work with __match_any_sync

c++cudagpuptxgpu-warp

Score: -1

Answers: 1

Fabio T.

Reputation: 129

activemask() vs ballot_sync()

cudagpu-warp

Score: 9

Answers: 1

einpoklum

Reputation: 132108

How do I do the converse of shfl.idx (i.e. warp scatter instead of warp gather)?

cudashuffleptxgpu-warp

Score: 2

Answers: 2

nanofarad

Reputation: 41281

Why is my CUDA warp shuffle sum using the wrong offset for one shuffle step?

cudacompiler-optimizationgpu-warp

Score: 4

Answers: 1

Timocafé

Reputation: 765

Warp shuffling for CUDA

cudashuffleptxgpu-warp

Score: 1

Answers: 3

einpoklum

Reputation: 132108

Are threads in a multi-dimensional CUDA kernel blocks packed to fill warps?

cudanvidiamultidimensional-cubegpu-warp

Score: 1

Answers: 1

Silicomancer

Reputation: 9196

Monitor active warps and threads during a divergent CUDA run

c++cudagpgpugpu-warpthread-divergence

Score: 0

Answers: 2

sg_man

Reputation: 833

In CUDA, how can I get this warp's thread mask in conditionally executed code (in order to execute e.g., __shfl_sync or <cg>.shfl?

cudagpu-warpgpu-cooperative-groups

Score: 0

Answers: 1

Gabriel

Reputation: 9442

How are 2D / 3D CUDA blocks divided into warps?

cudagpgpugpu-warp

Score: 21

Answers: 2

Johan

Reputation: 76724

What's the alternative for __match_any_sync on compute capability 6?

cudagpu-warp

Score: 4

Answers: 1

user6671981

Reputation:

Why use thread blocks larger than the number of cores per multiprocessor

cudanvidiagpu-warp

Score: 0

Answers: 1

PreviousPage 1Next

StackOverflow Questions for Tag: gpu-warp

Warp Reduce primitives to threads sharing same value

Does __shfl_sync in CUDA always operate on registers, or does it involve shared memory or global memory in some certain situations?

Understanting thread utilization in the CUDA reduction examples

Pre 8.x equivalent of __reduce_max_sync() in CUDA

How to partition data in a warp based on a predicate so all keep items are consecutive

What is warp shuffling in CUDA and why is it useful?

Thread/warp local lock in cuda

On today&#39;s GPUs, can warps be recombined dynamically?

Compute per-warp histogram without shared memory

CUDA __shfl_down_sync does not work with __match_any_sync

__activemask() vs __ballot_sync()

How do I do the converse of shfl.idx (i.e. warp scatter instead of warp gather)?

Why is my CUDA warp shuffle sum using the wrong offset for one shuffle step?

Warp shuffling for CUDA

Are threads in a multi-dimensional CUDA kernel blocks packed to fill warps?

Monitor active warps and threads during a divergent CUDA run

In CUDA, how can I get this warp&#39;s thread mask in conditionally executed code (in order to execute e.g., __shfl_sync or &lt;cg&gt;.shfl?

How are 2D / 3D CUDA blocks divided into warps?

What&#39;s the alternative for __match_any_sync on compute capability 6?

Why use thread blocks larger than the number of cores per multiprocessor

On today's GPUs, can warps be recombined dynamically?

activemask() vs ballot_sync()

In CUDA, how can I get this warp's thread mask in conditionally executed code (in order to execute e.g., __shfl_sync or <cg>.shfl?

What's the alternative for __match_any_sync on compute capability 6?