TechQA.

Question

Bank Conflict Issue in CUDA Shared Memory Access

score 56 · Answer 1 · 2024-03-28T02:21:36.570000

0

Answer

56

Views

Bank Conflict Issue in CUDA Shared Memory Access

56 views Asked by Sunjnn At 28 March 2024 at 02:21

score 40 · Answer 2 · 2024-03-09T14:01:17.903000

Nsight Compute Range Replay mode usage

40 views Asked by 최보열 At 09 March 2024 at 14:01

score 46 · Answer 3 · 2024-03-04T16:06:52.917000

Nsight Compute can not non-interactive Profiler in Windows

46 views Asked by Ether At 04 March 2024 at 16:06

score 90 · Answer 4 · 2024-02-07T11:40:08.030000

How do I analyze register spills with Nsight Compute?

90 views Asked by Marko Grdinić At 07 February 2024 at 11:40

score 173 · Answer 5 · 2023-10-20T12:52:04.597000

use NCU with tensorRT, but got No kernels were profiled

173 views Asked by wenxin li At 20 October 2023 at 12:52

score 209 · Answer 6 · 2023-07-14T00:21:00.733000

CUDA math function register usage

209 views Asked by Chris Uchytil At 14 July 2023 at 00:21

score 356 · Answer 7 · 2023-07-12T19:10:27.310000

Roofline Model with CUDA Manual vs. Nsight Compute

356 views Asked by Cherry Toska At 12 July 2023 at 19:10

score 119 · Answer 8 · 2023-06-29T03:42:29.353000

Unbalanced Memory Read & Write in CUDA

119 views Asked by Alex Chen At 29 June 2023 at 03:42

score 187 · Answer 9 · 2023-03-27T12:55:28.653000

L2 Fabric cache hit rate of CUDA kernels on A100

187 views Asked by Shulai At 27 March 2023 at 12:55

score 306 · Answer 10 · 2023-03-01T12:58:55.083000

With the NSight Compute profiler, can I check cache hit rates for a specific region of memory?

306 views Asked by einpoklum At 01 March 2023 at 12:58

score 824 · Answer 11 · 2022-09-11T14:15:05.590000

Why is the Compute Throughput’s value different from the actual Performance / Peak Performance?

824 views Asked by TherLF At 11 September 2022 at 14:15

score 380 · Answer 12 · 2022-05-15T14:28:29.507000

Can I skip ahead to profile a specific invocation of a specific kernel?

380 views Asked by einpoklum At 15 May 2022 at 14:28

score 2416 · Answer 13 · 2022-04-28T11:25:15.090000

ncu-ui won't run: Could not load the Qt platform plugin "xcb" in "" even though it was found

2.4k views Asked by einpoklum At 28 April 2022 at 11:25

score 2689 · Answer 14 · 2021-10-16T21:09:41.623000

Nsight Compute says: "Profiling is not supported on this device" - why?

2.6k views Asked by einpoklum At 16 October 2021 at 21:09

score 870 · Answer 15 · 2021-08-09T14:21:38.107000

Filter on partial kernel name with Nsight Compute

870 views Asked by John At 09 August 2021 at 14:21

score 656 · Answer 16 · 2021-07-01T22:43:52.120000

Using ncu to profile pagefault in unified memory

656 views Asked by Daniel At 01 July 2021 at 22:43

score 795 · Answer 17 · 2021-03-14T14:15:15.367000

Which GPU execution dependencies have fixed latency (causing 'Wait' stalls)?

795 views Asked by einpoklum At 14 March 2021 at 14:15

score 507 · Answer 18 · 2021-02-17T10:03:50.887000

Shared memory loads not registered when using Tensor Cores

507 views Asked by rm95 At 17 February 2021 at 10:03

score 1575 · Answer 19 · 2021-02-16T23:07:01.863000

When does MIO Throttle stall happen?

1.5k views Asked by rm95 At 16 February 2021 at 23:07

score 3714 · Answer 20 · 2021-02-09T17:14:11.733000

What are the "long" and "short" scoreboards w.r.t. MIO/L1TEX?

3.7k views Asked by einpoklum At 09 February 2021 at 17:14

TechQA.

List Question

Bank Conflict Issue in CUDA Shared Memory Access

Nsight Compute Range Replay mode usage

Nsight Compute can not non-interactive Profiler in Windows

How do I analyze register spills with Nsight Compute?

use NCU with tensorRT, but got No kernels were profiled

CUDA math function register usage

Roofline Model with CUDA Manual vs. Nsight Compute

Unbalanced Memory Read & Write in CUDA

L2 Fabric cache hit rate of CUDA kernels on A100

With the NSight Compute profiler, can I check cache hit rates for a specific region of memory?

Why is the Compute Throughput’s value different from the actual Performance / Peak Performance?

Can I skip ahead to profile a specific invocation of a specific kernel?

ncu-ui won't run: Could not load the Qt platform plugin "xcb" in "" even though it was found

Nsight Compute says: "Profiling is not supported on this device" - why?

Filter on partial kernel name with Nsight Compute

Using ncu to profile pagefault in unified memory

Which GPU execution dependencies have fixed latency (causing 'Wait' stalls)?

Shared memory loads not registered when using Tensor Cores

When does MIO Throttle stall happen?

What are the "long" and "short" scoreboards w.r.t. MIO/L1TEX?

Popular Questions

Trending Questions