NVIDIA Developer Forums

cudaMemcpy() and L2 cache.

Accelerated Computing CUDA CUDA Programming and Performance

Robert_Crovella May 17, 2016, 9:01pm 2

ALL accesses to DRAM take place through the L2 cache. And the L2 cache cannot be disabled.

1 Like

What the difference using L2 by cudaMemcpy() and from kernel?

Topic		Replies	Views	Activity
What the difference using L2 by cudaMemcpy() and from kernel? General Topics and Other SDKs	1	511	May 4, 2023
cudaMemcpy DeviceToDevice and L2 cache usage CUDA Programming and Performance	2	163	April 15, 2025
Difference between L2 read/write transactions and L2_L1 read/write transactions ? CUDA Programming and Performance	3	1660	August 28, 2019
Question about GPU L2 cache memory access。 Nsight Compute cuda , kernel	5	1125	February 21, 2024
L2 Hit Rate(Texture Reads) becomes 100% when modifying memory never used CUDA Programming and Performance	7	2766	March 17, 2018
L2 Cache mechanism for streaming data? CUDA Programming and Performance	1	834	August 25, 2022
L1 and L2 cache hit rate CUDA Programming and Performance	8	6791	February 3, 2016
Memory transaction size CUDA Programming and Performance	1	1798	February 12, 2017
Fermi L2 cache How fast is the L2 cache? How do I access it? CUDA Programming and Performance	11	26347	December 2, 2011
cudaMallocHost caching behavior CUDA Programming and Performance	4	811	March 1, 2019