From 166bbe6863b104dbdb76c548793a1eb47ba1063a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1=C5=A1=20Oberhuber?= <oberhuber.tomas@gmail.com>
Date: Mon, 8 Mar 2021 14:40:13 +0100
Subject: [PATCH] Added comment to Adaptive CSR kernel about finetunning of
CUDA block size depending on the size of Value type.
---
.../Algorithms/Segments/details/CSRAdaptiveKernelParameters.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/src/TNL/Algorithms/Segments/details/CSRAdaptiveKernelParameters.h b/src/TNL/Algorithms/Segments/details/CSRAdaptiveKernelParameters.h
index 0f00fbd808..843f2f7d52 100644
--- a/src/TNL/Algorithms/Segments/details/CSRAdaptiveKernelParameters.h
+++ b/src/TNL/Algorithms/Segments/details/CSRAdaptiveKernelParameters.h
@@ -15,6 +15,8 @@ namespace TNL {
namespace Segments {
namespace details {
+// This can be used for tunning the number of CUDA threads per block depending on the size of Value
+// TODO: Perform some tests
static constexpr int CSRAdaptiveKernelParametersCudaBlockSizes[] = { 256, 256, 256, 256, 256, 256 };
template< int SizeOfValue = 1,
--
GitLab