From 7c7535bacf46dde577d29053b8ec49dfa53958f8 Mon Sep 17 00:00:00 2001 From: Tor Aamodt Date: Sun, 18 Jul 2010 16:11:08 -0800 Subject: finished refactor+implement call/return passing (now working on simple example included in this CL) [git-p4: depot-paths = "//depot/gpgpu_sim_research/fermi/distribution/": change = 6886] --- benchmarks/CUDA/template/template_kernel.cu | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'benchmarks/CUDA/template') diff --git a/benchmarks/CUDA/template/template_kernel.cu b/benchmarks/CUDA/template/template_kernel.cu index 14610d8..adac694 100644 --- a/benchmarks/CUDA/template/template_kernel.cu +++ b/benchmarks/CUDA/template/template_kernel.cu @@ -33,7 +33,7 @@ __device__ float foo( float bar ) { - return 2.0f*bar; + return bar+1.0f; } __global__ void @@ -51,7 +51,7 @@ testKernel( float* g_idata, float* g_odata) // read in input data from global memory // use the bank checker macro to check for bank conflicts during host // emulation - SDATA(tid) = foo(g_idata[tid])/2.0; + SDATA(tid) = foo(g_idata[tid])-1.0f; __syncthreads(); printf("thread tid=%u reads %f from g_idata[]\n", tid, sdata[tid]); -- cgit v1.3