<html>
<head>
<base href="https://llvm.org/bugs/" />
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW --- - CUDA __device__ lambdas generating incorrect code"
href="https://llvm.org/bugs/show_bug.cgi?id=26483">26483</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>CUDA __device__ lambdas generating incorrect code
</td>
</tr>
<tr>
<th>Product</th>
<td>clang
</td>
</tr>
<tr>
<th>Version</th>
<td>unspecified
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</td>
</tr>
<tr>
<th>OS</th>
<td>Linux
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>normal
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>-New Bugs
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedclangbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>justin.lebar@gmail.com
</td>
</tr>
<tr>
<th>CC</th>
<td>llvm-bugs@lists.llvm.org
</td>
</tr>
<tr>
<th>Classification</th>
<td>Unclassified
</td>
</tr></table>
<p>
<div>
<pre>The following CUDA code, which passes a __device__ functor to a kernel,
compiles fine but does not give the correct output.
When I compile and run it, I get the following output:
Error 3: invalid device function
Error: 0 2883584
Error: 1 0
Error: 2 2883584
Error: 3 0
Error: 4 3105024
Error: 5 0
Error: 6 0
Error: 7 0
Error: 8 0
Error: 9 0
Error: 10 0
Error: 11 0
Error: 12 0
Error: 13 0
Error: 14 0
Error: 15 0
The "error 3 invalid device function" is probably the root of the problem; it
seems like we may not be running any device code at all.
#include <cuda_runtime.h>
#include <cuda.h>
#include <cstdio>
template<class Functor>
__global__ void foo(const Functor f) {
f(blockIdx.x*blockDim.x+threadIdx.x);
}
template<class Functor>
void run(const Functor& f) {
foo<Functor> <<<256,256>>> (f);
}
int main() {
int* h_a = new int[256*256];
int d_c = 9;
int* d_a;
cudaMalloc(&d_a,256*256*sizeof(int));
run([=] (int i) __device__ {
d_a[i] = d_c;
});
cudaMemcpy(h_a,d_a,256*256*sizeof(int),cudaMemcpyDeviceToHost);
printf("Error 3: %s\n",cudaGetErrorString(cudaGetLastError()));
for(int i=0;i<16;i++)
if(h_a[i]!=d_c)
printf("Error: %i %i\n",i,h_a[i]);
}</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>