Databases
StackOverflow2013
Postsdbo
POOpenCL. Weird bug with array addressing/pointers
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POOpenCL. Weird bug with array addressing/pointers
    primarykey
    data
    text
    <p>I have this simple kernel for testing.</p> <pre><code>__kernel void nfa(__global const int *a, __global int *output) { output[0] = a[0]; } </code></pre> <p>Note: This is running on a cpu, and memory is probably on the host. It results in this error.</p> <p><strong>* glibc detected <em></strong> ./program: malloc(): smallbin double linked list corrupted: 0x0000000000a4a540 *</em>*</p> <p>I suspect somehow this is corrupting a part of the program, since it's accessing host memory. But as far I know all memory allocated correctly. It is on the stack, but stays in scope while running.</p> <p>However if I do this:</p> <pre><code> __kernel void nfa(__global const int *a, __global int *output) { a = a; output[0] = a[0]; } </code></pre> <p>It results in the answer 2, which is correct since a is an array with [2, 4, 8];</p> <p>An allocation to its self fixed the issue...</p> <p>This is also fine, resulting in 4.</p> <pre><code> __kernel void nfa(__global const int *a, __global int *output) { output[0] = a[1]; } </code></pre> <p>It seems as though just accessing a[0], without assigning to its self causes the problem.</p> <p>Does anyone know what's going on?</p> <p>I'm on linux with AMD OpenCL Drivers(With an intel CPU, but I have a AMD Card).</p> <p>Edit:</p> <p>The code the buffer is created with(Condensed down, there is other code between the array and buffer):</p> <pre><code>int a[3]; a[0] = 2; a[1] = 4; a[2] = 8; cl::Buffer bufferA = cl::Buffer(context, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR , sizeof(int) * 3, &amp;a); </code></pre>
    singulars
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PTQuestion
    1. USUK-AL
    1. USUK-AL
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PO
      singulars
      1. PTAnswer
    1. VO
      singulars
      1. POOpenCL. Weird bug with array addressing/pointers
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. POOpenCL. Weird bug with array addressing/pointers
      1. This table or related slice is empty.
      1. VTUpMod
    1. CODid you allocate the memory with malloc, or with clCreateBuffer()? I believe the latter is required, even if you're on a CPU device.
      singulars
      1. POOpenCL. Weird bug with array addressing/pointers
      1. USDrew Hall
    2. COMemory is on the stack. clCreateBuffer is used(Well its equivalent i'm using c++ bindings), and the clCreateBuffer method is passed the pointer to the array, it uses the flag "CL_MEM_USE_HOST_PT". As far as i'm aware this can result in either using the memory directly, and copying it to the GPU. Since i'm not using the GPU, it is on the host.
      singulars
      1. POOpenCL. Weird bug with array addressing/pointers
      1. USUK-AL
    3. COCan we see your clSetKernelArg() call?
      singulars
      1. POOpenCL. Weird bug with array addressing/pointers
      1. USDrew Hall
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload