Databases
StackOverflow2013
Postsdbo
POCUDA memory transfer issue
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POCUDA memory transfer issue
    primarykey
    data
    text
    <p>I am trying to execute a code which first transfers data from CPU to GPU memory and vice-versa. In spite of increasing the volume of data, the data transfer time remains the same as if no data transfer is actually taking place. I am posting the code.</p> <pre><code>#include &lt;stdio.h&gt; /* Core input/output operations */ #include &lt;stdlib.h&gt; /* Conversions, random numbers, memory allocation, etc. */ #include &lt;math.h&gt; /* Common mathematical functions */ #include &lt;time.h&gt; /* Converting between various date/time formats */ #include &lt;cuda.h&gt; /* CUDA related stuff */ #include &lt;sys/time.h&gt; __global__ void device_volume(float *x_d,float *y_d) { int index = blockIdx.x * blockDim.x + threadIdx.x; } int main(void) { float *x_h,*y_h,*x_d,*y_d,*z_h,*z_d; long long size=9999999; long long nbytes=size*sizeof(float); timeval t1,t2; double et; x_h=(float*)malloc(nbytes); y_h=(float*)malloc(nbytes); z_h=(float*)malloc(nbytes); cudaMalloc((void **)&amp;x_d,size*sizeof(float)); cudaMalloc((void **)&amp;y_d,size*sizeof(float)); cudaMalloc((void **)&amp;z_d,size*sizeof(float)); gettimeofday(&amp;t1,NULL); cudaMemcpy(x_d, x_h, nbytes, cudaMemcpyHostToDevice); cudaMemcpy(y_d, y_h, nbytes, cudaMemcpyHostToDevice); cudaMemcpy(z_d, z_h, nbytes, cudaMemcpyHostToDevice); gettimeofday(&amp;t2,NULL); et = (t2.tv_sec - t1.tv_sec) * 1000.0; // sec to ms et += (t2.tv_usec - t1.tv_usec) / 1000.0; // us to ms printf("\n %ld\t\t%f\t\t",nbytes,et); et=0.0; //printf("%f %d\n",seconds,CLOCKS_PER_SEC); // launch a kernel with a single thread to greet from the device //device_volume&lt;&lt;&lt;1,1&gt;&gt;&gt;(x_d,y_d); gettimeofday(&amp;t1,NULL); cudaMemcpy(x_h, x_d, nbytes, cudaMemcpyDeviceToHost); cudaMemcpy(y_h, y_d, nbytes, cudaMemcpyDeviceToHost); cudaMemcpy(z_h, z_d, nbytes, cudaMemcpyDeviceToHost); gettimeofday(&amp;t2,NULL); et = (t2.tv_sec - t1.tv_sec) * 1000.0; // sec to ms et += (t2.tv_usec - t1.tv_usec) / 1000.0; // us to ms printf("%f\n",et); cudaFree(x_d); cudaFree(y_d); cudaFree(z_d); return 0; } </code></pre> <p>Can anybody help me with this issue?</p> <p>Thanks</p>
    singulars
    1. PO
      singulars
      1. PTAnswer
    1. This table or related slice is empty.
    1. PTQuestion
    1. USTomasz Dzięcielewski
    1. USVaibhav Sundriyal
    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
    2. PO
      singulars
      1. PTAnswer
    1. VO
      singulars
      1. POCUDA memory transfer issue
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. POCUDA memory transfer issue
      1. This table or related slice is empty.
      1. VTApproveEditSuggestion
    1. COI believe `cudaMemcpy` is asynchronous - you would need to run at least a dummy kernel before collecting the second time stamp.
      singulars
      1. POCUDA memory transfer issue
      1. USPaul R
    2. COWell, cudaMemcpy is synchronous and cudaMemcpyAsync is the async variant. I have tested this code on a different C1050 tesla GPU and it worked just fine. But with a different GTX 250, it is not giving a proper result.
      singulars
      1. POCUDA memory transfer issue
      1. USVaibhav Sundriyal
    3. COOK - by asynchronous I mean that the call returns as soon as the DMA transfer is initiated, so the transfer is not complete until some time later. Calling a kernel will then block until any pending DMA transfers are complete. Try adding a dummy kernel and see if it helps.
      singulars
      1. POCUDA memory transfer issue
      1. USPaul R
 

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