Databases
StackOverflow2013
Postsdbo
PO
 

Note that there are some explanatory texts on larger screens.

plurals
  1. PO
    primarykey
    data
    text
    <p>Let's suppose you have a kernel that computes the three values. Each thread in your configuration will compute the three values for each (r,c) pair.</p> <pre><code>__global__ value_kernel(Y, H, X, W) { r = blockIdx.x + Y; c = threadIdx.x + W; chan1value = ... chan2value = ... chan3value = ... } </code></pre> <p>I don't believe you can calculate the sum (completely in parallel, at least) in the above kernel. You won't be able to use += like you have above. You could put it all in one kernel if you have only one thread in each block (row) do the sum and mean, like this...</p> <pre><code>__global__ both_kernel(Y, H, X, W) { r = blockIdx.x + Y; c = threadIdx.x + W; chan1value = ... chan2value = ... chan3value = ... if(threadIdx.x == 0) { ch1RowSum = 0; ch2RowSum = 0; ch3RowSum = 0; for(i=0; i&lt;blockDim.x; i++) { ch1RowSum += chan1value; ch2RowSum += chan2value; ch3RowSum += chan3value; } ch1Mean = ch1RowSum / blockDim.x; ch2Mean = ch2RowSum / blockDim.x; ch3Mean = ch3RowSum / blockDim.x; } } </code></pre> <p>but it's probably better to use the first value kernel and then a second kernel for both sums and means... It's possible to further parallelize the kernel below, and if it's separate you can focus on that when you're ready.</p> <pre><code>__global__ sum_kernel(Y,W) { r = blockIdx.x + Y; ch1RowSum = 0; ch2RowSum = 0; ch3RowSum = 0; for(i=0; i&lt;W; i++) { ch1RowSum += chan1value; ch2RowSum += chan2value; ch3RowSum += chan3value; } ch1Mean = ch1RowSum / W; ch2Mean = ch2RowSum / W; ch3Mean = ch3RowSum / W; } </code></pre>
    singulars
    1. This table or related slice is empty.
    1. POCUDA: embedded for loop kernel
      singulars
      1. PTQuestion
    1. PTAnswer
    1. USjmilloy
    1. USjmilloy
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. POCUDA: embedded for loop kernel
      singulars
      1. PTQuestion
    1. This table or related slice is empty.
    1. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTAcceptedByOriginator
    1. COi mentioned that you can parallelize the sum/mean... what you want is a reduction. (one of many available examples: http://supercomputingblog.com/cuda/cuda-tutorial-3-thread-communication/)
      singulars
      1. PO
      1. USjmilloy
    2. COI notice that you didnt actually use the values of R or C here. Should there be some sort of check to do a No Op if like blockIdx.x*threadIdx.x is greater than blockIdx.x+Y? what would that look like exactly?
      singulars
      1. PO
      1. USDerek
    3. COyou didn't use r or c in your code, either. You have three values that are calculated for each grid location (r,c). You will probably want to have one thread calculate one set of values. When you launch your kernel, launch with r blocks and c threads per block. Then, in each thread you can calculate r and c from the block id and thread id. I assume r and c are used to access input for the value calculations and/or they are used directly. (if you might launch a kernel with more threads or blocks than you need, then sure check your bounds... something like if(r >= #rows || c >= #cols) return;)
      singulars
      1. PO
      1. USjmilloy
 

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