I have an OpenCL Kernel with multiple work items. Let's assume for discussion, that I have a 2-D Workspace with x*y elements working on an equally sized, but sparce, array of input elements. Few of these input elements produce a result, that I want to keep, most don't. I want to enqueue another kernel, that only takes the kept results as an input.
Is it possible in OpenCL to append results to some kind of list to pass them as input to another Kernel or is there a better idea to reduce the volume of the solution space? Furthermore: Is this even a good question to ask with the programming model of OpenCL in mind?
What I would do if the amount of result data is a small percentage (ie: 0-10%) is use local atomics and global atomics, with a global counter.
Data interface between kernel 1 <----> Kernel 2:
int counter //used by atomics to know where to write
data_type results[counter]; //used to store the results
Kernel1:
Create a kernel function that does the operation on the data
Work items that do produce a result:
Save the result to local memory, and ensure no data races occur using local atomics in a local counter.
Use the work item 0 to save all the local results back to global memory using global atomics.
Kernel2:
Work items lower than "counter" do work, the others just return.
Related
I have written an openCL kernel that takes 25million points and checks them relative to two lines, (A & B). It then outputs two lists; i.e. set A of all of the points found to be beyond line A, and vice versa.
I'd like to run the kernel repeatedly, updating the input points with each of the line results sets in turn (and also updating the checking line). I'm guessing that reading the two result sets out of the kernel, forming them into arrays and then passing them back in one at a time as inputs is quite a slow solution.
As an alternative, I've tested keeping a global index in the kernel that logs which points relate to which line. This is updated at each line checking cycle. During each iteration, the index for each point in the overall set is switched to 0 (no line), A or B or so forth (i.e. the related line id). In subsequent iterations only points with an index that matches the 'live' set being checked in that cycle (i.e. tagged with A for set A) are tested further.
The problem is that, in each iteration, the kernels still have to check through the full index (i.e. all 25m points) to discover wether or not they are in the 'live' set. As a result, the speed of each cycle does not significantly improve as the size of the results set decrease over time. Again, this seems a slow solution; whilst avoiding passing too much information between GPU and CPU it instead means that a large number of the work items aren't doing very much work at all.
Is there an alternative solution to what I am trying to do here?
You could use atomics to sort the outputs into two arrays. Ie if we're in A then get my position by incrementing the A counter and put me into A, and do the same for B
Using global atomics on everything might be horribly slow (fast on amd, slow on nvidia, no idea about other devices) - instead you can use a local atomic_inc in a 0'd local integer to do exactly the same thing (but for only the local set of x work-items), and then at the end do an atomic_add to both global counters based on your local counters
To put this more clearly in code (my explanation is not great)
int id;
if(is_a)
id = atomic_inc(&local_a);
else
id = atomic_inc(&local_b);
barrier(CLK_LOCAL_MEM_FENCE);
__local int a_base, b_base;
int lid = get_local_id(0);
if(lid == 0)
{
a_base = atomic_add(a_counter, local_a);
b_base = atomic_add(b_counter, local_b);
}
barrier(CLK_LOCAL_MEM_FENCE);
if(is_a)
a_buffer[id + a_base] = data;
else
b_buffer[id + b_base] = data;
This involves faffing around with atomics which are inherently slow, but depending on how quickly your dataset reduces it might be much faster. Additionally if B data is not considered live, you can omit getting the b ids and all the atomics involving b, as well as the write back
I am parallelizing a certain dynamic programming problem using AVX2/SSE instructions.
In the main iteration of my calculation, I calculate column in matrix where each cell is a structure of AVX2 registers (_m256i). I use values from the previous matrix column as input values for calculating the current column. Columns can be big, so what I do is I have an array of structures (on stack), where each structure has two _m256i elements.
Structure:
struct Cell {
_m256i first;
_m256i second;
};
An then I have array like this: Cell prevColumn [N]. N will tipically be few hundreds.
I know that _m256i basically represents an avx2 register, so I am wondering how should I think about this array, how does it behave, since N is much larger than 16 (which is number of avx registers)? Is it a good practice to create such an array, or is there some better approach that i should use when storing a lot of _m256i values that are going to be reused real soon?
Also, is there any aligning I should be doing with this structures? I read a lot about aligning, but I am still not sure how and when to do it exactly.
It's better to structure your code to do everything it can with a value before moving on. Small buffers that fit in L1 cache aren't going to be too bad for performance, but don't do that unless you need to.
I think it's more typical to write your code with buffers of int [] type, rather than __m256i type, but I'm not sure. Either way works, and should get the compile to generate efficient code. But the int [] way means less code has to be different for the SSE, AVX2, and AVX512 version. And it might make it easier to examine things with a debugger, to have your data in an array with a type that will get the data formatted nicely.
As I understand it, the load/store intrinsics are partly there as a cast between _m256i and int [], since AVX doesn't fault on unaligned, just slows down on cacheline boundaries. Assigning to / from an array of _m256i should work fine, and generate load/store instructions where needed, otherwise generate vector instructions with memory source operands. (for more compact code and fewer fused-domain uops.)
After compacting an array(putting required elements from an input array into an output array) by doing a scan operation, there might be some empty spaces left in the output(compacted) array in a contiguous form after the required elements are placed. Is there a way to free these empty spaces in the OpenCL kernel code itself without going back in the host(just for the sake of deleting)...?
for eg I have an input array of 100 elements with some no.s greater than 50 and some of them less than 50 and want to store the no.s more than 50 in a different array and do further processing only on those elements in that array, and I don't know the size of this output array since I don't know how many no.s are actually greater than 50(so I declare the size of this array to be 100)... then after performing a scan I get the output array with all elements more than 50... but there might be some continuous spaces empty in the output array after the storage of these elements... then how do we delete these spaces... Is there a way of doing this in the kernel code itself...? Or do we have to come back in the Host code for this...?
How do we deal with such compacted arrays to do further processing if we can't delete the remaining spaces in the kernel code itself and also if we don't want to go back in the host code..?
There is no simple solution to your problem I'm afraid.
What I think you might do, is to have a counter of the elements in each array. You can increment the counter first locally with atomic_inc() and then globally with atomic_add().
This way at the end of your kernel execution the total number of elements in each array will be present.
You can also use this atomic operation as an index for the array. This way you can write to the output without any "hole" in your array. However you will probably lose some speed due to abusing of atomic operations I'm afraid.
I am implementing an algorithm with OpenCL. I will loop in C++ many times and call a same OpenCL kernel each time. The kernel will generate the input data of next iteration and the number of these data. Currently, I read back this number in each loop for two usages:
I use this number to decide how many work items I need for next loop; and
I use this number to decide when to exit the loop (when the number is 0).
I found the reading takes most of time of the loop. Is there any way to avoid it?
Generally speaking, if you need to call a kernel repeatedly, and the exit condition is dependent to the result generated by the kernel (not fixed number loops), how can you do it efficiently? Is there anything like the occlusion query in OpenGL that you can just do some query instead of reading back from GPU?
Reading a number back from a GPU Kernel will always take 10s - 1000s microseconds or more.
If the controlling number is always reducing, you can keep in global memory, and test it against the global id and decide if the kernel does work or not on each iteration. Use a global memory barrier to sync all the threads ...
kernel void x(global int * the_number, constant int max_iterations, ... )
{
int index = get_global_id(0);
int count = 0; // stops an infinite loop
while( index < the_number[0] && count < max_iterations )
{
count++;
// loop code follows
....
// Use one thread decide what to do next
if ( index == 0 )
{
the_number[0] = ... next value
}
barrier( CLK_GLOBAL_MEM_FENCE ); // Barrier to sync threads
}
}
You have a couple of options here:
If possible, you can simply move the loop and the conditional into the kernel? Use a scheme where additional work items do nothing depending on the input for the current iteration.
If 1. isn't possible, I would recommend that you store the data generated by the "decision" kernel in a buffer and use that buffer to "direct" your other kernels.
Both these options will allow you to skip the readback.
I'm just finishing up some research where we had to tackle this exact problem!
We discovered a couple of things:
Use two (or more) buffers! Have the first iteration of the kernel
operate on data in b1, then the next on b2, then on b1 again. In
between each kernel call, read back the result of the other buffer
and check to see if it's time to stop iterating. Works best when the kernel takes longer than a read. Use a profiling tool to make sure you aren't waiting on reads (and if you are, increase the number of buffers).
Over shoot! Add a finishing check to each kernel, and call it
several (100s) of times before copying data back. If your kernel is
low-cost, this can work very well.
I'm trying to write a histogram kernel in OpenCL to compute 256 bin R, G, and B histograms of an RGBA32F input image. My kernel looks like this:
const sampler_t mSampler = CLK_NORMALIZED_COORDS_FALSE |
CLK_ADDRESS_CLAMP|
CLK_FILTER_NEAREST;
__kernel void computeHistogram(read_only image2d_t input, __global int* rOutput,
__global int* gOutput, __global int* bOutput)
{
int2 coords = {get_global_id(0), get_global_id(1)};
float4 sample = read_imagef(input, mSampler, coords);
uchar rbin = floor(sample.x * 255.0f);
uchar gbin = floor(sample.y * 255.0f);
uchar bbin = floor(sample.z * 255.0f);
rOutput[rbin]++;
gOutput[gbin]++;
bOutput[bbin]++;
}
When I run it on an 2100 x 894 image (1,877,400 pixels) i tend to only see in or around 1,870,000 total values being recorded when I sum up the histogram values for each channel. It's also a different number each time. I did expect this since once in a while two kernels probably grab the same value from the output array and increment it, effectively cancelling out one increment operation (I'm assuming?).
The 1,870,000 output is for a {1,1} workgroup size (which is what seems to get set by default if I don't specify otherwise). If I force a larger workgroup size like {10,6}, I get a drastically smaller sum in my histogram (proportional to the change in workgroup size). This seemed strange to me, but I'm guessing what happens is that all of the work items in the group increment the output array value at the same time, and so it just counts as a single increment?
Anyways, I've read in the spec that OpenCL has no global memory syncronization, only syncronization within local workgroups using their __local memory. The histogram example by nVidia breaks up the histogram workload into a bunch of subproblems of a specific size, computes their partial histograms, then merges the results into a single histogram after. This doesn't seem like it'll work all that well for images of arbitrary size. I suppose I could pad the image data out with dummy values...
Being new to OpenCL, I guess I'm wondering if there's a more straightforward way to do this (since it seems like it should be a relatively straightforward GPGPU problem).
Thanks!
As stated before, you write into a shared memory unsynchronized and non atomic. This leads to errors. If the picture is big enough, I have a suggestion:
Split your work group into a one dimensional one for cols or rows. Use each kernel to sum up the histogram for the col or row and afterwards sum it globally with atomic atom_inc. This brings the most sum ups in private memory which is much faster and reduces atomic ops.
If you work in two dimensions you can do it on parts of the picture.
[EDIT:]
I think, I have a better answer: ;-)
Have a look to: http://developer.download.nvidia.com/compute/opencl/sdk/website/samples.html#oclHistogram
They have an interesting implementation there...
Yes, you're writing to a shared memory from many work-items at the same time, so you will lose elements if you don't do the updates in a safe way (or worse ? Just don't do it). The increase in group size actually increases the utilization of your compute device, which in turn increases the likelihood of conflicts. So you end up losing more updates.
However, you seem to be confusing synchronization (ordering thread execution order) and shared memory updates (which typically require either atomic operations, or code synchronization and memory barriers, to make sure the memory updates are visible to other threads that are synchronized).
the synchronization+barrier is not particularly useful for your case (and as you noted is not available for global synchronization anyways. Reason is, 2 thread-groups may never run concurrently so trying to synchronize them is nonsensical). It's typically used when all threads start working on generating a common data-set, and then all start to consume that data-set with a different access pattern.
In your case, you can use atomic operations (e.g. atom_inc, see http://www.cmsoft.com.br/index.php?option=com_content&view=category&layout=blog&id=113&Itemid=168). However, note that updating a highly contended memory address (say, because you have thousands of threads trying all to write to only 256 ints) is likely to yield poor performance. All the hoops typical histogram code goes through are there to reduce the contention on the histogram data.
You can check
The histogram example from AMD Accelerated Parallel Processing (APP) SDK.
Chapter 14 - Image Histogram of OpenCL Programming Guide book (ISBN-13: 978-0-321-74964-2).
GPU Histogram - Sample code from Apple