I ran "./install.sh install" as https://software.intel.com/en-us/articles/sdk-for-opencl-gsg instructed, but I get this error:

Intel Graphics Driver Installer for OpenCL on CentOS

Running on Generic distribution

The installer will download (unless they already exist):
* OpenCL user-mode driver (87MiB) 
* Linux 4.7 kernel (66MiB) 

Setup requires ~10GiB disk space, and takes ~1h on a 4-cores machine.
(~8.9GiB for building the Linux kernel, ~1GiB for other files)

Do you want to continue? [y/n] y
Found existing workspace at /root/intel-opencl...
~/intel-opencl /home/dandan
Installing user mode driver...
Downloading user mode driver...
SRB5.0_linux64.zip already exists, skipping download...
unzip -o SRB5.0_linux64.zip
Archive:  SRB5.0_linux64.zip
  inflating: intel-opencl-cpu-r5.0-63503.x86_64.rpm  
  inflating: intel-opencl-cpu-r5.0-63503.x86_64.tar.xz  
  inflating: intel-opencl-cpu-r5.0-63503.x86_64.tar.xz.sig  
  inflating: intel-opencl-devel-r5.0-63503.x86_64.rpm  
  inflating: intel-opencl-devel-r5.0-63503.x86_64.tar.xz  
  inflating: intel-opencl-devel-r5.0-63503.x86_64.tar.xz.sig  
  inflating: intel-opencl-r5.0-63503.x86_64.rpm  
  inflating: intel-opencl-r5.0-63503.x86_64.tar.xz  
  inflating: intel-opencl-r5.0-63503.x86_64.tar.xz.sig  
  inflating: vpg_ocl_linux_rpmdeb.public  
Looking for previously installed user-mode driver...
rm -rf /etc/ld.so.conf.d/libintelopencl.conf
rm -rf /etc/OpenCL/vendors/intel.icd
rm -rf /etc/profile.d/libintelopencl.sh
rm -rf /opt/intel/opencl
tar -C intel-opencl -Jxf intel-opencl-cpu-r5.0-63503.x86_64.tar.xz
tar -C intel-opencl -Jxf intel-opencl-devel-r5.0-63503.x86_64.tar.xz
tar -C intel-opencl -Jxf intel-opencl-r5.0-63503.x86_64.tar.xz
cp -R intel-opencl/* /
ldconfig
Downloading Linux kernel 4.7...
linux-4.7.tar.xz already exists, skipping download...
Installing Linux kernel 4.7...
linux-4.7 kernel directory exists, skipping extract...
~/intel-opencl/linux-4.7 ~/intel-opencl /home/dandan
patch -p1 < /opt/intel/opencl/kernel-4.7.patch
can't find file to patch at input line 19
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|From 18c0bd9224ea2764698bc42ceba47f1be4ed7967 Mon Sep 17 00:00:00 2001
|From: =?UTF-8?q?Micha=C5=82=20Winiarski?= <michal.winiarski@intel.com>
|Date: Wed, 21 Oct 2015 13:11:24 +0200
|Subject: [PATCH 01/17] drm/i915: Add L3_LRA_1 Register to cmdparser whitelist
|MIME-Version: 1.0
|Content-Type: text/plain; charset=UTF-8
|Content-Transfer-Encoding: 8bit
|
|Change-Id: I4fba318061b175d481ea2f2f02593d70930daf23
|Signed-off-by: Michał Winiarski <michal.winiarski@intel.com>
|---
| drivers/gpu/drm/i915/i915_cmd_parser.c |    1 +
| 1 files changed, 1 insertions(+), 0 deletions(-)
|
|diff --git a/drivers/gpu/drm/i915/i915_cmd_parser.c b/drivers/gpu/drm/i915/i915_cmd_parser.c
|index a337f33..85f0f2f 100644
|--- a/drivers/gpu/drm/i915/i915_cmd_parser.c
|+++ b/drivers/gpu/drm/i915/i915_cmd_parser.c
--------------------------
File to patch: 

What should I enter at "FIle to patch"? As I realized, it said "Intel Graphics Driver Installer for OpenCL on CentOS", but I'm running on Debian, which is similar enough to Ubuntu. How can I fix this?

Thanks!

I tried to install SRB 5.0 opencl driver for Intel(R) Celeron(R) CPU  J1900, but clinfo can not find the gpu device , only the cpu device can be found. (see the attachment clinfo_1123.txt).

here is the output from lspci -nn -s 0:02.0

00:02.0 VGA compatible controller [0300]: Intel Corporation Atom Processor Z36xxx/Z37xxx Series Graphics & Display [8086:0f31] (rev 0e)

Any suggestion?

thanks

I've found the OpenCL runtime application for SUSE linux,

but it was difficult to get the OpenCL SDK application for SUSE linux.

There were only for CentOS and Ubuntu.

How can I download the OpenCL SDK applicatioin for SUSE linux?

Moreover, is the copyright of the OpenCL applications in Intel for open source?

I'm planning to set up opencl on my linux. After looking around, it looks like the steps should be installing the opencl driver and then the sdk, but there're also conflicting information.

First of all, it seems that the opencl drivers only support up to 7th generation cpu (https://software.intel.com/en-us/articles/opencl-drivers#latest_linux_dr...) and I'm using 8th generation, and according to this thread (https://software.intel.com/en-us/forums/opencl/topic/390630), I only need to install the SDK.

On some other random sites, they suggest installing just the runtime package or just some generic ubuntu packages (https://askubuntu.com/questions/850281/opencl-on-ubuntu-16-04-intel-sand...).

What should I do?

Hi,

I have installed OpenCL on my linux (CentOS 7.4) machine and also installed the code builder plug-in (version 7.0.0.2568) with Eclipse platform.

But, I am not finding the OpenCL debugger in the code-builder menu, I could only see OpenCL Kernel Development, along with Help and platform info.

Can someone help me, whether I missed anything here? Or is it not supported on Eclipse platform?

Thanks & Regards, Venkat

Hello!

I tried to debug my first OpenCL application with the Visual Studio Plugin "OpencC Code Builder" (Intel SDK for OpenCL Application 2017 Beta, 7.0.0.2567). The Data View does not work properly because the raw data associated with the object buffer does not appear inside the view. The history combo box is also empty.

Is there a solution to the problem with Data View?

Thanks!

Hi !

I have made this nice OpenCL kernel source code file to C/C++ char string converter:

https://github.com/MatejGomboc/OpenC...ring-converter

I hope you will find it useful !

Best regards.

Hello all,

I am a total beginner in Opencl/ GPU computing and I decided to start off by working on the Vector addition using Opencl (V 1.2). I was able to successfully run the code and then, decided to move further and write the code to compute the execution time of Vector addition on GPU using the Opencl Event. However, I ran into a segmentation fault of some sort and being a newbie to this stuff, despite trying for quite some time to fix this on my own, I'm not able to get over the error.

Furthermore, I'm still learning the various terms and methods involved in Opencl and thus, fixing the error has been difficult for me.  I really want to move forward with my opencl coding, so any help in this regard would be much appreciated. I've attached screenshots of portions of the code for reference.

​Thanks

Hi 

i am sorry if the issue has been discussed before, but i couldn't find the solution the problem i am facing. 

I am running a simple vector_add openCL code through emulator for Intel FPGA SDK for openCL. The issue i am facing is that the program cannot find CL_DEVICE("ERROR: CL_DEVICE_NOT_FOUND "). where as if i run a helloworld program using intel SDK for OpenCL, it runs seamlessly.

the example code is here 

https://www.altera.com/content/dam/altera-www/global/en_US/others/suppor...

Please help me make devices visible for intel FPGA SDK for openCL

Regards

jeveria

I'm observing an ~25% performance drop in a kernel that is ~4.5 msecs of pure compute with one final write to an OpenGL render buffer.

The measurement is via a cl_event.  

Whether the kernels are loaded as binaries or as source doesn't matter.

Did something change in the OpenCL driver with Vulkan integration?  

I'm wondering how performance can vary so much when the kernel hasn't changed and its only interaction with "external" resources is limited to a single store to a 1 megapixel renderbuffer?

OpenCL support appears to be broken in this new driver.

I installed .4944 several times and rebooted.  No OpenCL.

Tested on an HD 630.

https://communities.intel.com/message/526049#526049

Hi, mates,

i searched almost everywhere, who know where to download it or the new version? all links to download the sdk in intel.com are invalid.

 Thanks.

I ran into a bug when I use the Intel OpenCL SDK (verison 7.0.0.2567) on Visual Studio 2015. I defined a struct "Obj" of an array of 5 int. Then I pass five variable of type "Obj" to my opencl kernel program as "__private" variables. If I build my kernel in debug mode (with argument "-g -s filepath"), then some of my variables can not be passed correctly. The code example is on https://github.com/flm8620/intel_opencl_bug/blob/master/main.cpp

Kernel program is:

struct Obj {
    int a[5];
};

__kernel void test(
    __global double* output,
    __private struct Obj param1,
    __private struct Obj param2,
    __private struct Obj param3,
    __private struct Obj param4,
    __private struct Obj param5
    //__private struct Obj param6
)
{
    int gl = get_global_id(0);
    const int N = 5;
    if (gl == 0) {
        for (int i = 0; i < N; i++)
            output[i] = param1.a[i];
        for (int i = 0; i < N; i++)
            output[i + N * 1] = param2.a[i];
        for (int i = 0; i < N; i++)
            output[i + N * 2] = param3.a[i];
        for (int i = 0; i < N; i++)
            output[i + N * 3] = param4.a[i];
        for (int i = 0; i < N; i++)
            output[i + N * 4] = param5.a[i];
    }
}

To verify this, I copy passed variables in kernel to a output variable and print it out at host side:

int main() {
    bool debug = true;
    find_cl(debug);
    const int N = 5;
    struct Obj
    {
        cl_int a[N];
    };
    Obj param1{ {1,12,123,1234,12345} };
    Obj param2{ {1,12,123,1234,12345} };
    Obj param3{ {1,12,123,1234,12345} };
    Obj param4{ {1,12,123,1234,12345} };
    Obj param5{ {1,12,123,1234,12345} };

    double output[N * 5];

    cl::Buffer output_b(context, CL_MEM_WRITE_ONLY, N * 5 * sizeof(double));

    kernel.setArg(0, output_b);
    kernel.setArg(1, param1);
    kernel.setArg(2, param2);
    kernel.setArg(3, param3);
    kernel.setArg(4, param4);
    kernel.setArg(5, param5);


    cl::CommandQueue queue(context, device);

    queue.enqueueNDRangeKernel(kernel, cl::NullRange, { size_t(1) }, { size_t(1) });
    queue.enqueueReadBuffer(output_b, false, 0, N * 5 * sizeof(double), output);
    queue.finish();

    for (int i = 0; i < N * 5; i++)
        std::cout << output[i] << std::endl;

    return 0;
}

The output is 

Detected 3 platforms :
NVIDIA CUDA
Intel(R) OpenCL
Experimental OpenCL 2.1 CPU Only Platform
Found CPU platform: Intel(R) OpenCL, has devices:
  1: Intel(R) Xeon(R) CPU E3-1231 v3 @ 3.40GHz
Use device: Intel(R) Xeon(R) CPU E3-1231 v3 @ 3.40GHz
1
12
123
1234
12345
1
12
123
1234
12345
1
12
123
1234
12345
1
12
123
1234
0
123
1234
0
0
3.90955e+07

I don't think it's linked to struct align because the first four arguments are correctly passed. If I pass six instead of five variables, the program will crash.

If I change the first line in main() to

bool debug = false;

Then everything works. 

Hi,

encountered some strange bug and prepared the demonstration.

1. Simple kernel:
--------------------------

#pragma OPENCL EXTENSION cl_intel_printf : enable

__kernel void glitch( 
	__local uint *scratch
) {

	uint k,v;
	k=0;
	v=0; 

	printf(" START\n");
	uint id = (uint)get_global_id(0);
	if (id==0) {
		scratch[k] = v;
	}
	barrier( CLK_LOCAL_MEM_FENCE );
	while ( scratch[k]<2 ) {
		if (id==0) {
			scratch[k]++;
		}
		barrier( CLK_LOCAL_MEM_FENCE );
	}
	printf(" FINISH\n");
}

---------------------------

2. Run with natural grid dimensions: single group 16x1x1 , or 256x1x1
and local memory size, for example, 1024

3. what expected:
the items are started (you see 16 lines of START)
the first of them increments the counter in local memory
and all items are finished (you see 16 lines of FINISH)

4. but FINISH repeat forever!
maybe some stack corruption?
When you switch off the printf and remove its pragma,
the kernel simply does not return which is the same.

---------------------------
cases:

Intel CPU with "-cl-opt-disable" runs fine

Nvidia GPU - no problem with or without optimization.

Intel CPU with optimization and workgroup sizes 8 or 1 runs fine. 2, 4, 16, 64, 256 are bad.
(8 may be the special case: the number of threads in my CPU, i7-4790)

Intel OpenCL SDK and CPU driver are re-installed today.

---------------------------

additional demonstration:
I added two #define's (see 1st and 2nd lines of the code)

GLITCH=0 whows how to overcome with nonzero index for local memory array
and initial value of counter from get_global_id(1)

zero index is bad.

initial value of counter = explicitly written 0 is bad.

GLITCH2 shows that the item counter does not go after 8:

 FINISH: z=0 y=0 x=0
 FINISH: z=0 y=0 x=1
 FINISH: z=0 y=0 x=2
 FINISH: z=0 y=0 x=3
 FINISH: z=0 y=0 x=4
 FINISH: z=0 y=0 x=5
 FINISH: z=0 y=0 x=6
 FINISH: z=0 y=0 x=7
 FINISH: z=0 y=0 x=0
 FINISH: z=0 y=0 x=1
 FINISH: z=0 y=0 x=2
 FINISH: z=0 y=0 x=3
 FINISH: z=0 y=0 x=4
 FINISH: z=0 y=0 x=5
 FINISH: z=0 y=0 x=6
 FINISH: z=0 y=0 x=7
 FINISH: z=0 y=0 x=0
...etc.

but if you don't call printf with x,y,z, then again everything is OK.

-----
Rather strange.
Would like to know if it works for you?
I don't see any rough errors in my code...

Regards, Petr

I am trying to use the CodeBuilder for Eclipse Neon on ubuntu 16.04.3.

I have installed the driver and kernel patch (not sure if it is really required though) and I can compile opencl kernels and programs. However I would like to use the SDK tools for eclipse. When I install the intel_sdk_for_opencl_2017_7.0.0.25.11_x64 sdk for eclipse and put the plugin in the dropins folder. I get the new toolbars etc but it complains of being unable to find machine or device etc. such as:

"Failed to update machine list:

Could not load required libraries; please make sure to set the correct path under the Code Builder for OpenCL preference page"

and similar errors for platforms and devices.

I have the paths pointing to:

OpenCL Preferences:

OpenCL Bin Directory: /opt/intel/opencl/SDK/bin

OpenCL Lib Directory: /opt/intel/opencl/SDK/lib64

in the eclipse preferences/CodeBuilder for OpenCL menu dialog.

I noticed that there are .dll and .exe files included in the linux install folders which is odd as they also prevent me from using them directly in a standalone command line fashion.

Is there something wrong with the distribution? or do I need to do something to get this to work.

cheers and thanks in advance.

The new Intel OpenCL GPU driver (NEO) driver is now available on github (https://github.com/intel/compute-runtime). Going forward this driver replaces both the SRB5 closed-source driver (aka Classic) and the Beignet open-source driver.

To learn about our motivation for this change, visit http://01.org/compute-runtime. For more information on this driver, source code, and binary releases, please visit our github page. We invite you to post any questions and issues there as well.

Hi ,

I have a question related to the compiler generated information.

I have compiled the kernel and see the following information.I am working on Gen9 GT2 system.

1> Maximum work-group size: 256

1> Compiler work-group size: (0, 0, 0)

1> Local memory size: 0

1> Preferred multiple of work-group size: 8

1> Minimum amount of private memory: 768

1> Amount of spill memory used by the kernel: 1024

I assume that private memory is mapped to the registers by the compiler. My workgroup size is "1". each EU thread has 128 GRFs and each GRF size is 256-bits . the total memory size of the private space is 4096 bytes. I see my kernel requires only 768 bytes of the GRF space which is much less than 4096 bytes , but compiler is generating a spill code of 1024 bytes. 

could you please explain me why this is happening?

Best Regards,

Rajesh

Hello,

 I have been trying to install the OpenCL driver on Ubuntu 16.04 using the scripts on Intel's website and ending up with a few errors. And I'm quite a noob to linux, so it will be nice if someone can help me out on this. My system is an Intel N4200 with 505 HD Graphiics which the script claims support. Here are the errors:

Downloading user mode driver...
SRB5.0_linux64.zip already exists, skipping download...
unzip -o SRB5.0_linux64.zip
Archive:  SRB5.0_linux64.zip
  inflating: intel-opencl-cpu-r5.0-63503.x86_64.rpm  
  inflating: intel-opencl-cpu-r5.0-63503.x86_64.tar.xz  
  inflating: intel-opencl-cpu-r5.0-63503.x86_64.tar.xz.sig  
  inflating: intel-opencl-devel-r5.0-63503.x86_64.rpm  
  inflating: intel-opencl-devel-r5.0-63503.x86_64.tar.xz  
  inflating: intel-opencl-devel-r5.0-63503.x86_64.tar.xz.sig  
  inflating: intel-opencl-r5.0-63503.x86_64.rpm  
  inflating: intel-opencl-r5.0-63503.x86_64.tar.xz  
  inflating: intel-opencl-r5.0-63503.x86_64.tar.xz.sig  
  inflating: vpg_ocl_linux_rpmdeb.public  
Looking for previously installed user-mode driver...
rm -rf /etc/ld.so.conf.d/libintelopencl.conf
rm -rf /etc/OpenCL/vendors/intel.icd
rm -rf /etc/profile.d/libintelopencl.sh
rm -rf /opt/intel/opencl
tar -C intel-opencl -Jxf intel-opencl-cpu-r5.0-63503.x86_64.tar.xz
tar: ./opt/intel/opencl/libtbbmalloc.so: Cannot create symlink to ‘libtbbmalloc.so.2’: Operation not permitted
tar: ./opt/intel/opencl/libtbb.so: Cannot create symlink to ‘libtbb.so.2’: Operation not permitted
tar: Exiting with failure status due to previous errors

Thanks in advance.

Hi,

I get a segflt when running Intel OpenCL code samples.  In fact, i have never successfully run _any_ OpenCL examples.

I'm using: 
S/W: Fedora 25 running unmodified Fedora kernel 4.12.11-200.fc25.x86_64
H/W : Processor: E3-1505L v5 (Skylake)  PCH: CM238

Web sites I've been referring to for help:

https://software.intel.com/en-us/articles/opencl-drivers
https://software.intel.com/en-us/intel-opencl-support/code-samples

The packages/versions of Intel OpenCL I’m using are:
 
intel-opencl-cpu-r4.1-61547.x86_64
intel-opencl-devel-r4.1-61547.x86_64
intel-opencl-r4.1-61547.x86_64

I have basically followed  http://registrationcenter-download.intel.com/akdlm/irc_nas/11396/intel-opencl-4.1-installation.pdf
However, It’s unclear if I need to patch my kernel.  I tried applying the supplied patch (kernel-4.7.patch), but  applying it 
to my kernel (4.12.11) appears to be very difficult or impossible.

Examples of some of the programs I unsuccessfully tried:
 
 1.  https://software.intel.com/sites/default/files/managed/59/01/intel_ocl_gemm.zip  
After unzipping, it compiles OK; but running it gives a Segflt:
 
sh-4.3# ./GEMM -p 0 -d 1
Platforms (1):
    [0] Intel(R) OpenCL [Selected]
Devices (2):
    [0] Intel(R) Xeon(R) CPU E3-1505L v5 @ 2.00GHz
    [1] Intel(R) HD Graphics [Selected]
Build program options: "-DT=float -DTILE_SIZE_M=1 -DTILE_GROUP_M=16 -DTILE_SIZE_N=128 -DTILE_GROUP_N=1 -DTILE_SIZE_K=8"
Segmentation fault (core dumped)

sh-4.3# ./GEMM -p 0 -d 0
Platforms (1):
    [0] Intel(R) OpenCL [Selected]
Devices (2):
    [0] Intel(R) Xeon(R) CPU E3-1505L v5 @ 2.00GHz [Selected]
    [1] Intel(R) HD Graphics
Segmentation fault (core dumped)
 

 2.  intel_ocl_median_filter.zip
Same thing- compiles, but segflt when running.
 

 3.  intel_ocl_tone_mapping_multidevice.zip  
Same thing – segflt when running:
 
sh-4.3# pwd
/root/opencl.examples/tone.mapping/ToneMappingMultiDevice
sh-4.3# ./ToneMappingMultiDevice  -t gpu
Platforms (1):
    [0] Intel(R) OpenCL [Selected]
CalculateFStopsParameter result = 0.379187
Trying to run on Processor Graphics
Segmentation fault (core dumped)
sh-4.3# ./ToneMappingMultiDevice  -t cpu
Platforms (1):
    [0] Intel(R) OpenCL [Selected]
CalculateFStopsParameter result = 0.379187
Trying to run on CPU
Segmentation fault (core dumped)

Question 1 Is this version expected to work with my hardware/CPU/GPU ?
Question 2 Is this expected to work with my 4.12, unpatched, kernel ?
Question 3 wondering if moving to a later version of the above three OpenCL packages would make a difference ?
 
Thanks a lot!
Martin Rogers

Hi,

Today I downloaded the latest version of Intel SDK for OpenCL. I very like it. It's nice to look inside the device.

I have found couple strange thing with the Command Queue View.
1. It shows commands a bit strange way. I believe there is some problems with the font.
2. Some commands are found at 2 stages at the same time: WRITE_IMAGE(4), MAP_BUFFER(8), MAP_BUFFER(11), MAP_BUFFER(13). It's a bit hard to recognize them with the font issue. Is it expected?
3. Time after time I've got garbage in mapped region, when working kernel size is small (= kernel loading/submitting time is much bigger than kernel running time). The buffer contains bytes like "_Z55intel_sub_group_avc_ref_set_motion_vector_cost_functionmDv2_jh33intel_sub_group_avc_re". Looks like a wrong region was mapped. No error/warnings were return from any OpenCL function before getting the garbage. The OpenCL log view from MSVC contains only CL_SUCCESS.

Image may be NSFW.
Clik here to view.

A brief description of my application. The application creates an asynchronous command queue on the Intel device, writes YUV data to 8-bit frames allocated on the device (asynchronous clEnqueueWriteImage), asynchronously runs kernel vme.cl downloaded from the Intel site, and maps buffers with results into the host memory (asynchronous clEnqueueMapBuffer). All cl_event objects are passed correctly between the stages. At the end applications synchronizes with latest cl_event and reads the data. Commands are submitted from various threads. Is it possible that OpenCL library has problems with calls from different threads? When I map buffers with results only once (at the starting of the application), the everything runs fine. But it contradicts with OpenCL description of clEnqueueMapBuffer: "The behavior of writes by a kernel executing on a device to a mapped region of a memory object is undefined". So I have to map buffers after the kernel execution and to unmap them after finishing using.

My PC is running under regular Windows 10 Pro 64 bit, 24Gb RAM, i7 6700, GPU driver version is 22.20.16.4836. I'm running MSVC 2015 version 14.0.25420.01 Update 3.

I'll be very appreciated for any help.

Regards,
Victor