Task is This one

Running Debian Stretch with Beignet. Hardware is an i7-3770 with HD Graphics 4000. CLInfo and BOINC (7.6.22) both found the graphics without any issue. However the task above crashed straight away.

Unlike the other message thread on this I am running more up to date versions of the kernel and Beignet.


Error from the task
OpenCL version by Raistmer, r3352

Number of OpenCL platforms: terminate called after throwing an instance of 'std::bad_cast'
what(): std::bad_cast
SIGABRT: abort called
Stack trace (16 frames):
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah(boinc_catch_signal+0x62)[0x63d122]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x10660)[0x7fa1b00df660]
/lib/x86_64-linux-gnu/libc.so.6(gsignal+0x37)[0x7fa1afd5e507]
/lib/x86_64-linux-gnu/libc.so.6(abort+0x16a)[0x7fa1afd5f8da]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x7c0c95]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x776526]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x776553]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x7764ce]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x7bf832]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x5ecd0a]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x41e8f9]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x43caf8]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x451258]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x4159d7]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf0)[0x7fa1afd4b870]
../../projects/setiweb.ssl.berkeley.edu_beta/setiathome_8.05_x86_64-pc-linux-gnu__opencl_intel_gpu_sah[0x4173c9]

Exiting...


CLInfo
Number of platforms 1
Platform Name Intel Gen OCL Driver
Platform Vendor Intel
Platform Version OpenCL 1.2 beignet 1.1.1
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_byte_addressable_store cl_khr_spir cl_khr_icd
Platform Extensions function suffix Intel

Platform Name Intel Gen OCL Driver
Number of devices 1
Device Name Intel(R) HD Graphics IvyBridge GT2
Device Vendor Intel
Device Vendor ID 0x8086
Device Version OpenCL 1.2 beignet 1.1.1
Driver Version 1.1.1
Device OpenCL C Version OpenCL C 1.2 beignet 1.1.1
Device Type GPU
Device Profile FULL_PROFILE
Max compute units 16
Max clock frequency 1000MHz
Device Partition (core)
Max number of sub-devices 1
Supported partition types None, None, None
Max work item dimensions 3
Max work item sizes 512x512x512
Max work group size 512
Preferred work group size multiple 16
Preferred / native vector sizes
char 16 / 8
short 8 / 8
int 4 / 4
long 2 / 2
half 0 / 8 (n/a)
float 4 / 4
double 0 / 2 (n/a)
Half-precision Floating-point support (n/a)
Single-precision Floating-point support (core)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add No
Support is emulated in software No
Correctly-rounded divide and sqrt operations No
Double-precision Floating-point support (n/a)
Address bits 32, Little-Endian
Global memory size 2147483648 (2GiB)
Error Correction support No
Max memory allocation 1073741824 (1024MiB)
Unified memory for Host and Device Yes
Minimum alignment for any data type 128 bytes
Alignment of base address 1024 bits (128 bytes)
Global Memory cache type Read/Write
Global Memory cache size 8192
Global Memory cache line 64 bytes
Image support Yes
Max number of samplers per kernel 16
Max size for 1D images from buffer 65536 pixels
Max 1D or 2D image array size 2048 images
Max 2D image size 8192x8192 pixels
Max 3D image size 8192x8192x2048 pixels
Max number of read image args 128
Max number of write image args 8
Local memory type Global
Local memory size 65536 (64KiB)
Max constant buffer size 134217728 (128MiB)
Max number of constant args 8
Max size of kernel argument 1024
Queue properties
Out-of-order execution No
Profiling Yes
Prefer user sync for interop Yes
Profiling timer resolution 80ns
Execution capabilities
Run OpenCL kernels Yes
Run native kernels Yes
SPIR versions <printDeviceInfo:135: get SPIR versions size : error -30>
printf() buffer size 1048576 (1024KiB)
Built-in kernels __cl_copy_region_align4;__cl_copy_region_align16;__cl_cpy_region_unalign_same_offset;__cl_copy_region_unalign_dst_offset;__cl_copy_region_unalign_src_offset;__cl_copy_buffer_rect;__cl_copy_image_1d_to_1d;__cl_copy_image_2d_to_2d;__cl_copy_image_3d_to_2d;__cl_copy_image_2d_to_3d;__cl_copy_image_3d_to_3d;__cl_copy_image_2d_to_buffer;__cl_copy_image_3d_to_buffer;__cl_copy_buffer_to_image_2d;__cl_copy_buffer_to_image_3d;__cl_fill_region_unalign;__cl_fill_region_align2;__cl_fill_region_align4;__cl_fill_region_align8_2;__cl_fill_region_align8_4;__cl_fill_region_align8_8;__cl_fill_region_align8_16;__cl_fill_region_align128;__cl_fill_image_1d;__cl_fill_image_1d_array;__cl_fill_image_2d;__cl_fill_image_2d_array;__cl_fill_image_3d;
Device Available Yes
Compiler Available Yes
Linker Available Yes
Device Extensions cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_byte_addressable_store cl_khr_spir cl_khr_icd

NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) Intel Gen OCL Driver
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) Success [Intel]
clCreateContext(NULL, ...) [default] Success [Intel]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name Intel Gen OCL Driver
Device Name Intel(R) HD Graphics IvyBridge GT2
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) Success (1)
Platform Name Intel Gen OCL Driver
Device Name Intel(R) HD Graphics IvyBridge GT2

ICD loader properties
ICD loader Name OpenCL ICD Loader
ICD loader Vendor OCL Icd free software
ICD loader Version 2.2.8
ICD loader Profile OpenCL 1.2

Unusual error you got there :

Number of OpenCL platforms: terminate called after throwing an instance of 'std::bad_cast'

Especially as there is no cast in the code where the error shows up, maybe some underlying function not ok. Need to find solution.

http://lunatics.kwsn.info/index.php/topic,1796.msg60476.html#msg60476
Could you retry running clinfo with clinfo from AMD just for reference ? Does it crash in the same place as the intel_gpu app ?

Also important : "gcc -v" and which versions of OpenCL are supported by your "libOpenCL.so" in use : "objdump -p /path/to/lib/libOpenCL.so". Check "/etc/OpenCL/vendors/intel-beignet.icd" for "path to lib".

Example:

objdump -p libOpenCL.so

libOpenCL.so: file format elf64-x86-64

Program Header:
LOAD off 0x0000000000000000 vaddr 0x0000000000000000 paddr 0x0000000000000000 align 2**21
filesz 0x000000000001bfc4 memsz 0x000000000001bfc4 flags r-x
LOAD off 0x000000000001c314 vaddr 0x000000000021c314 paddr 0x000000000021c314 align 2**21
filesz 0x0000000000000e24 memsz 0x0000000000000e6c flags rw-
DYNAMIC off 0x000000000001cd90 vaddr 0x000000000021cd90 paddr 0x000000000021cd90 align 2**3
filesz 0x0000000000000220 memsz 0x0000000000000220 flags rw-
NOTE off 0x0000000000000200 vaddr 0x0000000000000200 paddr 0x0000000000000200 align 2**2
filesz 0x0000000000000024 memsz 0x0000000000000024 flags r--
TLS off 0x000000000001c314 vaddr 0x000000000021c314 paddr 0x000000000021c314 align 2**2
filesz 0x0000000000000000 memsz 0x0000000000000004 flags r--
EH_FRAME off 0x0000000000019374 vaddr 0x0000000000019374 paddr 0x0000000000019374 align 2**2
filesz 0x000000000000047c memsz 0x000000000000047c flags r--
STACK off 0x0000000000000000 vaddr 0x0000000000000000 paddr 0x0000000000000000 align 2**3
filesz 0x0000000000000000 memsz 0x0000000000000000 flags rw-
RELRO off 0x000000000001c314 vaddr 0x000000000021c314 paddr 0x000000000021c314 align 2**0
filesz 0x0000000000000cec memsz 0x0000000000000cec flags r--

Dynamic Section:
NEEDED libdl.so.2
NEEDED libc.so.6
NEEDED ld-linux-x86-64.so.2
SONAME libOpenCL.so.1
INIT 0x0000000000004cd0
FINI 0x00000000000164e4
INIT_ARRAY 0x000000000021c318
INIT_ARRAYSZ 0x0000000000000008
FINI_ARRAY 0x000000000021c320
FINI_ARRAYSZ 0x0000000000000008
HASH 0x0000000000000228
GNU_HASH 0x00000000000006f8
STRTAB 0x0000000000001bf8
SYMTAB 0x0000000000000ba8
STRSZ 0x0000000000000c5d
SYMENT 0x0000000000000018
PLTGOT 0x000000000021d000
PLTRELSZ 0x0000000000000348
PLTREL 0x0000000000000007
JMPREL 0x0000000000004988
RELA 0x0000000000002b10
RELASZ 0x0000000000001e78
RELAENT 0x0000000000000018
VERDEF 0x00000000000029b8
VERDEFNUM 0x0000000000000006
VERNEED 0x0000000000002a80
VERNEEDNUM 0x0000000000000003
VERSYM 0x0000000000002856
RELACOUNT 0x000000000000013e

Version definitions:
1 0x01 0x08a40d11 libOpenCL.so.1
2 0x00 0x028568b0 OPENCL_1.0
3 0x00 0x028568b1 OPENCL_1.1
OPENCL_1.0
4 0x00 0x028568b2 OPENCL_1.2
OPENCL_1.1
5 0x00 0x028561b0 OPENCL_2.0
OPENCL_1.2
6 0x00 0x028561b1 OPENCL_2.1
OPENCL_2.0

Version References:
required from libdl.so.2:
0x09691a75 0x00 11 GLIBC_2.2.5
required from ld-linux-x86-64.so.2:
0x0d696913 0x00 09 GLIBC_2.3
required from libc.so.6:
0x09691974 0x00 12 GLIBC_2.3.4
0x06969194 0x00 10 GLIBC_2.14
0x0d696914 0x00 08 GLIBC_2.4
0x09691a75 0x00 07 GLIBC_2.2.5

Was Beignet driver built from source on host or installed from repository ?

---

_\|/_
U r s

Everything is from the repo. Nothing compiled on this machine so probably not relevant to get local gcc version (Stretch is up to gcc v5). The clinfo is the one from the repo, I will need to see if they have an AMD version.

$ cat intel-beignet-x86_64-linux-gnu.icd
/usr/lib/x86_64-linux-gnu/beignet//libcl.so

No objdump on the machine

Everything is from the repo. Nothing compiled on this machine so probably not relevant to get local gcc version (Stretch is up to gcc v5). The clinfo is the one from the repo, I will need to see if they have an AMD version.

Did some reading and am now convinced about the reason for the error : your system is gcc 5 based and expects by default a newer version of the stdandard c++ library, which is no longer automatically backward compatible with earlier versions.
The <std::vector>.size() function, where the error showed up, has been changed significantly in the new standard library.
(see http://www.gnu.org/software/gcc/gcc-5/changes.html#libstdcxx for reference)
Solution :
Set _GLIBCXX_USE_CXX11_ABI to 0 when including any headers from standard library, so new standrad library can choose old methods, which are still supported.

Will change that in the next version of the intel_gpu app for linux.

---

_\|/_
U r s

Everything is from the repo. Nothing compiled on this machine so probably not relevant to get local gcc version (Stretch is up to gcc v5). The clinfo is the one from the repo, I will need to see if they have an AMD version.

Did some reading and am now convinced about the reason for the error : your system is gcc 5 based and expects by default a newer version of the stdandard c++ library, which is no longer automatically backward compatible with earlier versions.

<snipped>

When the stdc6 library turned up on my Raspberry Pi B+ it managed to kill the Einstein BRP4 1.06 app that it had been running for months. I had to go back to the Jessie release so I could run it. It sounds like quite a few things will break because of this. The repo stuff will probably be okay because they will be rebuilt but the rest...

https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=22839680

I got no answers , but more failures, whats that ?

Just asked the question below on the Android apps News thread and, as you can see, it was suggested I try here, or at least I think this is where Claggy meant :-)

In the last round of testing I joined an Android tablet up to Beta, in addition to my usual Windows clients. In the computer details it lists both a CPU and GPU (INTEL Intel(R) HD Graphics (535MB) OpenCL: 1.2). I can see, from the Event Log on the tablet, that it is requesting tasks for the GPU, but it's never got any.

Is there an Intel GPU app for Android? If there is, do I need to change/add a configuration file get it to work?

Any advice would be much appreciated :-)

Until your post i'd never heard of Intel GPU support being available under Android,
generic GPU support under Android has been almost non-existent because of lack of manufacture provided drivers,
It's good to see that Boinc Intel GPU device detection does in fact work,
Regarding applications, i haven't heard of any Android Intel GPU apps being in development,
Since Android is fairly closely related to Linux, you could ask in the Linux/Beignet thread if Urs can supply an Linux app to Eric for deploying as a Android app and see what happens.

Claggy

Is this something you would like to test? If so my tablet is available :-) http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=75528

Cheers

SusieQ

https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=22839680

I got no answers , but more failures, whats that ?

Number of OpenCL platforms:				 terminate called after throwing an instance of 'std::bad_cast'
  what():  std::bad_cast
SIGABRT: abort called

That error happens with the new libstdc++ based on gcc5.x

---

_\|/_
U r s

Number of OpenCL platforms:				 terminate called after throwing an instance of 'std::bad_cast'
  what():  std::bad_cast
SIGABRT: abort called

That error happens with the new libstdc++ based on gcc5.x

Is there a new build that corrects this?

You mentioned before you were looking at changing something to make it work with later GCC versions.

Set _GLIBCXX_USE_CXX11_ABI to 0 when including any headers from standard library, so new standrad library can choose old methods, which are still supported.

Number of OpenCL platforms:				 terminate called after throwing an instance of 'std::bad_cast'
  what():  std::bad_cast
SIGABRT: abort called

That error happens with the new libstdc++ based on gcc5.x

Is there a new build that corrects this?

You mentioned before you were looking at changing something to make it work with later GCC versions.

Set _GLIBCXX_USE_CXX11_ABI to 0 when including any headers from standard library, so new standrad library can choose old methods, which are still supported.

New beta app was released last week. Please check if it works now gcc5 based system here at beta test.

---

_\|/_
U r s

New beta app was released last week. Please check if it works now gcc5 based system here at beta test.

One completed successfully and validated. Will run more tonight my time.

New beta app was released last week. Please check if it works now gcc5 based system here at beta test.

One completed successfully and validated. Will run more tonight my time.

About 14 or 15 done so far on the iGPU. One inconclusive at the moment, but both of us have -9 result overflow the rest have validated or are pending.

New beta app was released last week. Please check if it works now gcc5 based system here at beta test.

One completed successfully and validated. Will run more tonight my time.

About 14 or 15 done so far on the iGPU. One inconclusive at the moment, but both of us have -9 result overflow the rest have validated or are pending.

Can you guess why the CPU usage for iGPU tasks is that high ? It's 6 times higher than on my Ci7 3770T.

---

_\|/_
U r s

About 14 or 15 done so far on the iGPU. One inconclusive at the moment, but both of us have -9 result overflow the rest have validated or are pending.

Can you guess why the CPU usage for iGPU tasks is that high ? It's 6 times higher than on my Ci7 3770T.

No idea. Possibly Beignet driver.

I was running 7 CPU and one iGPU at the same time. I find OpenCL likes a dedicated core regardless of it running under Linux or Windows.

About 14 or 15 done so far on the iGPU. One inconclusive at the moment, but both of us have -9 result overflow the rest have validated or are pending.

Can you guess why the CPU usage for iGPU tasks is that high ? It's 6 times higher than on my Ci7 3770T.

No idea. Possibly Beignet driver.

I was running 7 CPU and one iGPU at the same time. I find OpenCL likes a dedicated core regardless of it running under Linux or Windows.

Same over here 7 on CPU and one on iGPU, so thats not it.

---

_\|/_
U r s

I ran some Einstein work on the same machine and their tasks also came in somewhat slower than a similar windows machine.

It was a clean install in the last week. I did a net install from Debian Jessie then a dist-upgrade to Stretch. Nothing else on it that I know of that should slow it down.

Got Debian Stretch running on host id 75365 from one of the daily snapshots, but had to install a lot of libs to get BOINC 7.6.22 compile on it. The repository version 7.6.31 did not work properly.

Will watch CPU usage closely !

---

_\|/_
U r s

Got Debian Stretch running on host id 75365 from one of the daily snapshots, but had to install a lot of libs to get BOINC 7.6.22 compile on it. The repository version 7.6.31 did not work properly.

Will watch CPU usage closely !

The BOINC-client seemed to work fine for me. I remove X11-* before installing BOINC. I don't use the manager. If you have issues with the client I am sure Gianfranco would like to know.

To switch from Jessie to Stretch all you should need to do is change your sources.list and do a dist-upgrade.