Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for
Archives Discussions
- AMD Community
- Communities
- Developers
- Devgurus Archives
- Archives Discussions
- bug in compiling double for Cypress on linux?
Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Mute
- Printer Friendly Page
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
11-18-2011
03:58 PM
bug in compiling double for Cypress on linux?
compilation segfaults when targeting Cypress, but not when targeting CPU
Any ideas? Is this a known bug?
this error:
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff36b4069 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
Missing separate debuginfos, use: debuginfo-install glibc-2.14-5.x86_64 libX11-1.4.3-1.fc15.x86_64 libXau-1.0.6-2.fc15.x86_64 libXext-1.2.0-2.fc15.x86_64 libXinerama-1.1.1-2.fc15.x86_64 libgcc-4.6.1-9.fc15.x86_64 libgfortran-4.6.1-9.fc15.x86_64 libgomp-4.6.1-9.fc15.x86_64 libjpeg-turbo-1.1.1-1.fc15.x86_64 libpng-1.2.46-1.fc15.x86_64 libquadmath-4.6.1-9.fc15.x86_64 libstdc++-4.6.1-9.fc15.x86_64 libtiff-3.9.5-1.fc15.x86_64 libxcb-1.7-2.fc15.x86_64 numactl-2.0.7-1.fc15.x86_64 xorg-x11-drv-catalyst-libs-11.9-3.fc15.x86_64 zlib-1.2.5-3.fc15.x86_64
(gdb) bt
#0 0x00007ffff36b4069 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#1 0x00007ffff36b5c27 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#2 0x00007ffff36b6641 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#3 0x00007ffff3996e5f in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#4 0x00007ffff3e28950 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#5 0x00007ffff3e28a62 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#6 0x00007ffff3e28bf9 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#7 0x00007ffff320656b in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#8 0x00007ffff3208bc7 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#9 0x00007ffff3222f18 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#10 0x00007ffff324339f in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#11 0x00007ffff320d25d in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#12 0x00007ffff325f565 in ?? () from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#13 0x00007ffff320335d in clBuildProgram ()
from /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/lib/x86_64/libamdocl64.so
#14 0x00000000004038db in cl::Program::build (this=0xa2aec0,
devices=std::vector of length 1, capacity 1 = {...}, options=0x407ab9 "-I.", notifyFptr=0, data=0x0)
at /home/tesch/src/AMD-APP-SDK-v2.5-RC2-lnx64/include/CL/cl.hpp:2675
#15 0x0000000000402782 in clsetup (sourcecount=1) at Clutil.cpp:79
-----------
is cause by this code :
---------
//#define REAL_T_IS_FLOAT
#define REAL_T_IS_DOUBLE
#ifdef REAL_T_IS_FLOAT
typedef float real_t;
typedef float3 vec3_t;
#else
#pragma OPENCL EXTENSION cl_amd_fp64 : enable // subset of cl_khr_fp64
typedef double real_t;
typedef double3 vec3_t;
#endif
__kernel void
bloch_steps(int nsteps)
{
real_t T2i;
for (int i = 0; i < nsteps; i++) {
vec3_t b1;
real_t et2 = exp(-1 * T2i);
real_t theta;
theta = length(b1);
b1 *= 1. / theta;
}
}
----------
when compiling on this device (clinfo output):
[tesch@localhost mtsrc]$ ~/src/AMD-APP-SDK-v2.5-RC2-lnx64/bin/x86_64/clinfo
Number of platforms: 2
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 1.1 AMD-APP-SDK-v2.5 (684.213)
Platform Name: AMD Accelerated Parallel Processing
Platform Vendor: Advanced Micro Devices, Inc.
Platform Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 1.1 LINUX
Platform Name: Intel(R) OpenCL
Platform Vendor: Intel(R) Corporation
Platform Extensions: cl_khr_fp64 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_gl_sharing cl_khr_byte_addressable_store cl_intel_printf cl_ext_device_fission cl_khr_icd
Platform Name: AMD Accelerated Parallel Processing
Number of devices: 2
Device Type: CL_DEVICE_TYPE_GPU
Device ID: 4098
Device Topology: PCI[ B#1, D#0, F#0 ]
Max compute units: 18
Max work items dimensions: 3
Max work items[0]: 256
Max work items[1]: 256
Max work items[2]: 256
Max work group size: 256
Preferred vector width char: 16
Preferred vector width short: 8
Preferred vector width int: 4
Preferred vector width long: 2
Preferred vector width float: 4
Preferred vector width double: 2
Native vector width char: 16
Native vector width short: 8
Native vector width int: 4
Native vector width long: 2
Native vector width float: 4
Native vector width double: 2
Max clock frequency: 725Mhz
Address bits: 32
Max memory allocation: 134217728
Image support: Yes
Max number of images read arguments: 128
Max number of images write arguments: 8
Max image 2D width: 8192
Max image 2D height: 8192
Max image 3D width: 2048
Max image 3D height: 2048
Max image 3D depth: 2048
Max samplers within kernel: 16
Max size of kernel argument: 1024
Alignment (bits) of base address: 32768
Minimum alignment (bytes) for any datatype: 128
Single precision floating point capability
Denorms: No
Quiet NaNs: Yes
Round to nearest even: Yes
Round to zero: Yes
Round to +ve and infinity: Yes
IEEE754-2008 fused multiply-add: Yes
Cache type: None
Cache line size: 0
Cache size: 0
Global memory size: 536870912
Constant buffer size: 65536
Max number of constant args: 8
Local memory type: Scratchpad
Local memory size: 32768
Kernel Preferred work group size multiple: 64
Error correction support: 0
Unified memory for Host and Device: 0
Profiling timer resolution: 1
Device endianess: Little
Available: Yes
Compiler available: Yes
Execution capabilities:
Execute OpenCL kernels: Yes
Execute native function: No
Queue properties:
Out-of-Order: No
Profiling : Yes
Platform ID: 0x7f45fad3b060
Name: Cypress
Vendor: Advanced Micro Devices, Inc.
Device OpenCL C version: OpenCL C 1.1
Driver version: CAL 1.4.1546
Profile: FULL_PROFILE
Version: OpenCL 1.1 AMD-APP-SDK-v2.5 (684.213)
Extensions: cl_khr_fp64 cl_amd_fp64 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_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_atomic_counters_32 cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_popcnt
-------------
but NOT when compiling for this device: (clinfo output continued)
-------------
Device Type: CL_DEVICE_TYPE_CPU
Device ID: 4098
Max compute units: 6
Max work items dimensions: 3
Max work items[0]: 1024
Max work items[1]: 1024
Max work items[2]: 1024
Max work group size: 1024
Preferred vector width char: 16
Preferred vector width short: 8
Preferred vector width int: 4
Preferred vector width long: 2
Preferred vector width float: 4
Preferred vector width double: 0
Native vector width char: 16
Native vector width short: 8
Native vector width int: 4
Native vector width long: 2
Native vector width float: 4
Native vector width double: 0
Max clock frequency: 800Mhz
Address bits: 64
Max memory allocation: 4215897088
Image support: Yes
Max number of images read arguments: 128
Max number of images write arguments: 8
Max image 2D width: 8192
Max image 2D height: 8192
Max image 3D width: 2048
Max image 3D height: 2048
Max image 3D depth: 2048
Max samplers within kernel: 16
Max size of kernel argument: 4096
Alignment (bits) of base address: 1024
Minimum alignment (bytes) for any datatype: 128
Single precision floating point capability
Denorms: Yes
Quiet NaNs: Yes
Round to nearest even: Yes
Round to zero: Yes
Round to +ve and infinity: Yes
IEEE754-2008 fused multiply-add: No
Cache type: Read/Write
Cache line size: 64
Cache size: 65536
Global memory size: 16863588352
Constant buffer size: 65536
Max number of constant args: 8
Local memory type: Global
Local memory size: 32768
Kernel Preferred work group size multiple: 1
Error correction support: 0
Unified memory for Host and Device: 1
Profiling timer resolution: 1
Device endianess: Little
Available: Yes
Compiler available: Yes
Execution capabilities:
Execute OpenCL kernels: Yes
Execute native function: Yes
Queue properties:
Out-of-Order: No
Profiling : Yes
Platform ID: 0x7f45fad3b060
Name: AMD Phenom(tm) II X6 1100T Processor
Vendor: AuthenticAMD
Device OpenCL C version: OpenCL C 1.1
Driver version: 2.0
Profile: FULL_PROFILE
Version: OpenCL 1.1 AMD-APP-SDK-v2.5 (684.213)
Extensions: cl_khr_fp64 cl_amd_fp64 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_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_device_fission cl_amd_device_attribute_query cl_amd_vec3 cl_amd_media_ops cl_amd_popcnt cl_amd_printf
2 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
11-18-2011
04:53 PM
This looks a lot like a crash I've seen that Micah says is fixed in an upcoming release. I bet if you comment out the length that it won't crash.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
11-18-2011
09:48 PM
Yep, commenting it out does stop the crash, also removing the for loop, and removing various other surrounding statements or declarations. I removed a ton of code from the original function to boil down the offending code to just this.
In the end my workaround is to compute the length with sqrt(a.x * a.x + a.y * a.y...) .. but still it would be much better if it just worked like it's supposed to, and I didn't spend any time at all finding workarounds for compiler bugs.
