Comments (6)
I've just pushed a change that dumps the build log if this error comes up. If you still have that platform around and can reproduce the issue, you should now be able to just pipe stderr to a file, and get a (hopefully) meaningful log.
from clinfo.
=== CL_PROGRAM_BUILD_LOG ===
Compilation started
1:6:1: error: using image access qualifier with non-image type
1:5:16: note: expanded from macro 'KRN'
1:4:39: note: expanded from macro '_KRN'
1:1:26: note: expanded from macro 'GWO'
1:6:1: error: using image access qualifier with non-image type
1:5:16: note: expanded from macro 'KRN'
1:4:54: note: expanded from macro '_KRN'
1:2:26: note: expanded from macro 'GRO'
1:6:1: error: using image access qualifier with non-image type
1:5:16: note: expanded from macro 'KRN'
1:4:69: note: expanded from macro '_KRN'
1:2:26: note: expanded from macro 'GRO'
Compilation failed
from clinfo.
Should be fixed in the latest commit, and duh, I should have known better.
from clinfo.
Yep. Here's an Intel MIC for the curious:
Device Name Intel(R) Many Integrated Core Acceleration Card
Device Vendor Intel(R) Corporation
Device Version OpenCL 1.2 (Build 76921)
Driver Version 1.2
Device OpenCL C Version OpenCL C 1.2
Device Type Accelerator
Device Profile FULL_PROFILE
Max compute units 236
Max clock frequency 1052MHz
Device Partition (core)
Max number of sub-devices 0
Supported partition types none
Max work item dimensions 3
Max work item size[0] 8192
Max work item size[1] 8192
Max work item size[2] 8192
Max work group size 8192
Preferred work group size multiple 128
Preferred / native vector sizes
char 1 / 16
short 1 / 16
int 1 / 16
long 1 / 8
half 1 / 0 (n/a)
float 1 / 16
double 1 / 8 (cl_khr_fp64)
Half-precision Floating-point support (n/a)
Single-precision Floating-point support
Denormals Yes
Infinity and NANs Yes
Round to nearest Yes
Round to zero Yes
Round to infinity Yes
IEEE754-2008 fused multiply-add Yes
Correctly-rounded divide and sqrt operations No
Support is emulated in software No
Double-precision Floating-point support (cl_khr_fp64)
Denormals Yes
Infinity and NANs Yes
Round to nearest Yes
Round to zero Yes
Round to infinity Yes
IEEE754-2008 fused multiply-add Yes
Correctly-rounded divide and sqrt operations No
Support is emulated in software No
Address bits 64, Little-Endian
Global memory size 6053617664 ( 5.638GB)
Error Correction support Yes
Max memory allocation 2017869824 ( 1.879GB)
Unified memory for Host and Device No
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 262144 ( 256KB)
Global Memory cache line 64 bytes
Image support No
Local memory type Global
Local memory size 32768 ( 32KB)
Max constant buffer size 131072 ( 128KB)
Max number of constant args 128
Max size of kernel argument 1024
Queue properties
Out-of-order execution Yes
Profiling Yes
Profiling timer resolution 1052ns
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
SPIR versions <printDeviceInfo:742: get CL_DEVICE_SPIR_VERSIONS size : error -30>
Prefer user sync for interops No
printf() buffer size 1048576 ( 1024KB)
Device Built-in kernels
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_fp64
BTW why is Single-precision Floating-point support
empty for all devices? Should it not say "n/a" like many others?
from clinfo.
BTW why is Single-precision Floating-point support empty for all devices?
No wait, that's a headline, right? Maybe you should end them with ":"
from clinfo.
Yes, it's a headline. If it's not available, it will have a n/a next to it (like the Half-precision in your case, or double precision when lacking fp64 support in any form. Yes, it's not the clearest approach. I did consider adding the ":" but it looked strange since they were the only ones. But you just gave me an idea.
from clinfo.
Related Issues (20)
- NULL platform behavior HOT 1
- Information about supported image formats HOT 2
- Typo: "comfornace"
- Preferred work group size multiple (kernel) not working with latest Nvidia compiler HOT 5
- getWGsizes error: syntax error at 'kernel' HOT 2
- clinfo hangs on configurations with two AMD GPU and open source rocm HOT 1
- JSON output is invalid HOT 1
- OpenCL device not reporting with Mali-Gpu HOT 3
- make appimage or binary file, it can run every HOT 1
- On some systems, Clinfo reports twice as many platforms as actually exist. HOT 7
- Clinfo hangs on mesa 23 HOT 5
- Links to both Windows Binaries are broken HOT 4
- Fails to use nvidia ICD. HOT 2
- [feature] Shell completions HOT 1
- Adding support to Android (Qualcomm Adreno) HOT 1
- Invalid JSON output
- Device version check is wrong? HOT 3
- core dump in polaris with rocm opencl version 6.0 HOT 6
- cannot find -lOpenCL while compilation HOT 3
- libOpenCL.so.1: no version information available (required by ./clinfo) HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from clinfo.