| View previous topic :: View next topic |
| Author |
Message |
justXi
Tux's lil' helper
Joined: 01 Aug 2007
Posts: 85
|
 Posted: Sun Mar 18, 2018 4:17 pm Post subject: ROC - Radeon Open Compute |
 |
|
Hello,
I am interessted in getting ROC from AMD to work and emerge on Gentoo.
Is someone working on it?
I created a few ebuilds and uploaded them to https://bugs.gentoo.org/650804.
It is possbile to emerge the "thunk-interface", "runtime" and "opencl".
It works with kernel 4.13 from https://github.com/RadeonOpenCompute/ROCK-Kernel-Driver. (firmware is taken from the ubuntu packages)
Running "clinfo" using the library "libamdocl64.so" gives some information about the device.
My next step is to check OpenCL programs with the installed libs.
I am working also on an ebuild for the compiler "hcc" (https://github.com/RadeonOpenCompute/hcc), all is building and installing fine.
But the compiler does not work, because of missing files.
If there are already ebuilds to test or a Gentoo project working on that, please let me know where to find it.
If not and there are people who are interessted in ROC too, sharing informations would be nice. |
|
| Back to top |
|
 |
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3345
Location: Rasi, Finland
|
| Posted: Sun Mar 18, 2018 4:42 pm Post subject: |
|
|
This topic is relevant to my interests.
Did you already get OpenCL to work? If you did, can you paste the output of clinfo?
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
justXi
Tux's lil' helper
Joined: 01 Aug 2007
Posts: 85
|
| Posted: Sun Mar 18, 2018 7:59 pm Post subject: |
|
|
This is the output of clinfo (from portage) as root:
| Code: |
Number of platforms 1
Platform Name AMD Accelerated Parallel Processing
Platform Vendor Advanced Micro Devices, Inc.
Platform Version OpenCL 2.0 AMD-APP.internal.dbg (2545.0)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd cl_amd_object_metadata cl_amd_event_callback
Platform Extensions function suffix AMD
Platform Name AMD Accelerated Parallel Processing
Number of devices 1
Device Name gfx803
Device Vendor Advanced Micro Devices, Inc.
Device Vendor ID 0x1002
Device Version OpenCL 1.2
Driver Version 2545.0 (HSA1.1,LC)
Device OpenCL C Version OpenCL C 2.0
Device Type GPU
Device Available Yes
Device Profile FULL_PROFILE
Device Board Name (AMD) Baffin [Radeon RX 460/560D / Pro 450/455/460/560]
Device Topology (AMD) PCI-E, 23:00.0
Max compute units 14
SIMD per compute unit (AMD) 4
SIMD width (AMD) 16
SIMD instruction width (AMD) 1
Max clock frequency 1176MHz
Graphics IP (AMD) 8.3
Device Partition (core)
Max number of sub-devices 14
Supported partition types none specified
Max work item dimensions 3
Max work item sizes 1024x1024x1024
Max work group size 256
Compiler Available Yes
Linker Available Yes
Preferred work group size multiple 64
Wavefront width (AMD) 64
Preferred / native vector sizes
char 4 / 4
short 2 / 2
int 1 / 1
long 1 / 1
half 1 / 1 (cl_khr_fp16)
float 1 / 1
double 1 / 1 (cl_khr_fp64)
Half-precision Floating-point support (cl_khr_fp16)
Denormals No
Infinity and NANs No
Round to nearest No
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add No
Support is emulated in software No
Single-precision Floating-point support (core)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero Yes
Round to infinity Yes
IEEE754-2008 fused multiply-add Yes
Support is emulated in software No
Correctly-rounded divide and sqrt operations Yes
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
Support is emulated in software No
Address bits 64, Little-Endian
Global memory size 4294967296 (4GiB)
Global free memory (AMD) 4192256 (3.998GiB)
Global memory channels (AMD) 4
Global memory banks per channel (AMD) 4
Global memory bank width (AMD) 256 bytes
Error Correction support No
Max memory allocation 3650722201 (3.4GiB)
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 16384 (16KiB)
Global Memory cache line size 64 bytes
Image support No
Local memory type Local
Local memory size 65536 (64KiB)
Local memory syze per CU (AMD) 65536 (64KiB)
Local memory banks (AMD) 32
Max constant buffer size 3650722201 (3.4GiB)
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 1ns
Profiling timer offset since Epoch (AMD) 0ns (Thu Jan 1 01:00:00 1970)
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Thread trace supported (AMD) No
printf() buffer size 4194304 (4MiB)
Built-in kernels
Device 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_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_fp16 cl_khr_gl_sharing cl_amd_device_attribute_query cl_amd_media_ops cl_amd_media_ops2 cl_khr_subgroups cl_khr_depth_images cl_amd_copy_buffer_p2p
NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) No platform
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) No platform
clCreateContext(NULL, ...) [default] No platform
clCreateContext(NULL, ...) [other] Success [AMD]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) No platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) No platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) No platform
|
|
|
| Back to top |
|
|
justXi
Tux's lil' helper
Joined: 01 Aug 2007
Posts: 85
|
| Posted: Sun Mar 18, 2018 8:02 pm Post subject: |
|
|
Output from clinfo shipped with the ROCm-OpenCL-Runtime package, as root:
| Code: |
Number of platforms: 1
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 2.0 AMD-APP.internal.dbg (2545.0)
Platform Name: AMD Accelerated Parallel Processing
Platform Vendor: Advanced Micro Devices, Inc.
Platform Extensions: cl_khr_icd cl_amd_object_metadata cl_amd_event_callback
Platform Name: AMD Accelerated Parallel Processing
Number of devices: 1
Device Type: CL_DEVICE_TYPE_GPU
Vendor ID: 1002h
Board name: Baffin [Radeon RX 460/560D / Pro 450/455/460/560]
Device Topology: PCI[ B#35, D#0, F#0 ]
Max compute units: 14
Max work items dimensions: 3
Max work items[0]: 1024
Max work items[1]: 1024
Max work items[2]: 1024
Max work group size: 256
Preferred vector width char: 4
Preferred vector width short: 2
Preferred vector width int: 1
Preferred vector width long: 1
Preferred vector width float: 1
Preferred vector width double: 1
Native vector width char: 4
Native vector width short: 2
Native vector width int: 1
Native vector width long: 1
Native vector width float: 1
Native vector width double: 1
Max clock frequency: 1176Mhz
Address bits: 64
Max memory allocation: 3650722201
Image support: No
Max size of kernel argument: 1024
Alignment (bits) of base address: 1024
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: Read/Write
Cache line size: 64
Cache size: 16384
Global memory size: 4294967296
Constant buffer size: 3650722201
Max number of constant args: 8
Local memory type: Scratchpad
Local memory size: 65536
Max pipe arguments: 16
Max pipe active reservations: 16
Max pipe packet size: 3650722201
Max global variable size: 3650722201
Max global variable preferred total size: 4294967296
Max read/write image args: 0
Max on device events: 0
Queue on device max size: 0
Max on device queues: 0
Queue on device preferred size: 0
SVM capabilities:
Coarse grain buffer: Yes
Fine grain buffer: Yes
Fine grain system: No
Atomics: No
Preferred platform atomic alignment: 0
Preferred global atomic alignment: 0
Preferred local atomic alignment: 0
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 on Host properties:
Out-of-Order: No
Profiling : Yes
Queue on Device properties:
Out-of-Order: No
Profiling : No
Platform ID: 0x7f1fe60b43e0
Name: gfx803
Vendor: Advanced Micro Devices, Inc.
Device OpenCL C version: OpenCL C 2.0
Driver version: 2545.0 (HSA1.1,LC)
Profile: FULL_PROFILE
Version: OpenCL 1.2
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_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_fp16 cl_khr_gl_sharing cl_amd_device_attribute_query cl_amd_media_ops cl_amd_media_ops2 cl_khr_subgroups cl_khr_depth_images cl_amd_copy_buffer_p2p
|
|
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3345
Location: Rasi, Finland
|
| Posted: Sun Mar 18, 2018 8:54 pm Post subject: |
|
|
I'm currently using the OpenCl implementation from amdgpu-pro drivers, while still using the opensource amdgpu driver for general graphics. I should maybe switch then...
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
justXi
Tux's lil' helper
Joined: 01 Aug 2007
Posts: 85
|
| Posted: Tue Mar 20, 2018 5:17 pm Post subject: |
|
|
Do you know what this means....
Is it OpenCL 2.0 capable or not?
| Code: |
Number of platforms: 1
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 2.0 AMD-APP.internal.dbg (2545.0)
Platform Name: AMD Accelerated Parallel Processing
Platform Vendor: Advanced Micro Devices, Inc.
Platform Extensions: cl_khr_icd cl_amd_object_metadata cl_amd_event_callback
...
Vendor: Advanced Micro Devices, Inc.
Device OpenCL C version: OpenCL C 2.0
Driver version: 2545.0 (HSA1.1,LC)
Profile: FULL_PROFILE
Version: OpenCL 1.2
|
|
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3345
Location: Rasi, Finland
|
| Posted: Tue Mar 20, 2018 6:33 pm Post subject: |
|
|
My guess is, it's cabable of both standards: OpenCL 1.2 and 2.0.
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
justXi
Tux's lil' helper
Joined: 01 Aug 2007
Posts: 85
|
|
| Back to top |
|
|
Accipiter
Tux's lil' helper
Joined: 24 Feb 2003
Posts: 87
Location: Buffalo, NY
|
| Posted: Sun Apr 22, 2018 8:04 pm Post subject: |
|
|
| I'm also very interested in giving these a spin. I have a little experience with ebuild development, and would like to help if I can. |
|
| Back to top |
|
|
justXi
Tux's lil' helper
Joined: 01 Aug 2007
Posts: 85
|
| Posted: Mon Apr 23, 2018 4:19 pm Post subject: |
|
|
| Accipiter wrote: | | I'm also very interested in giving these a spin. I have a little experience with ebuild development, and would like to help if I can. |
I think any help is appreciated.
I want to run the ebuilds on a fresh installation to find missing dependencies.
And I want to try the git-sources-4.17-rc2 if there is support for dGPU. |
|
| Back to top |
|
|
MarcoMarin
Apprentice
Joined: 12 Mar 2014
Posts: 151
|
| Posted: Wed May 02, 2018 10:31 pm Post subject: help |
|
|
yet another aspiring amd openCL'er here reporting..
my card probably won't be able to run the new ROC stuff but I'll settle for studying openCL.
I have an old Gentoo box which had already managed to run openCL stuff, but it runs the old fglx driver. On a new Gentoo installation I'm trying to get openCL on the new drivers. Everything works (fbcon, etc) except emerging the amdgpu-pro-opencl libraries.
I download it manually, as per Zucca awesome work (on another thread) and the good soul who placed it on main gentoo overlay.
But portage can't find the file:
1st) Asks me to place it in /var/tmp/(some random string)/portage/dev-libs (etc, the distdir directory), but it always asks for a different random string!
2nd) If I say Y to emerge, then it gets more reasonable and asks for it to be placed in the normal directory (/var/tmp/portage/dev-libs/amdgpu-pro-opencl(version)/distdir) BUT when I do that, it says it cannot find it and deletes the file.
So far I managed to emerge ocl-icd, but I'm stuck without the libraries.
What am I doing wrong? |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3345
Location: Rasi, Finland
|
| Posted: Wed May 02, 2018 11:27 pm Post subject: Re: help |
|
|
| MarcoMarin wrote: | 1st) Asks me to place it in /var/tmp/(some random string)/portage/dev-libs (etc, the distdir directory), but it always asks for a different random string!
2nd) If I say Y to emerge, then it gets more reasonable and asks for it to be placed in the normal directory (/var/tmp/portage/dev-libs/amdgpu-pro-opencl(version)/distdir) BUT when I do that, it says it cannot find it and deletes the file. |
Can you post the output of emerge command of what it is telling you to do?
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
Hu
Moderator
Joined: 06 Mar 2007
Posts: 21633
|
| Posted: Thu May 03, 2018 2:01 am Post subject: Re: help |
|
|
| MarcoMarin wrote: | Everything works (fbcon, etc) except emerging the amdgpu-pro-opencl libraries.
I download it manually, as per Zucca awesome work (on another thread) and the good soul who placed it on main gentoo overlay.
But portage can't find the file:
1st) Asks me to place it in /var/tmp/(some random string)/portage/dev-libs (etc, the distdir directory), but it always asks for a different random string! |
This is a known bug with a messy explanation. In short, the package shows you $DISTDIR as seen by the function telling you to do the download. Historically, this was the real DISTDIR, but the specification never promised that. Recently, the developers involved decided that since the specification never promised this would work, they would change DISTDIR handling and declare that any ebuild using DISTDIR in this phase, including for diagnostic output like this, was buggy and needed to be changed. See [Tracker] Ebuilds accessing DISTDIR in pkg_* phases among others. Personally, this feels like a case where the reality of ebuilds showing this variable should override the technicality that it was not guaranteed to work.
As I understand it, you are supposed to (1) bug the maintainer of the ebuild to make it not print this variable, since the variable is misleading instead of useful and (2) ignore the apparent value of DISTDIR and instead put the file in your real, persistent DISTDIR (default value /usr/portage/distfiles). |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3345
Location: Rasi, Finland
|
| Posted: Thu May 03, 2018 10:17 am Post subject: Re: help |
|
|
| Hu wrote: | | This is a known bug with a messy explanation. In short, the package shows you $DISTDIR as seen by the function telling you to do the download. Historically, this was the real DISTDIR, but the specification never promised that. Recently, the developers involved decided that since the specification never promised this would work, they would change DISTDIR handling and declare that any ebuild using DISTDIR in this phase, including for diagnostic output like this, was buggy and needed to be changed. |
Why I don't remember stumbling into this even when I have made an ebuild which specifically "abuses" DISTDIR..?
Oh well. Sourcing or grepping make.conf for the destination directory isn't nice either. Although one can assume it's /usr/share/portage/distfiles if there isn't one defined in make.conf. But there's a change that someone uses portage bashrc to set it for example. So all methods of getting the persistent DISTDIR inside pkg_* -functions are considered hacks now?
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
MarcoMarin
Apprentice
Joined: 12 Mar 2014
Posts: 151
|
| Posted: Fri May 04, 2018 12:31 am Post subject: |
|
|
It is curious though that the 2nd one is "correct", I have only a vague idea of how ebuilds work, so the best I could suggest would be to try to use the 2nd kind (of function?), assuming this is not all automated of course.
okay, I placed the archive on /usr/portage/distfiles (the directory "/usr/share/portage/distfiles doesn't seem to exist  ), will test first thing in the morning. Let me know whether you still need the output from emerge. |
|
| Back to top |
|
|
Hu
Moderator
Joined: 06 Mar 2007
Posts: 21633
|
| Posted: Fri May 04, 2018 1:40 am Post subject: |
|
|
| When did you last execute the pkg_nofetch of that ebuild, Zucca? I think this change in Portage is relatively recent. The only way to use $DISTDIR in pkg_nofetch is as advice to the user, which mitigates the negative effect of this change. Even so, I think it is a bad change. |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Fri May 04, 2018 4:00 pm Post subject: |
|
|
| Hu wrote: | | The only way to use $DISTDIR in pkg_nofetch is as advice to the user, which mitigates the negative effect of this change. Even so, I think it is a bad change. |
Concur. We have a DISTDIR, it is variable from the point-of-view of the ebuild, and there is no reason to ban its usage.
Honestly, it reminds me of the idiocy the gcc developers used to pull around UB (and perhaps still do.)
Effectively it ends up being developers "thinking through" policy decisions that are not their business; a "modern" failing. |
|
| Back to top |
|
|
MarcoMarin
Apprentice
Joined: 12 Mar 2014
Posts: 151
|
| Posted: Fri May 04, 2018 10:18 pm Post subject: |
|
|
It works! \o/
Here is some output: | Code: | >>> Unpacking source...
>>> Unpacking amdgpu-pro-17.50-511655.tar.xz to /var/tmp/portage/dev-libs/amdgpu-pro-opencl-17.50.511655/work
>>> Unpacking opencl-amdgpu-pro-icd_17.50-511655_amd64.deb to /var/tmp/portage/dev-libs/amdgpu-pro-opencl-17.50.511655/work/amdgpu-pro-17.50-511655
>>> Unpacking data.tar.xz to /var/tmp/portage/dev-libs/amdgpu-pro-opencl-17.50.511655/work/amdgpu-pro-17.50-511655
>>> Unpacking libdrm-amdgpu-amdgpu1_2.4.82-511655_amd64.deb to /var/tmp/portage/dev-libs/amdgpu-pro-opencl-17.50.511655/work/amdgpu-pro-17.50-511655
>>> Unpacking data.tar.xz to /var/tmp/portage/dev-libs/amdgpu-pro-opencl-17.50.511655/work/amdgpu-pro-17.50-511655
>>> Unpacking ids-amdgpu_1.0.0-511655_all.deb to /var/tmp/portage/dev-libs/amdgpu-pro-opencl-17.50.511655/work/amdgpu-pro-17.50-511655
>>> Unpacking data.tar.xz to /var/tmp/portage/dev-libs/amdgpu-pro-opencl-17.50.511655/work/amdgpu-pro-17.50-511655
>>> Source unpacked in |
Apparently, only the files "opencl-amdgpu-pro-icd, data.tar.xz, libdrm-amdgpu-amdgpu, ids-amdgpu" are needed (although there is more going on behind the scenes, "data" for instance is used 3 times), in case we could carve out the over half a Gigabyte archive into a more manageable size.
Then I grabbed an official openCL hello_world and issued "gcc cl_util.c hello_world.c -I/usr/lib/OpenCL/global/include/ -lOpenCL -L/usr/lib64/OpenCL/vendors/ocl-icd/"
Maybe we could add the libs to the default path as a nicety? (or are we indeed supposed to do that for extraneous libraries?)
I had to "mv ./libOpenCL.so.1 /usr/lib" though, so at least a symlink there would be needed. Here's the output:
| Code: | There are 1 platforms.
There are 1 GPU devices.
clCreateContext() succeeded.
clCreateCommandQueue() failed: CL_OUT_OF_HOST_MEMORY |
No idea what's going on.  |
|
| Back to top |
|
|
MarcoMarin
Apprentice
Joined: 12 Mar 2014
Posts: 151
|
| Posted: Tue May 22, 2018 10:26 pm Post subject: |
|
|
trying to install Xorg, apparently had some progress with this:
| Code: | ./a.out: /usr/lib64/OpenCL/vendors/mesa/libOpenCL.so.1: no version information available (required by ./a.out)
There are 1 platforms.
There are 1 GPU devices.
clCreateContext() succeeded.
clCreateCommandQueue() succeeded.
clCreateProgramWithSource() suceeded.
clBuildProgram() suceeded.
clCreateKernel() suceeded.
clCreateBuffer() succeeded.
clSetKernelArg() succeeded.
clEnqueueNDRangeKernel() suceeded.
clFinish() succeeded.
clEnqueueReadBuffer() suceeded.
pi = 3.141590
|
Then I checked eselect opencl, and it had MESA selected!
Now if I eselect back to ocl-icd, it segfaults! O.o
Let me know if anyone wants that source code (hello world). Gonna try some libraries next. |
|
| Back to top |
|
|
schwarzygesetzlos
Apprentice
Joined: 11 Dec 2004
Posts: 185
Location: Funeralopolis
|
| Posted: Thu Dec 06, 2018 5:46 pm Post subject: |
|
|
@justXi: Thanks for your great ebuild! Saved me lots of time to install ROCm manually! I hope it will find it's way in portage.
I only got one problem, OpenCL via ROCm works with my Radeon RX 570, 580 and Radeon Nano but not on my Radeon RX Vega 56. The card is detected by rocminfo but clinfo shows 0 cards.
| Code: | =====================
HSA System Attributes
=====================
Runtime Version: 1.1
System Timestamp Freq.: 1000.000000MHz
Sig. Max Wait Duration: 18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count)
Machine Model: LARGE
System Endianness: LITTLE
==========
HSA Agents
==========
*******
Agent 1
*******
Name: AMD A8-6600K APU with Radeon(tm) HD Graphics
Vendor Name: CPU
Feature: None specified
Profile: FULL_PROFILE
Float Round Mode: NEAR
Max Queue Number: 0(0x0)
Queue Min Size: 0(0x0)
Queue Max Size: 0(0x0)
Queue Type: MULTI
Node: 0
Device Type: CPU
Cache Info:
Chip ID: 0(0x0)
Cacheline Size: 64(0x40)
Max Clock Frequency (MHz):3700
BDFID: 0(0x0)
Compute Unit: 4(0x4)
Features: None
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: KERNARG, FINE GRAINED
Size: 16441176(0xfadf58) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Alignment: 4KB
Acessible by all: TRUE
Pool 2
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 16441176(0xfadf58) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Alignment: 4KB
Acessible by all: TRUE
ISA Info:
N/A
*******
Agent 2
*******
Name: gfx900
Vendor Name: AMD
Feature: KERNEL_DISPATCH
Profile: BASE_PROFILE
Float Round Mode: NEAR
Max Queue Number: 128(0x80)
Queue Min Size: 4096(0x1000)
Queue Max Size: 131072(0x20000)
Queue Type: MULTI
Node: 1
Device Type: GPU
Cache Info:
L1: 16(0x10) KB
Chip ID: 26751(0x687f)
Cacheline Size: 64(0x40)
Max Clock Frequency (MHz):1590
BDFID: 768(0x300)
Compute Unit: 56(0x38)
Features: KERNEL_DISPATCH
Fast F16 Operation: FALSE
Wavefront Size: 64(0x40)
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Waves Per CU: 40(0x28)
Max Work-item Per CU: 2560(0xa00)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 4294967295(0xffffffff)
y 4294967295(0xffffffff)
z 4294967295(0xffffffff)
Max number Of fbarriers Per Workgroup:32
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 8372224(0x7fc000) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Alignment: 4KB
Acessible by all: FALSE
Pool 2
Segment: GROUP
Size: 64(0x40) KB
Allocatable: FALSE
Alloc Granule: 0KB
Alloc Alignment: 0KB
Acessible by all: FALSE
ISA Info:
ISA 1
Name: amdgcn-amd-amdhsa--gfx900
Machine Models: HSA_MACHINE_MODEL_LARGE
Profiles: HSA_PROFILE_BASE
Default Rounding Mode: NEAR
Default Rounding Mode: NEAR
Fast f16: TRUE
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 4294967295(0xffffffff)
y 4294967295(0xffffffff)
z 4294967295(0xffffffff)
FBarrier Max Size: 32
*** Done *** |
I am running gentoo-sources-4.19.7 + media-libs/ROCm-OpenCL-Runtime-1.9.9999. Any ideas what I may have forgotten?
_________________
Talos II. [Gentoo Linux] | PMac G5 11,2. PMac G4 3,6. PBook G4 5,8. [MorphOS 3.18 / Gentoo Linux] | Vampire V4 SA [ApolloOS / Amiga OS 3.2.2] |
|
| Back to top |
|
|
|
Portage & Programming
