Radeon Software 18.20 Preview Offers Early Support For Ubuntu 18.04 LTS & RHEL 7.5

[twriter]

Thanks for the response, I'll try and get that done by today

Also, I checked with our QA folks and RX480 was tested with 100% pass rate across install, base graphics, 3D, and multimedia.

[dungeon]

Also, I checked with our QA folks and RX480 was tested with 100% pass rate across install, base graphics, 3D, and multimedia.

He mentioned CL, so might be that legacy compute break something.

[PackRat]

So, if they add closed-source support to that card, are you gonna buy a recent AMD card (e.g. RX 480) and quit using the older one (thereby throwing their efforts to waste (unless somebody else has that card))?

Amd makes Etherium drivers but I can't get a driver that works with the proprietary applications Autodeskmaya and Sidefx houdini for Linux. I'm using windows because the drivers works there.

[Marc Driftmeyer]

Running the OpenCL stack from the preview on Debian Sid using the 4.15.0-1 Kernel and all is well. Blender chewing out mixed CPU/GPGPU renders from master on 2.79.x and 2.80.x.Platform Name AMD Accelerated Parallel Processing
Number of devices 1
Device Name Ellesmere
Device Vendor Advanced Micro Devices, Inc.
Device Vendor ID 0x1002
Device Version OpenCL 1.2 AMD-APP (2633.3)
Driver Version 2633.3
Device OpenCL C Version OpenCL C 1.2
Device Type GPU
Device Board Name (AMD) AMD Radeon (TM) RX 480 Graphics
Device Topology (AMD) PCI-E, 01:00.0
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Linker Available Yes
Max compute units 36
SIMD per compute unit (AMD) 4
SIMD width (AMD) 16
SIMD instruction width (AMD) 1
Max clock frequency 1338MHz
Graphics IP (AMD) 8.0
Device Partition (core)
Max number of sub-devices 36
Supported partition types none specified
Max work item dimensions 3
Max work item sizes 1024x1024x1024
Max work group size 256
Preferred work group size (AMD) 256
Max work group size (AMD) 1024
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 8331276288 (7.759GiB)
Global free memory (AMD) 8112396 (7.737GiB)
Global memory channels (AMD) 8
Global memory banks per channel (AMD) 16
Global memory bank width (AMD) 256 bytes
Error Correction support No
Max memory allocation 4244635648 (3.953GiB)
Unified memory for Host and Device No
Minimum alignment for any data type 128 bytes
Alignment of base address 2048 bits (256 bytes)
Global Memory cache type Read/Write
Global Memory cache size 16384 (16KiB)
Global Memory cache line size 64 bytes
Image support Yes
Max number of samplers per kernel 16
Max size for 1D images from buffer 134217728 pixels
Max 1D or 2D image array size 2048 images
Base address alignment for 2D image buffers 256 bytes
Pitch alignment for 2D image buffers 256 pixels
Max 2D image size 16384x16384 pixels
Max 3D image size 2048x2048x2048 pixels
Max number of read image args 128
Max number of write image args 8
Local memory type Local
Local memory size 32768 (32KiB)
Local memory syze per CU (AMD) 65536 (64KiB)
Local memory banks (AMD) 32
Max number of constant args 8
Max constant buffer size 4244635648 (3.953GiB)
Preferred constant buffer size (AMD) 16384 (16KiB)
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) 1525466316024086734ns (Fri May 4 13:38:36 2018)
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Thread trace supported (AMD) Yes
Number of async queues (AMD) 2
Max real-time compute queues (AMD) 0
Max real-time compute units (AMD) 0
SPIR versions 1.2
printf() buffer size 4194304 (4MiB)
Built-in kernels
Device 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_3d_image_writes cl_khr_byte_addressable_store cl_khr_fp16 cl_khr_gl_sharing cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_image2d_from_buffer cl_khr_spir cl_khr_gl_event

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 [MESA]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) Success (1)
Platform Name Clover
Device Name AMD Radeon (TM) RX 480 Graphics (POLARIS10 / DRM 3.23.0 / 4.15.0-1-amd64, LLVM 6.0.0)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name Clover
Device Name AMD Radeon (TM) RX 480 Graphics (POLARIS10 / DRM 3.23.0 / 4.15.0-1-amd64, LLVM 6.0.0)
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 Clover
Device Name AMD Radeon (TM) RX 480 Graphics (POLARIS10 / DRM 3.23.0 / 4.15.0-1-amd64, LLVM 6.0.0)

[dungeon]

Amd makes Etherium drivers but I can't get a driver that works with the proprietary applications Autodeskmaya and Sidefx houdini for Linux. I'm using windows because the drivers works there.

AFAIK these apps runs profiles usually, tell AMD exact versions and names of the apps so they might do something about these if that is missing

When it comes to certifcations, as i see here for Maya 2018.3 Windows is there marked as Tested (not Certified) for Radeon R9 280/X with Crimson Edition 16.6.2 (nearly 2 years old driver for a new version of Maya sounds like a copy/paste to me ) with all APIs:

https://knowledge.autodesk.com/sites...ox2018_v03.pdf

While for Linux only AMD workstation cards are there, so not sure... maybe AMD only supports workstation cards on Maya for Linux or they just do certifications for workstation cards for Linux, but not Testing of consumer cards

[StillStuckOnSI]

So, if they add closed-source support to that card, are you gonna buy a recent AMD card (e.g. RX 480) and quit using the older one (thereby throwing their efforts to waste (unless somebody else has that card))?

As someone still in possession of a GCN 1.0 GPU, the biggest missing piece not provided by the open source drivers is modern compute (aka OpenCL >= 1.2). fglrx used to provide this, but the (very welcome!) push towards unifying under amdgpu has left SI cards high-and-dry, so to speak, with regards to said compute features. Short of buying a new GPU, the only options are to downgrade to a 3.x series kernel or run Windows.

At the end of the day, GCN 1.0 cards were just in the wrong place at the wrong time with OpenCL. AMD has given technical arguments as to why SI can't simply be supported in ROCM, and the legacy compute implementation in the new proprietary driver seems to be just that--legacy. Unfortunate for sure, but certainly understandable.

[bridgman]

At the end of the day, GCN 1.0 cards were just in the wrong place at the wrong time with OpenCL. AMD has given technical arguments as to why SI can't simply be supported in ROCM, and the legacy compute implementation in the new proprietary driver seems to be just that--legacy. Unfortunate for sure, but certainly understandable.

We are calling the older code path "legacy" because we are just starting to replace it with PAL (beginning with Vega in the 18.10 release) and needed a name for it since we now had more than one option. That said, I believe it is what we are using for everything up to Polaris right now.

I don't know what the timeline is for enabling PAL in earlier GPUs (it supports everything back to SI AFAIK) but I'll ask if it is worth installing on SI with the PAL option selected to see if/how it works. In a lot of cases the main obstacle to switching code paths quickly is more about the testing/tuning/certification effort than about "making it run".

[StillStuckOnSI]

We are calling the older code path "legacy" because we are just starting to replace it with PAL (beginning with Vega in the 18.10 release) and needed a name for it since we now had more than one option. That said, I believe it is what we are using for everything up to Polaris right now.

I don't know what the timeline is for enabling PAL in earlier GPUs (it supports everything back to SI AFAIK) but I'll ask if it is worth installing on SI with the PAL option selected to see if/how it works. In a lot of cases the main obstacle to switching code paths quickly is more about the testing/tuning/certification effort than about "making it run".

Thanks for the quick reply! I've attempted to compile and run some basic OpenCL kernels on the legacy implementation, but last I checked it required some rather bizarre workarounds to make a simple vecadd kernel function on SI. Do you know if there is a forum for raising OpenCL-related bugs?

[PackRat]

AFAIK these apps runs profiles usually, tell AMD exact versions and names of the apps so they might do something about these if that is missing

When it comes to certifcations, as i see here for Maya 2018.3 Windows is there marked as Tested (not Certified) for Radeon R9 280/X with Crimson Edition 16.6.2 (nearly 2 years old driver for a new version of Maya sounds like a copy/paste to me ) with all APIs:

https://knowledge.autodesk.com/sites...ox2018_v03.pdf

While for Linux only AMD workstation cards are there, so not sure... maybe AMD only supports workstation cards on Maya for Linux or they just do certifications for workstation cards for Linux, but not Testing of consumer cards

Fglrx 15.9 (rhel and centos 7.2 )works with both both Maya and Houdini even unreal engine 4 (no Vulkan) and has Opencl. I have 2 280x's was using the other card Opencl for Houdini fx on opensuse 42.1. I don't know if I can run fglrx and mesa with centos 7.2 since I have 2 cards . Suse use to be autodesk certified... They got no love from amd with radeon software. Ue4 and houdini engine is not officially supported but can work (have not tried). Facefx plugin for ue4 on Linux is on hold. The developers does not know what clang compiler to support X86 or arm ,maybe they should ask one of the bearded ladies at llvm.

Can't wait to get a new Nvidia card (I have a geforce 660 too) when the game ships we will be like sorry Amd cards are not supported.

[extraymond]

We are calling the older code path "legacy" because we are just starting to replace it with PAL (beginning with Vega in the 18.10 release) and needed a name for it since we now had more than one option. That said, I believe it is what we are using for everything up to Polaris right now.

I don't know what the timeline is for enabling PAL in earlier GPUs (it supports everything back to SI AFAIK) but I'll ask if it is worth installing on SI with the PAL option selected to see if/how it works. In a lot of cases the main obstacle to switching code paths quickly is more about the testing/tuning/certification effort than about "making it run".

Thx for updating!

Is PAL something different than rocm? Will Polaris be supported by PAL?
There seems to be so many drivers spinning around lately...