Radeon Software For Linux 20.10 Driver Released

[Pseus]

I thought that they added support for Ubuntu 20.04 which will be released in less than a week.
Disappointing as always.
Probably it's my fault since I forget they are nowhere near to Nvidia's release plans.
In 2 years or more they will be probably ready.

Even older versions of AMDGPU-PRO work just fine on a fully up-to-date ArchLinux, so Ubuntu 20.04 support, even if not official, should be there. AMDGPU-PRO replaces userspace components of the driver, so there's no kernel support involved that needs to happen for you to use it. And even then, unless you have a very specific use case, or a bug, you should stick to the full open source driver using Mesa.

[finalzone]

Yes it should... AMDGPU-PRO OpenCL uses the PAL back end which should have been unaffected by the changes required to aligning ROCm APU and dGPU support.

That works on Ryzen 2500U. Only missing is the name of GPU (Radeon™ Vega 8 Graphics as listed on the website) listed as AMD) Unknown AMD GPU as seen below from clinfo

clinfo
Number of platforms 1
Platform Name AMD Accelerated Parallel Processing
Platform Vendor Advanced Micro Devices, Inc.
Platform Version OpenCL 2.1 AMD-APP (3075.10)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd cl_amd_event_callback cl_amd_offline_devices
Platform Host timer resolution 1ns
Platform Extensions function suffix AMD

Platform Name AMD Accelerated Parallel Processing
Number of devices 1
Device Name gfx902
Device Vendor Advanced Micro Devices, Inc.
Device Vendor ID 0x1002
Device Version OpenCL 2.0 AMD-APP (3075.10)
Driver Version 3075.10 (PAL,HSAIL)
Device OpenCL C Version OpenCL C 2.0
Device Type GPU
Device Board Name (AMD) Unknown AMD GPU
Device Topology (AMD) PCI-E, 03:00.0
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Linker Available Yes
Max compute units 8
SIMD per compute unit (AMD) 4
SIMD width (AMD) 16
SIMD instruction width (AMD) 1
Max clock frequency 1100MHz
Graphics IP (AMD) 9.2
Device Partition (core)
Max number of sub-devices 8
Supported partition types None
Supported affinity domains (n/a)
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 2684354560 (2.5GiB)
Global free memory (AMD) 2551712 (2.434GiB)
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 912680550 (870.4MiB)
Unified memory for Host and Device Yes
Shared Virtual Memory (SVM) capabilities (core)
Coarse-grained buffer sharing Yes
Fine-grained buffer sharing Yes
Fine-grained system sharing No
Atomics No
Minimum alignment for any data type 128 bytes
Alignment of base address 2048 bits (256 bytes)
Preferred alignment for atomics
SVM 0 bytes
Global 0 bytes
Local 0 bytes
Max size for global variable 821412352 (783.4MiB)
Preferred total size of global vars 2684354560 (2.5GiB)
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 64
Max number of read/write image args 64
Max number of pipe args 16
Max active pipe reservations 16
Max pipe packet size 912680550 (870.4MiB)
Local memory type Local
Local memory size 65536 (64KiB)
Local memory syze per CU (AMD) 65536 (64KiB)
Local memory banks (AMD) 32
Max number of constant args 8
Max constant buffer size 912680550 (870.4MiB)
Preferred constant buffer size (AMD) 16384 (16KiB)
Max size of kernel argument 1024
Queue properties (on host)
Out-of-order execution No
Profiling Yes
Queue properties (on device)
Out-of-order execution Yes
Profiling Yes
Preferred size 262144 (256KiB)
Max size 8388608 (8MiB)
Max queues on device 1
Max events on device 1024
Prefer user sync for interop Yes
Number of P2P devices (AMD) 0
P2P devices (AMD) <printDeviceInfo:147: get number of CL_DEVICE_P2P_DEVICES_AMD : error -30>
Profiling timer resolution 1ns
Profiling timer offset since Epoch (AMD) 1587258282366124911ns (Sat Apr 18 18:04:42 2020)
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Thread trace supported (AMD) Yes
Number of async queues (AMD) 4
Max real-time compute queues (AMD) 0
Max real-time compute units (AMD) 0
printf() buffer size 4194304 (4MiB)
Built-in kernels (n/a)
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_khr_gl_depth_images 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_subgroups cl_khr_gl_event cl_khr_depth_images cl_khr_mipmap_image cl_khr_mipmap_image_writes 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) Success (1)
Platform Name AMD Accelerated Parallel Processing
Device Name gfx902
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name AMD Accelerated Parallel Processing
Device Name gfx902
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 AMD Accelerated Parallel Processing
Device Name gfx902

That issue plagues for two years since their release.

By the way, amdgpu-install script has a broken installation of local repo for either Red Hat Enterprise 8 and its upstream Fedora which I use. I had to manually set up a local repository. Bonus point for hosting the package to make installation easier.

[Naquatis]

I'll ask about it - I had heard about some accuracy issues with Navi OpenCL but wasn't aware of any crashing. Does the problem happen even with simple renders on Blender or is there some specific model that triggers the issue ?

Some times it renders some frames and some times it crashes instantly. But every time it hangs or reboots while Blender say "computing distribution".

journalctl -b show something like:

kernel: mce: [Hardware Error]: Machine check events logged
kernel: mce: [Hardware Error]: CPU 10: Machine Check: 0 Bank 5: bea0000000000108
kernel: mce: [Hardware Error]: TSC 0 ADDR 7f593df6da9a MISC d012000100000000 SYND 4d000000 IPID 500b000000000
kernel: mce: [Hardware Error]: PROCESSOR 2:870f10 TIME 1587477683 SOCKET 0 APIC 5 microcode 8701013

After the machine reboots because of the crash.

Other than that the system runs extremely stable day and night.

Ok I disabled global C-State and removed every CPU overclocking option + used PCIe Gen3 but even that still crash the system while using AMDGPU Pro PAL OpenCL.

I did some more steps like 2 single symmetrical PSU cable connections. Disabling XMP and setting everything to normal but not a single Bios setting will help to stabilize AMDGPU Pro OpenCL PAL on this RX 5700 XT.

And even the latest Bios update for the X570 Aorus Xtreme did not help.

I changed the PCI Slot and to some reason the PCIe 4.0 x16/x8/x4 Slot works more stable than the PCIe 4.0 Turbo B-Clock IC but guess what .. it is still crashing.

And my last test using the PCIe 4.0 x16/x8/x4 Slot as PCIe 3.0 + disabled global c-states.

Yep, it look like I found a stable setting for the mimimi AMDGPU Pro OpenCL PAL RX 5700 XT rendering.

[bridgman]

OK, interesting... that's a CPU fault not a GPU fault. I have seen that MCE signature reported previously, although not sure what it means.

Let me see what I can find out.

[Naquatis]

OK, interesting... that's a CPU fault not a GPU fault. I have seen that MCE signature reported previously, although not sure what it means.

Let me see what I can find out.

Now as I found a setting that did not crash the whole system it does not mean Blender itself will not crash while creating render kernels. I tried to render a scene I already rendered on a GTX 1080 TI and GTX 1060 without problems.:

https://www.youtube.com/watch?v=bLDL...HQcjcV&index=5

whenever I start rendering the RX 5700 XT react slidely different .. some times creating the render kernel stalls .. some times while trying to render an animation it losing textures ..

.. well I lost enough time on 20.10 while using RX 5700 XT ..

pacman -R $(pacman -Qg Radeon_Software_for_Linux | cut -f2 -d" ")

.. nice back to plain Mesa ..

.. but no problem I simply buy a bigger AMD CPU