version? gpu id? effect?

Message boards : Number crunching : version? gpu id? effect?
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4

AuthorMessage
Fizz Project Donor

Send message
Joined: 11 Jun 01
Posts: 52
Credit: 24,840,628
RAC: 0
Canada
Message 2012459 - Posted: 18 Sep 2019, 21:03:06 UTC - in response to Message 2012455.  

As far as I know you are correct. The MESA drivers do not work for compute. But I don't think that BOINC would show them in the detected output. Could be wrong here. I am not familiar with the output of inxi -G. I am more used to the output of clinfo. Which reminds me I have not seen the requested full output of clinfo from him.


Here is the clinfo:

Number of platforms                               1
  Platform Name                                   AMD Accelerated Parallel Processing
  Platform Vendor                                 Advanced Micro Devices, Inc.
  Platform Version                                OpenCL 2.1 AMD-APP (2906.7)
  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                                     Ellesmere
  Device Vendor                                   Advanced Micro Devices, Inc.
  Device Vendor ID                                0x1002
  Device Version                                  OpenCL 1.2 AMD-APP (2906.7)
  Driver Version                                  2906.7
  Device OpenCL C Version                         OpenCL C 1.2 
  Device Type                                     GPU
  Device Board Name (AMD)                         AMD Radeon (TM) RX 470 Graphics
  Device Topology (AMD)                           PCI-E, 0b:00.0
  Device Profile                                  FULL_PROFILE
  Device Available                                Yes
  Compiler Available                              Yes
  Linker Available                                Yes
  Max compute units                               32
  SIMD per compute unit (AMD)                     4
  SIMD width (AMD)                                16
  SIMD instruction width (AMD)                    1
  Max clock frequency                             1250MHz
  Graphics IP (AMD)                               8.0
  Device Partition                                (core)
    Max number of sub-devices                     32
    Supported partition types                     None
  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                              8308727808 (7.738GiB)
  Global free memory (AMD)                        8085408 (7.711GiB)
  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)        1568833062247484712ns (Wed Sep 18 14:57:42 2019)
  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_amd_bus_addressable_memory 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 [AMD]
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT)  Success (1)
    Platform Name                                 AMD Accelerated Parallel Processing
    Device Name                                   Ellesmere
  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                                   Ellesmere
  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                                   Ellesmere

ID: 2012459 · Report as offensive
Fizz Project Donor

Send message
Joined: 11 Jun 01
Posts: 52
Credit: 24,840,628
RAC: 0
Canada
Message 2012461 - Posted: 18 Sep 2019, 21:07:17 UTC - in response to Message 2012456.  
Last modified: 18 Sep 2019, 21:09:38 UTC

Well Fizz has finally completed his 1st GPU task, https://setiathome.berkeley.edu/result.php?resultid=8059067020.
Cheers.


Yay! Thanks for the info!

Now i'm just left with the puzzle of why going to an older kernel caused the gpu to finally be detected. Why does it work under 4.15 but not under 5.0?

-Fizz
ID: 2012461 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2012462 - Posted: 18 Sep 2019, 21:08:15 UTC - in response to Message 2012459.  

Clinfo output looks correct. You have processed your first gpu task correctly. The app provided is one I am not familiar with and does not look like it performs very well. Eventually the scheduler should send you some SoG tasks and the app and that one should process much faster. Make sure you give the gpu task some cpu support. Drop your cpu usage down to 90% so the gpu tasks doesn't have to fight for cpu support over all the cpu tasks running.

You need to read the thread I posted about suggested ATI gpu command lines which will speed up the applications.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2012462 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2012463 - Posted: 18 Sep 2019, 21:10:55 UTC - in response to Message 2012461.  
Last modified: 18 Sep 2019, 21:12:05 UTC

Well Fizz has finally completed his 1st GPU task, https://setiathome.berkeley.edu/result.php?resultid=8059067020.
Cheers.


Yay! Thanks for the info!

Now i'm just left with the puzzle of why going to an older kernel caused the gpu to finally be detected.

-Fizz

Well the Linux AMD drivers are in a daily flux. Seems there is always a news article posted every day about new drivers or new kinds of graphics support added to them for future kernels. You might want to start reading the daily news articles at phoronix.com. https://www.phoronix.com/scan.php?page=home
https://www.phoronix.com/scan.php?page=search&q=AMDGPU
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2012463 · Report as offensive
Fizz Project Donor

Send message
Joined: 11 Jun 01
Posts: 52
Credit: 24,840,628
RAC: 0
Canada
Message 2012466 - Posted: 18 Sep 2019, 21:18:27 UTC - in response to Message 2012462.  
Last modified: 18 Sep 2019, 21:18:49 UTC

The app provided is one I am not familiar with and does not look like it performs very well. Eventually the scheduler should send you some SoG tasks and the app and that one should process much faster.


To what app(s) are you referring?

Make sure you give the gpu task some cpu support. Drop your cpu usage down to 90% so the gpu tasks doesn't have to fight for cpu support over all the cpu tasks running.

You need to read the thread I posted about suggested ATI gpu command lines which will speed up the applications.


I will check that out. I haven't looked at my configuration for quite a while. The last couple years it's been set to do nothing unless free. I'm running a bunch of planetary orbit simulations for my school work and they can take a long time to process. So i have kept the cpu cycles free for that, only letting Seti run wild when nothing else is going on.

But if limiting the cpu to a maximum of 90% helps the gpu, and in turn ups the overall output, i'm happy to do that.

Next, to figure out how to get the seti screensaver working... heh.


-Fizz
ID: 2012466 · Report as offensive
Fizz Project Donor

Send message
Joined: 11 Jun 01
Posts: 52
Credit: 24,840,628
RAC: 0
Canada
Message 2012467 - Posted: 18 Sep 2019, 21:21:33 UTC - in response to Message 2012463.  

You might want to start reading the daily news articles at phoronix.com. https://www.phoronix.com/scan.php?page=home
https://www.phoronix.com/scan.php?page=search&q=AMDGPU


Indeed, i have checked them out. Hard to keep up with it all, and what's packaged with what, and in which distributions, etc. It's a workout.

But in none of those i bet is anything that said anything like "will prevent Boinc from detecting gpu for openCL applications". Heh. It seems one just has to try, and revert if not happy.

-Fizz
ID: 2012467 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2012468 - Posted: 18 Sep 2019, 21:23:22 UTC - in response to Message 2012466.  

You can find all the Seti apps here. https://setiathome.berkeley.edu/apps.php

Look for the Seti MB V8 Linux apps for what is available for ATI/AMD.

There is no screensaver for Linux for either cpu or gpu tasks. I did find one recent post where someone managed to kludge the screensaver into working on Linux. Not worth the effort in my opinion. I want my gpus doing nothing but crunching. I don't need eye candy.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2012468 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2012469 - Posted: 18 Sep 2019, 21:26:09 UTC - in response to Message 2012467.  
Last modified: 18 Sep 2019, 21:27:56 UTC

You might want to start reading the daily news articles at phoronix.com. https://www.phoronix.com/scan.php?page=home
https://www.phoronix.com/scan.php?page=search&q=AMDGPU


Indeed, i have checked them out. Hard to keep up with it all, and what's packaged with what, and in which distributions, etc. It's a workout.

But in none of those i bet is anything that said anything like "will prevent Boinc from detecting gpu for openCL applications". Heh. It seems one just has to try, and revert if not happy.

-Fizz

Well in one review for the RX 480, so close to what you are running, the review mentioned that so and so OpenCL drivers did not run correctly and they fell back on the AMDGPU-Pro drivers or something.

[Edit] Phoronix usually does Compute tests in their testing suite and mentions when the drivers don't work correctly. Look at the recent 5700XT review for an example.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2012469 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 2012524 - Posted: 19 Sep 2019, 10:24:09 UTC - in response to Message 2012467.  

You might want to start reading the daily news articles at phoronix.com. https://www.phoronix.com/scan.php?page=home
https://www.phoronix.com/scan.php?page=search&q=AMDGPU


Indeed, i have checked them out. Hard to keep up with it all, and what's packaged with what, and in which distributions, etc. It's a workout.

But in none of those i bet is anything that said anything like "will prevent Boinc from detecting gpu for openCL applications". Heh. It seems one just has to try, and revert if not happy.

-Fizz
I'm really surprised it's even working. The Video driver is a Kernel module, it's compiled for a certain kernel. When you make large kernel changes it's very possible the module will not work with the new kernel, you would need to reinstall the driver for the new kernel in that case. I've really never heard of someone mixing the Open source driver with the AMD driver, much less it actually working, if you want to call CPU speed on a GPU working. I'd suggest two Systems, one standard system that's left alone with the Full AMD driver for BOINC, the other that you can experiment with changing kernels and mixing drivers. I've tested all those Linux ATI/AMD Apps Urs has at SETI, they are all within a few seconds of each other on a properly working system. This is how a properly working system would run, https://setiathome.berkeley.edu/results.php?hostid=8316299, those BLC32s would take around 7.5 minutes on a 470 or 570. My first guess would be the mixed driver you're using, but, I'd also guess you've made other changes you haven't mentioned, so, it could be anything.
ID: 2012524 · Report as offensive
Fizz Project Donor

Send message
Joined: 11 Jun 01
Posts: 52
Credit: 24,840,628
RAC: 0
Canada
Message 2012559 - Posted: 19 Sep 2019, 21:25:25 UTC - in response to Message 2012524.  

I've found some other instances of people using the base amdgpu (non-pro) in conjunction with the OpenCL. This thread has an interesting discussion about it, though it's for Einstein@Home, not Seti.

https://einsteinathome.org/content/quick-guide-how-install-opencl-amd-gpus-linux-kubuntu-1804-and-similar-distro

-Fizz
ID: 2012559 · Report as offensive
Fizz Project Donor

Send message
Joined: 11 Jun 01
Posts: 52
Credit: 24,840,628
RAC: 0
Canada
Message 2012560 - Posted: 19 Sep 2019, 21:29:25 UTC - in response to Message 2012469.  
Last modified: 19 Sep 2019, 21:30:11 UTC

Well in one review for the RX 480, so close to what you are running, the review mentioned that so and so OpenCL drivers did not run correctly and they fell back on the AMDGPU-Pro drivers or something.

[Edit] Phoronix usually does Compute tests in their testing suite and mentions when the drivers don't work correctly. Look at the recent 5700XT review for an example.


I am hoping to (one day) get back into some gaming, and may want to update my graphics card to a 5800 or somesuch. As you mentioned earlier, there are some issues with it currently, but i can afford to wait (haven't even got simple games working through Wine yet). But yes keeping an eye on Phoronix for such news is a good idea. I'm slowly but steadily learning about all the details that go into being a linux user. Heh.

Oh, and thanks to you, TBar, and everyone else who has stuck with me through this thread to get this working. Was a bit more painful than i hoped, but we made it i think. :)


-Fizz
ID: 2012560 · Report as offensive
Previous · 1 · 2 · 3 · 4

Message boards : Number crunching : version? gpu id? effect?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.