Hello Everyone,

I just released the newest version of the MilkyWay@home application (1.38). The only major change to this version is an inclusion of the newest BOINC libraries in an attempt to improve compatibility with the new graphics cards.

Mac users will still receive the 1.36 application version for the time being while we work on securing a new Mac computer to build and test on for the future.

If you notice any issues with this application please post them here.

Happy Crunching,

Jake

Something's not right with the opencl_ati_101 app.

On my AMD Radeon R9 version 1.36 used to run 11 seconds, now version 1.38 needs more than 10 minutes (see http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1791907717)!?

Most of the time it's not even utilizing the graphic card but running solely on a full cpu core.

Greetings,
Senilix

Edit: I compared the stdout files of a 1.36 job and a 1.38 job. Looks like 1.38 doesn't detect my gpu anymore...

Ohhh, Thanks for this. Looks like the CPU version got released as the GPU version. I'll have this fixed in a few minutes.

Jake

[edit]

This should be fixed now. You may want to kill all your work units and see if the newly downloaded ones are working. If they aren't please let me know.

Jake
[/edit]

Nice work Jake thanks!

Hi, something going wrong with the new version MilkyWay@Home v1.39 (opencl_ati_101).
When I see new version I resume the project and down load new exe and 60 WUs. They mostly start with "Completed, validation inconclusive" and then "Completed, can't validate".
After 60 WUs only "18" WUs validated "10" invalid and others waiting for validation inconclusive.

I stop downloading new WUs.

---

Did something happen with v1.38? I'm getting v1.39. Everything is working good with v1.39.

I am getting errors such as this with 1.39 .... Please note that I did NOT manually abort

Name de_modfit_fast_19_3s_136_ModfitConstraints1_2_1471352126_40779419_2
Workunit 1314427731
Created 19 Sep 2016, 21:07:21 UTC
Sent 19 Sep 2016, 21:07:59 UTC
Report deadline 1 Oct 2016, 21:07:59 UTC
Received 19 Sep 2016, 21:09:05 UTC
Server state Over
Outcome Computation error
Client state Aborted by user
Exit status 201 (0xc9) EXIT_MISSING_COPROC
Computer ID 527314
Run time
CPU time
Validate state Invalid
Credit 0.00
Device peak FLOPS 2,135.76 GFLOPS
Application version MilkyWay@Home v1.39 (opencl_nvidia_101)
Stderr output

<core_client_version>7.4.36</core_client_version>

Hi, something going wrong with the new version MilkyWay@Home v1.39 (opencl_ati_101).
When I see new version I resume the project and down load new exe and 60 WUs. They mostly start with "Completed, validation inconclusive" and then "Completed, can't validate".
After 60 WUs only "18" WUs validated "10" invalid and others waiting for validation inconclusive.

I stop downloading new WUs.

Interestingly I test extra 30 WUs and found this. I change app_config and 2 wu per GPU to 1 wu. and see that a few ones invalid. most of them validated others waiting for validation. only 2 of them in "validation inconclusive"

continue testing with new WUs.

---

The nvidia opencl 101 ver. 1.38 is broken. GPU load is zero and the GPU clock stays at idle level and the units take nearly the fourfold time than they used to complete... :/

---

Aloha, Uli

Mine are all Computational errors now (two computers).

Below is typical error text



<core_client_version>7.2.42</core_client_version>
<![CDATA[
<message>
process exited with code 1 (0x1, -255)
</message>
<stderr_txt>
<search_application> milkyway_separation 1.38 Linux x86_64 double OpenCL </search_application>
Reading preferences ended prematurely
BOINC GPU type suggests using OpenCL vendor 'NVIDIA Corporation'
Setting process priority to 0 (13): Permission denied
Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4'
Switching to Parameter File 'astronomy_parameters.txt'
Using AVX path
Found 1 platform
Platform 0 information:
Name: NVIDIA CUDA
Version: OpenCL 1.2 CUDA 8.0.20
Vendor: NVIDIA Corporation
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_fp64 cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_copy_opts
Profile: FULL_PROFILE
Using device 0 on platform 0
Found 1 CL device
Device 'GeForce GTX 970' (NVIDIA Corporation:0x10de) (CL_DEVICE_TYPE_GPU)
Board:
Driver version: 361.42
Version: OpenCL 1.2 CUDA
Compute capability: 5.2
Max compute units: 13
Clock frequency: 1316 Mhz
Global mem size: 4294246400
Local mem size: 49152
Max const buf size: 65536
Double extension: cl_khr_fp64
Build log:
--------------------------------------------------------------------------------

ptxas info : 0 bytes gmem
ptxas info : Compiling entry function 'probabilities' for 'sm_52'
ptxas info : Function properties for probabilities
ptxas . 0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info : Used 106 registers, 420 bytes cmem[0], 152 bytes cmem[2]
--------------------------------------------------------------------------------
Error creating program (-6): CL_OUT_OF_HOST_MEMORY
Error creating summarization program from source
Failed to calculate likelihood
<background_integral> nan </background_integral>
<stream_integral> nan nan nan </stream_integral>
<background_likelihood> nan </background_likelihood>
<stream_only_likelihood> nan nan nan </stream_only_likelihood>
<search_likelihood> nan </search_likelihood>
20:37:38 (27201): called boinc_finish(1)

</stderr_txt>

I'm not getting any units for Mac. Everything says "Mon Sep 19 ... 2016 | Milkyway@Home | No tasks are available for MilkyWay@Home"

---

Plus SETI Classic = 21,082 WUs

Hey everyone,

I'll look into it a bit in the morning once the dust has started to settle a bit. Windows 1.38 GPU versions are known not to utilize the GPU, there is a 1.39 version released to fix this. It looks like the invalid results are being caused by the erroring work units not an issue with different answers. This should clear up on its own after the application stabilizes a it.

Looks like certain Linux 64 bit CL applications are erroring due to CL_OUT_OF_MEMORY. I am thinking of disabling the application Linux nvidia app for now until I know more.

As for supporting out of date cards, it seems updating the BOINC libraries nixed support for some cards. That's a bit out of my control, I might look into this a bit in a couple days.

Mac users should still be getting work units, I will take a look at why you stopped though.

On a side note, I fixed the server running out of work units, so that should no longer be a problem.

I think this hit all the major reply since I last checked. If I missed your post I am sorry. I'll try to read through again when I get a chance.


Jake

[edit]

I was able to reproduce the Linux 64bit GPU bug on a different computer so I should have a fix out for that hopefully tomorrow. I will revert that application to 1.36 for now though so there is something stable to crunch.

[/edit]

What about:
"Using a target frequency of 60.0" not conforming to "MilkyWay@Home preferences"?
Will it be coupled with preferences like in old versions, or will it be removed from logs (cause it doesn't give any info in reality if it's not coupled to preferences)?

On a side note, I fixed the server running out of work units, so that should no longer be a problem.

It seems it's not true. My PC still runs out of work. I have event log if you want. Last few days my PC computes only 30-35% of time.
Here's the link:
http://www.filedropper.com/boinceventlog

Workunits to my Macs started flowing around 2:00 am or earlier, depending on the machine. Thank you.

---

Plus SETI Classic = 21,082 WUs

Hey everyone,

Just got fix working for the Linux 64 bit OpenCL versions. Should have that pushed out in about a half hour. Sorry for that.

Jake

[edit]

You should start seeing 1.39 applications coming through now. Let me know if there is any trouble.

[/edit]

Jake -

How will I know when I get "the 64-bit fix"? I just downloaded some work and it is version 1.36. That is OK because they seem to be running without error.

I was expecting some new version number (1.40?, 1.41?).

Edit: I guess the fix is in 1.39

Edit: Just downloaded some 1.39 and they are running without error.

wb8ili,

Thanks for letting me know! I'm glad they are working well for you.

Jake

@Jake
MilkyWay@Home v1.39 (opencl_nvidia_101) are processing without error !
They are now waiting for validation. I stopped downloading to see whether the validation will happen. If yes I will proceed.

* Windows 10 home Core x64 Edition
* Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz
* NVIDIA GeForce GTX 1060 3GB (3072MB) driver: 372.70 OpenCL: 1.2

* runtime < 50 sec

Hi Jake,

There's too much fiddling to the server;
Serverpage got stuck
The validation-mountain is growing fast.....

Ooops........?!?

aad,

Nothing to worry about, the Validation queue was a mountain this morning, its down to a large hill now. Seems to be improving, but I will keep monitoring it.

Thanks for the report.

Jake