Posts by DeleteNull

41) Message boards : Number crunching : Trial Factoring tests (Message 6159)
Posted 13 Apr 2020 by DeleteNull

Yes with elf-file.
But in the tests before (zip without elf-file) was the same error.

Standalone test with wrapper and job.xml succeeded. So I don't know what the problem is. Sofware is o.k., but if the wrapper is started inside BOINC it fails.

42) Message boards : Number crunching : Trial Factoring tests (Message 6157)
Posted 13 Apr 2020 by DeleteNull

On both of my PC with AMD GPU: error 195 from wrapper, with or without elf-file?!

43) Message boards : Number crunching : Trial Factoring tests (Message 6156)
Posted 13 Apr 2020 by DeleteNull

I get an error (two):
A new attached PC: Download error (checksum fail)
Previously attached: The old ZIP (without elf-file) remains, if you replace the file manually the file is deleted (zip error).

It seems you have to create a new version, sorry.

44) Message boards : Number crunching : Trial Factoring tests (Message 6154)
Posted 13 Apr 2020 by DeleteNull

All my "standalone tests" are okay too. I think the problem ist that mfakto-x86_64 starts a child process to compile the mfakto_Kernels.elf file.

I seems that this forces the CHILD_FAILED error.

Can you include this file:
https://p-numbers.net/mfakto_Kernels.elf
in the ZIP?

(Just for a test)

45) Message boards : Number crunching : Trial Factoring tests (Message 6152)
Posted 13 Apr 2020 by DeleteNull

Thanks, now I get a new error:

process exited with code 195

This is an error from BOINC (wrapper)
195 is CHILD_FAILED,
Which means the wrapper failed to start the worker app.

46) Message boards : Number crunching : Trial Factoring tests (Message 6144)
Posted 12 Apr 2020 by DeleteNull

I have extended the parameter list for mfakto. (and compiled)

You can download it frome here:
https://p-numbers.net/mfakto_linux_x86_64_device.zip

HTH

the new code snippet:
else if(!strcmp((char*)"--device", argv[i]))
{
if(i+1 >= argc)
{
printf("ERROR: no device number specified for option \"--device\"\n");
return ERR_PARAM;
}
else if (argv[i+1][0] == '0') // parse to 00
{
devicenumber = strtol("00",&ptr,10);
if(*ptr || errno || devicenumber != strtol("00",&ptr,10) )
{
printf("ERROR: can't parse <device number> for option \"--device\"\n");
return ERR_PARAM;
}
}
i++;
}[/url]

47) Message boards : Number crunching : Trial Factoring tests (Message 6140)
Posted 12 Apr 2020 by DeleteNull

ERROR: unknown option '--device'

The application mfakto doesn't know the option --device

48) Message boards : Number crunching : Trial Factoring tests (Message 6137)
Posted 12 Apr 2020 by DeleteNull

It's running:
wolfgang@9600k:~/BOINC/projects/srbase.my-firewall.org_sr5$ ./wrapper_26012-v2_x86_64-pc-linux-gnu
mfakto 0.15pre6 (64bit build)

Runtime options
Inifile mfakto.ini
Verbosity 1
SieveOnGPU yes
MoreClasses yes
GPUSievePrimes 82486
GPUSieveProcessSize 24 Kib
GPUSieveSize 96 Mib
FlushInterval 8
WorkFile worktodo.txt
ResultsFile results.txt
Checkpoints enabled
CheckpointDelay 300 s
Stages enabled
StopAfterFactor class
PrintMode compact
V5UserID none
ComputerID none
TimeStampInResults no
VectorSize 2
GPUType AUTO
SmallExp no
UseBinfile mfakto_Kernels.elf
Compiletime options

Select device - Get device info:
WARNING: Unknown GPU name, assuming GCN. Please post the device name "gfx1012 (Advanced Micro Devices, Inc.)" to http://www.mersenneforum.org/showthread.php?t=15646 to have it added to mfakto. Set GPUType in mfakto.ini to select a GPU type yourself to avoid this warning.

OpenCL device info
name gfx1012 (Advanced Micro Devices, Inc.)
device (driver) version OpenCL 2.0 AMD-APP (3075.9) (3075.9 (PAL,LC))
maximum threads per block 1024
maximum threads per grid 1073741824
number of multiprocessors 11 (704 compute elements)
clock rate 1855 MHz

Automatic parameters
threads per grid 0
optimizing kernels for GCN

Compiling kernels.
GPUSievePrimes (adjusted) 82486
GPUsieve minimum exponent 1055144
Started a simple selftest ...
Selftest statistics
number of tests 30
successful tests 30

selftest PASSED!

got assignment: exp=78467119 bit_min=70 bit_max=71 (3.05 GHz-days)
Starting trial factoring M78467119 from 2^70 to 2^71 (3.05GHz-days)
Using GPU kernel "cl_barrett15_73_gs_2"
Date Time | class Pct | time ETA | GHz-d/day Sieve Wait
Apr 12 19:43 | 116 2.4% | 0.553 8m38s | 495.97 82486 0.00%

49) Message boards : Number crunching : Trial Factoring tests (Message 6135)
Posted 12 Apr 2020 by DeleteNull

@Rebirther: I am sorry to read that.

Tests with worktodo.txt:
In the self compiled stuff: o.k. runtime about 8 min.
In the unpacked ZIP: Failure.

In the ZIP is contained: mfakto_Kernels.elf
This will work only for the specific hardware and Linux for what this file was compiled for. This file has to be deleted ... it will be compiled during runtime in the target system.

After deletion of mfakto_Kernels.elf the test is also running for the ZIP content.

I have made my own app_info.xml.....but project doesn't accept this.

50) Message boards : Number crunching : Trial Factoring tests (Message 6131)
Posted 12 Apr 2020 by DeleteNull

I can do this around 18:00 UTC+2. Now it's time to visit the 80+ generation for Easter. (Corona social task force)

51) Message boards : Number crunching : Trial Factoring tests (Message 6129)
Posted 12 Apr 2020 by DeleteNull

Thanks.

results are failing:
13:20:23 (13094): wrapper: running ./mfakto-x64 ( -d 11 --device 0)

app_config.xml tells -d 11 but the argument --device 0 ist still there...

How can we deal with this?

52) Message boards : Number crunching : Trial Factoring tests (Message 6127)
Posted 12 Apr 2020 by DeleteNull

I have managed to install the AMD 20.10 driver for linux and the Radeon 5500XT now passes the selftest without errors.

Can you please allow new work for linux boxes with AMD GPU?

53) Message boards : Number crunching : Trial Factoring tests (Message 6123)
Posted 11 Apr 2020 by DeleteNull

Do you mean it's not recognized or it does not run?
Because even though my vega 8 isn't recognized it still runs.

The device is recognized and calculation starts but fails.
To test if this is a driver issue I have replaced the Radeon 5500XT with an R9 380 and now the selftest is completed with success.

But now I can't get new work to test if "AMD with Linux x86_64" is running or not?!

54) Message boards : Number crunching : Trial Factoring (Message 6076)
Posted 9 Apr 2020 by DeleteNull

The results of a Radeon 5500XT (Linux x86_64) you can see here https://p-numbers.net/Radeon_5500XT_Linux-x86_64.txt

55) Message boards : News : Trial Factoring - new subproject (beta) (Message 6059)
Posted 8 Apr 2020 by DeleteNull

o.k.?
https://www.gpu72.com/

We can name it GPU95 because 95 bit is the limit of this software ;)

56) Message boards : News : Trial Factoring - new subproject (beta) (Message 6057)
Posted 8 Apr 2020 by DeleteNull

There will be a name change of the project due a not permitted name from the website operator.

Which name is not permitted?
Who gives the permission?

57) Message boards : Number crunching : Trial Factoring tests (Message 6044)
Posted 7 Apr 2020 by DeleteNull

I will try an app_config.xml
With this app_config.xml I have no more device error:

<app_config>
<app>
<name>GPU72</name>
<max_concurrent>1</max_concurrent>
</app>
<app_version>
<app_name>GPU72</app_name>
<plan_class>opencl_ati_101</plan_class>
<cmdline>-d 11</cmdline>
</app_version>
</app_config>

...but Radeon 5500 XT is not compatible with this software (like AP27 in PrimeGrid)

58) Message boards : Number crunching : Trial Factoring tests (Message 6040)
Posted 7 Apr 2020 by DeleteNull

You can download the linux version from here:
https://p-numbers.net/mfakto_linux_x86_64.zip

59) Message boards : Number crunching : Trial Factoring (Message 6039)
Posted 7 Apr 2020 by DeleteNull

Just copy the file libcudart.so.10.1 into your /lib64 folder (or whatever your distribution needs)

Last time 'just copy this to that' worked for me was under DOS. This is not DOS, this is Linux. libcudart.so.10.1 is totally unknown yet in the repo that supplies my software.

It goes even further. Under Linux, libcudart.so.10.1 is NOT part of the official CUDA 10.1 toolkit, not even of the official CUDA 10.2 toolkit. Both come with libcudart.so.9.1. You can only get libcudart.so.10.1 by adding an experimental PPA (Mageia) to your softwarelist. As libcudart.so.10.1 also has (inter)dependencies, merely copying the file in a directory will not be enough -but I wrote that earlier.

I followed the instructions in FAQ and found libcudart.so.10.1 in one of the folders.
For all other PC's I copied libcudart.so.10.1 into the /lib64 folder and this works perferctly (for me).

60) Message boards : Number crunching : Trial Factoring (Message 6031)
Posted 5 Apr 2020 by DeleteNull

I'm not getting any work on this host:
http://srbase.my-firewall.org/sr5/show_host_detail.php?hostid=205419

Previous 20 · Next 20