Donate To Seti@HomeSeti@Home optimized science apps and information
 
Welcome, Guest. Please login or register.
24 Apr 2014, 06:51:01 pm

Login with username, password and session length
 
» Home
» Forums
» Downloads
» FAQ
» News

» Search site
 
 
 
If you've registered already but never got your activation email, please click here.
 
 
Seti@Home optimized science apps and information  |  Optimized Seti@Home apps  |  Windows  |  Topic: Installer v0.40 release notes 0 Members and 0 Guests are viewing this topic. « previous next »
Pages: [1] Go Down Print
Author Topic: Installer v0.40 release notes  (Read 6632 times)
William
Global Moderator
Knight Templar
*****
Offline Offline

Posts: 313



Installer v0.40 release notes
« on: 26 Mar 2012, 09:37:31 am »

Lunatics are pleased to announce the availability of the v0.40 installer in the Lunatics main download area
Please take care you use the right installer (32/64 bit) for your system.

Should you have any questions or problems - ask! Post in this thread or open a new one.

New in this installer:
Update of all Astropulse applications to support AP v6.01.

How to use the installer:
For the CPU applications, you need to know what your CPU supports. Either check the Boinc startup messages or check with CPU-Z downloadable from http://www.cpuid.com
Tick or untick CPU AP and MB applications as desired. Tick CUDA MB if desired.
Tick ATI (OpenCL) AP or MB if/as desired.
More details further down.

NB We use default values throughout the app_info.xml. If you have previously installed optimised applications and manually edited app_info.xml (e.g. <count> for multithreading GPUs, <flops>, <cmdline> for ATI) you WILL lose those edits on running the installer and have to redo them. You may wish to make a note of them. You will also lose all other manual edits to app_info.xml such as beta applications.

Known issues - general:
The installer is supposed to shut down the BOINC client whilst leaving the Manager running - noticable from the red 'disconnected' dot in the tray. After finishing the BOINC client should restart and the red dot vanish when the Manager reconnects. This make take a few minutes. On some systems however this fails to work. Please ensure BOINC has restarted.
Advanced users only: If you know you are going to edit app_info.xml, it may be easier to shutdown and restart BOINC manually, so you can get the changes into app_info.xml before the restart.


CPU MB apps - AKv8b2

If you have any doubt at all about what instruction sets your CPU supports, download CPU-Z from http://www.cpuid.com/softwares/cpu-z.html You will usually find that the SSSE3x application is fastest on the Intel processors which support SSSE3x, except on dual-core CPUs with especially fast memory subsystems,where SSE4.1 may be faster.
However, we have an indication, that on Intel i3, i5 and i7 processor series SSE3 may be the fastest variant at most angleranges. The effect is smallest with VHAR (shorties).


CPU AP apps - r555/r557

The new r555/r557 AP applications support AP v6.01 and v5.05 work.
If your CPU is only capable of SSE choose the r555 SSE application.
If your CPU supports AVX (Intel Sandybridge and AMD Bulldozer, Windows 7 SP1) choose the r557 SSE2 build as it is faster.
If you have a CPU with SSE2 and above but not AVX the r557 MAY be faster.
We have indication that on most hosts r557 is faster but no guarantees.
As always - your mileage may vary.


NVidia GPU (CUDA) app - x41g_cuda32

For compatible cards see http://developer.nvidia.com/cuda-gpus
Requires minimum CUDA 3.2 capable NVidia driver: 263.06 (260.99 on notebooks)
Download drivers from http://www.nvidia.com/Download/Find.aspx?lang=en-us (or your national equivalent). We recommend you avoid driver versions 295.73 and 296.10, which are not supported at some other BOINC projects.
Should run on all NVidia CUDA cards with at least 512MiB video RAM.
Actual free memory required is in the region of 200-235MB, depending on driver and OS flavour.
Running on a 256MiB card is just possible, but you will have to free up as much VRAM as possible, e.g. by disabling Aero.
Used video RAM can be checked with GPU-Z downloadable from http://www.techpowerup.com/downloads/2120/TechPowerUp_GPU-Z_v0.6.0.html
Watch closely on GPU-Z and in BOINC manager if the application finds enough memory.
Symptoms of insufficient memory are
- in BOINC manager: tasks start up and run a few seconds then go to 'waiting to run' and the next task tries to start
- on GPU-Z: the sensor for memory shows used VRAM ramping up and almost immediately dropping again.
In this case, there may be a BOINC bug putting the app in an infinite loop while it waits for more memory.
If you don't have enough memory, set BOINC to NNT, abort tasks and uninstall the CUDA application by rerunning the installer with CUDA MB UNticked.
If in doubt ask for help on the SETI Number Crunching forum.
To multithread on Fermi cards, find all instances of <count>1</count> in app_info.xml and decrease to 0.5 or 0.3 to run two or three tasks in parallel.
Again, ifá you are uncertain of how to do this correctly, ask on the above mentioned forum. 'Your mileage may vary' - it depends on your specific system which count will give the highest throughput.
x41g has improved reliability, stability and accuracy over x38g.
x41g_cuda32 has been tested with and is safe to use on GTX 680 (Kepler) GPUs.


ATI MB apps - r390; AP apps - r555

We offer the MB application in two different variants, depending on card:
plain - for most HD4xxx owners and those with HD5xxx and HD6xxx cards who suffer from driver incompatibilities.
HD5 - for most if not all owners of HD5xxx GPUs and HD6xxx GPUs.

The ATI MB application will not work on ATI cards with workgroup size 128 (e.g. HD43xx). To check workgroupsize run CLinfo, downloadable (at present) from {http://boinc.berkeley.edu/dl/clinfo.zip}
It will, however, probably work on HD53xx & HD63xx with workgroup size 128.
If you have problems running the HD5 flavour (anything from reduced performance to BSOD), you should run the plain variant.

The recommended apps (r555 for Astropulse, r390 for MultiBeam) need OpenCL support from the ATI driver - for recent driver versions, look for a driver *with* APP or OpenCL listed. For compatible drivers see http://support.amd.com/us/gpudownload/Pages/index.aspx
For cards, see http://developer.amd.com/gpu/AMDAPPSDK/pages/DriverCompatibility.aspx
On most systems GPU-Z (see above) will tell you if your setup supports OpenCL.
Minimum required driver: Something that is OpenCL capable. You can check with GPU-Z (see above).
Catalyst 10.7b with SDK
Catalyst 10.10 Accelerated Parallel Processing (APP) Technology Edition
The following drivers are known to be incompatible with the app:
11.10 and 11.11.
On some drivers the app shows increased CPU usage.
Drivers from 11.12 onwards work.
First official driver for HD7xxx is 12.3.

For cards without OpenCL support (from HD2xxxx on) use the 'hybrid' r555 Astropulse application. NB this is a combined CPU/GPU application and will only use the GPU for some of the calculations, while doing the rest on a CPU core.
Minimum driver version Catalyst 9.x.

NB using this installer will give you arbitrary default cmdline parameter entries.
If you are already running ATI applications: Make a note of what numbers you are using to be able to restore to those values PRIOR to running the installer.
This also applies to setting the <count> variable according to your -instances_per_device variable for multithreading. (i.e. a <count> setting of 0.5 requires a -instances_per device setting of 2)

The Hybrid AP application uses about 80% of a CPU core and does only about 20% of the calculations on the GPU. Parameters have been choosen to reflect this usage. However present BOINC can not reserve '0.8' of a CPU core and therefore reserves none. This means your CPUs will be overcommitted and runtimes and DCF will suffer. You do increase the overall output though.

Summary of available cmdline options (wording partially by Raistmer):

AP/MB:
-instances_per_device N how many tasks you want to run in parallel. Inverse of <count>.

-hpá gives the app high priority

-no_cpu_lock á prevents the app from using only a specific CPU core


AP only:
-v505á to process AP 5.05 tasks

-sbs 128 á áis the max size of single buffer that can be used in program. Lower limit is 128MB, upper - max size allowed particular card. [Note: accepted but not implemented]

-unroll 4 Optimal at half the number of Compute Units of the GPU. Lower values also reduce VRAM use. Decrease if you experience lags.

-ffa_block 2048á defines how many different periods GPU will process per single kernel call

-ffa_block_fetch 1024á defines how many threads will be used in FFA initial fetch kernel

ffa_block should be divisible by ffa_block_fetch. Going too high will result in premature 30/30 exit errors.


MB only:
-period_iterations_num 20á splits calls into smaller portions. Increase if experiencing lag.


General Notes:
With Windows Vista and Windows 7, do not install BOINC in 'Service' or 'Protected Application Execution' mode for GPU crunching, and do not connect to a Vista/7 computer with Windows Remote Desktop while it is running a GPU application.
Switching users can also cause the GPU app to fail [BOINC versions before 6.12.28].

Tuning cmd line parameters:
For AP: If you get lags or worse driver restarts, decrease unroll. If you run stable you can try to increase it. ffa_block should be a multiple of ffa_block_fetch. Start low e.g. -ffa_block 4096 -ffa_block_fetch 2048 and try increasing if it runs fine. Sticking to the power of 2 line at first can help.
There will be a system specific best combination - too low slows you down, too high slows you down as well.
Remember AP runtimes depend on blanking % so only compare runtimes on similar blanking %.
For MB: other way round - if you experience lag or driver restarts increase -period_iteration_num

Thanks go to Jason and Raistmer for their continued work to improve applications and to Josef W. Segur for invaluable help with everything.

We would also like to take this opportunity to thank our active alpha testers (in no particular order and apologies to anybody we've forgotten):
Claggy, Jamie, arkayn and Mike.

The Lunatics Installer team AD MMXII
« Last Edit: 30 Mar 2012, 09:48:19 am by barry » Logged
Fredericx51
Knight o' The Round Table
***
Offline Offline

Posts: 207


Knight Who Says Ni N!


Re: Installer v0.40 release notes
« Reply #1 on: 24 Apr 2012, 10:43:25 am »

Hi, I tried a lot of different Command-Line_parameters for MB on ATI 5870 GPU, after 10
seconds the driver (Cat. 11.5;IIRC) crashes, but does recover right away.
However GPU-Load, is 1 til 3% on GPU 0, second GPU doesn't get any (visible) load, progress
also, stalls.

I re"nstalled Rev. 0.40 and only used the Astropulse GPU part!
CPU, an I7-2600 works OK with the AVX executable, running on BOINC 7.0.25 and WINDOWS 7
64bit, both. HT is turned ON, too. DRAM (DDR3) runs @ 1600MHz (1.6V), CPU runs stock.

Also use 1 core, (2 threads) for GPU handling. Makes the PC more responcive and faster, doing
something else, like Browsing the Web; using PAINT and PrtScr or scrats-pad/note pad.

Logged
Fredericx51
Knight o' The Round Table
***
Offline Offline

Posts: 207


Knight Who Says Ni N!


Re: Installer v0.40 release notes
« Reply #2 on: 28 Apr 2012, 12:05:11 pm »

First, I updated the Cat. 11.4 driver to a newer, 12.4 (x64) version, re´nstalled LUNATICs v0.40 (64bit) SSE3 + AVX version
for my I7-2600 CPU, used the SSSE3x version for Q6600 & QX9650 CPU's.
Don't forget to install the right, x86 or x64 version of B.O.I.N.C. 7.0.25

Then CUDA executables for my 2 C2Q(X) hosts and  AMD ATI executables for MB and AstroPulse, don't forget to
have a look at the READ ME file and app_info.xml file, as you maybe want to change some parameters.

I noticed, older drivers below CAT. 12.4 gave uncontrollable performance and far to low GPU-Load!!!
Also, NOT keeping a CORE, or 2 threads (HT=ON), free gives a better/faster performance and NO driver crashes
or restarts. (With 2 EAH 5870 GPU's with the above mentioned driver!)

UPdated all SETI running hosts, to BOINC 7.0.25, x86 or x64 and ran the 0.40 LUNATIC's installer, great work, b.t.w.,
but it is important to UPdate all  nVidia and or AMD/ATI, drivers and then run the 0.40 installer.
 All hosts are running like a charm  Grin




Logged
Pages: [1] Go Up Print 
Seti@Home optimized science apps and information  |  Optimized Seti@Home apps  |  Windows  |  Topic: Installer v0.40 release notes « previous next »
Jump to:  


Quote!
Every man is guilty of all the good he didn't do.
- Voltaire

 
Site Statistics
Total Members:95
Total Posts:53,716
Total Topics:1,512
Downloads
..Some PHP stuff ToDo
Pages served
Today:3,344
Total:19,186,943
(since 6/26/2006)
Latest Member:
juan BFB
 
 
  
 
Seti@Home optimized science apps and information | Powered by Enigma 2.0 (RC1).
© 2003-2014, LSP Dev Team. All Rights Reserved.
Seti@Home optimized science apps and information Forums | Powered by SMF.
© 2005, Simple Machines LLC. All Rights Reserved.
Powered by MySQL Powered by PHP Valid XHTML 1.0! Valid CSS!