Posts by terencewee*
log in
1) Message boards : Windows : I have 32-bit fp GPU, is it good for collatz? (Message 13800)
Posted 2163 days ago by terencewee*
Your suggestion about testing with it on and off makes me think that you don't know if CF would have a improvement on collatz from the developer POV, so let them affirm or deny if this, what you write is your opinion and nothing else at the seems, if you know better explain me in detail, I love technical details.

Simmer down - mikey is trying to help. It's a process of elimination.
Look at the date mikey joined the project & his credits.

If an experienced person is giving me advice, I will be *very-grateful*.
2) Message boards : Number crunching : Linux ( ubuntu 10.10 64 bit) twice slower then Win (Message 12000)
Posted 2537 days ago by terencewee*
For now using 64 bit Linux app is waste of energy and time :(

I'm still running one linux x64 cruncher - will be switching to Win next few weeks.
3) Message boards : Number crunching : Linux 32-bit ATI App (Message 11882)
Posted 2547 days ago by terencewee*

If it's possible, install a 64 bit linux. That way you won't have that problem.

I was having some free time last week and had reinstalled the machine with 64bit linux (ubuntu 10.10 x64).

I'm happy to report after following required steps - we got (mini) joy!

Crunching normal WU will hang the system after a random number of hours. I can't be sure whether it's the app or the BOINC client, but I would go with the former as...

Crunching mini WU finishes in about 5hrs40min mark, works fine and validated (here and here).

@Crunch3r: since your are the original Crunch3r, I've named the system as cruncher02. :)

Thanks guys.
4) Message boards : Number crunching : Linux 32-bit ATI App (Message 11765)
Posted 2559 days ago by terencewee*
It seems like it isn't picking up the BRT_PERMIT_READ_WRITE_ALIASING variable for some reason. After setting it, when you run printenv does it show up? If not, have you tried logging out and then back in so it will reload the system profile?

research@jeeven:~$ printenv
SSH_CLIENT=xx.xx.xx.xx xxxxx xxxx
SSH_CONNECTION=xx.xx.xx.xx xxxxx xxxx
LESSOPEN=| /usr/bin/lesspipe %s
LESSCLOSE=/usr/bin/lesspipe %s %s

I had also tried restarting the machine.
BOINC-client is running as a daemon.

I know ATI made the suggestion that read-write was not suggested. I wonder if newer versions of the CAL no longer support that option. I'll have to do some research on that.

Anything else you want me to find out over here?

Thanks Slicker - hope to get collatz running on linux GPU-crunchers.
5) Message boards : Number crunching : Linux 32-bit ATI App (Message 11756)
Posted 2560 days ago by terencewee*
I'd been tinkering with linux GPU apps these few weeks as I wanted to learn as much first hand before committing to building new crunchers.

I am now trying out collatz-app for linux-32 ATI.

boinc-client start up messages:
1: 17-Mar-2011 02:27:45 (low) [] Starting BOINC client version 6.12.15 for i686-pc-linux-gnu
2: 17-Mar-2011 02:27:45 (low) [] Config: report completed tasks immediately
3: 17-Mar-2011 02:27:45 (low) [] Config: use all coprocessors
4: 17-Mar-2011 02:27:45 (low) [] Config: GUI RPC allowed from:
5: 17-Mar-2011 02:27:45 (low) [] Config:
6: 17-Mar-2011 02:27:45 (low) [] Config: localhost
7: 17-Mar-2011 02:27:45 (low) [] Config:
8: 17-Mar-2011 02:27:45 (low) [] log flags: file_xfer, sched_ops, task
9: 17-Mar-2011 02:27:45 (low) [] Libraries: libcurl/7.19.7 OpenSSL/0.9.8k zlib/ libidn/1.15
10: 17-Mar-2011 02:27:45 (low) [] Data directory: /var/lib/boinc-client
11: 17-Mar-2011 02:27:45 (low) [] Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU E4600 @ 2.40GHz [Family 6 Model 15 Stepping 13]
12: 17-Mar-2011 02:27:45 (low) [] Processor: 2.00 MB cache
13: 17-Mar-2011 02:27:45 (low) [] Processor features: fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc arch_perfmon pebs bts aperfmperf pni dtes64 monitor ds_cpl est tm2 ssse3 cx16 xtpr pdcm lahf_lm
14: 17-Mar-2011 02:27:45 (low) [] OS: Linux: 2.6.32-29-generic
15: 17-Mar-2011 02:27:45 (low) [] Memory: 1.96 GB physical, 5.75 GB virtual
16: 17-Mar-2011 02:27:45 (low) [] Disk: 287.75 GB total, 269.92 GB free
17: 17-Mar-2011 02:27:45 (low) [] Local time is UTC +8 hours
18: 17-Mar-2011 02:27:45 (low) [] ATI GPU 0: ATI Radeon HD 2300/2400/3200 (RV610) (CAL version 1.4.900, 256MB, 50 GFLOPS peak)
19: 17-Mar-2011 02:27:45 (low) [Collatz Conjecture] Found app_info.xml; using anonymous platform
20: 17-Mar-2011 02:27:45 (low) [SETI@home] Found app_info.xml; using anonymous platform

ATI RV610 detected, CAL 1.4.900 (Catalyst 10.12) - runs DNETC fine (albeit extremely slow, but hey! it works and WU got validated).

Downloaded [collatz_2.09_i686-pc-linux-gnu__ati14.tar.bz2 (Linux 32-bit ATI - EXPERIMENTAL!!!) 2.09 ATI CAL 1.4 01/17/2010 21:32 CST]

unpacked, copied *.so to /usr/lib, ldd = => (0x00a43000) => /lib/tls/i686/cmov/ (0x00207000) => /usr/lib/ (0x008c7000) => /usr/lib/fglrx/ (0x00779000) => /lib/tls/i686/cmov/ (0x0022d000)
/lib/ (0x00a09000) => /lib/tls/i686/cmov/ (0x0017b000) => /usr/lib/ (0x00d36000) => /lib/ (0x00ca3000) => /lib/tls/i686/cmov/ (0x0019f000) => /lib/tls/i686/cmov/ (0x0059c000) => /usr/lib/ (0x0049b000) => /usr/lib/ (0x005a0000) => /usr/lib/ (0x00471000) => /usr/lib/ (0x00b9a000) => /usr/lib/ (0x003db000)

Added BRT_PERMIT_READ_WRITE_ALIASING=1 into /etc/profile

Added collatz project and got WUs.

Computation error thru-out.

An example:

process exited with code 1 (0x1, -255)

Running Collatz Conjecture (3x+1) ATI GPU application version 2.09 by Gipsel (Win32, CAL 1.3)
instructed by BOINC client to use device 0
Reading input file ... done.
Checking 824633720832 numbers starting with 2368591747413562861928
CAL Runtime: 1.4.900
Found 1 CAL device

Device 0: ATI Radeon HD2350/2400/3200/3400 (RV610/620) 256 MB local RAM (remote 247 MB cached + 875 MB uncached)
GPU core clock: 0 MHz, memory clock: 0 MHz
40 shader units organized in 1 SIMDs with 8 VLIW units (5-issue), wavefront size 32 threads
not supporting double precision

Initializing lookup table (16384 kB) ... done
Starting WU on GPU 0
Copy lookup table to GPU memory (16384 kB)
Initialize step array on GPU (64 MB)
predicted runtime per iteration is 2 ms (33.3333 ms are allowed)
borders of the domains at 0 4096
No checkpoint data found.
0 Kernel Execution : Input stream is same as output stream.
Binding kernels read-write is prohibited.

02:28:45 (1514): called boinc_finish


I've tried commenting out export BRT_PERMIT_READ_WRITE_ALIASING=1; same result.

Can anyone offer help?


6) Message boards : News : Mini Collatz App Released (Message 11262)
Posted 2603 days ago by terencewee*

Let me clarify:-
ATI 5850 (normal WUs) and nVidia 8800gtx (slower, mini WUs) on the same machine.

7) Message boards : News : New Application Coming Soon (Message 11251)
Posted 2603 days ago by terencewee*

Collatz credit is based on a 100 credit per day machine earning 100 credits per day with the stock app running 32-bit Windows. DNETC credit is based upon.... well, they just picked a number bigger than Collatz but less than MW (see

Ahso-des-ka - we can't always please everyone like you said.

By my calculation, MW ratio is 1.86. Collatz (depending on WUs) falls between 1.98-2.14.
8) Message boards : News : Mini Collatz App Released (Message 11250)
Posted 2603 days ago by terencewee*

Thanks for giving us the choice to choose between mini and normal Collatz.

Recently I installed my (extremely-old-but-very-trusty) 8800gtx into one of my cruncher. I'm wondering whether it is possible to configure BOINC-client to fetch the normal WU for my ATI-5850 and the mini WU for 8800gtx-cuda... Is this possible?

Thanks again!
9) Message boards : News : New Application Coming Soon (Message 11226)
Posted 2605 days ago by terencewee*
I might change the validator to grant a little more credit to the large ones to encourage people to run those instead of the mini ones (either that or reduce the credit a little for the smaller ones).

Yes, please. DNetC ratio vs. time = ~2.61, while Collatz is ~1.95
I prefer Collatz due to GPU being cooler by 3 Celsius. :)
10) Message boards : Number crunching : GPU Utilization Varies (Message 10750)
Posted 2638 days ago by terencewee*
So now I know who's been sneaking up my back... :w

Awwww... I thought you wanted company waaaaay up high there.

At 35m+ it'll take me forever to "sneak" up on you - hope to keep you warm when I do get there. :)

To the Project!
11) Message boards : Number crunching : GPU Utilization Varies (Message 10749)
Posted 2638 days ago by terencewee*

Check you power setting for the laptop, they all have some kind of shut this or that off after x amount of time etc, I have set all of mine to stay on all the time and do not see the slowdowns you are seeing. Also turn off the screensaver, but physically turn the monitor off like you are now.



I've checked the notebook BIOS, but there is nothing for power management. But I've since narrowed it down to the GPU+driver.

Apparent when the LCD (internal or external) entered switch-off-mode, the GPU will automatically throttle to 50% for both speed and memory.

Putting the screen saver to blank but NOT switch off - the GPU continue to run at 100%. I think I'll need to go give HP+AMD a kick.

Thanks again.
12) Message boards : Number crunching : GPU Utilization Varies (Message 10743)
Posted 2640 days ago by terencewee*
I'm using a FirePro Mobility m7820 (a 5770 single precision part) in a HP Elitebook 8740w. Driver package is 8.743.3.2. Catalyst AI is set to Standard, PowerPlay is Plug-in = Maximum Performance.

I noticed this strange behavior:

When I'm using my notebook, the time to process a WU is ~14mins. The GPU load is 98% (GPU-Z).

When I'm not using the notebook but left it to process WUs (after the Blank screen saver kicks in and the screen switched off) - the time to process a WU is ~30mins. The GPU load is then 50%.

Can someone advice what to do to make the GPU run at peak?


Main page · Your account · Message boards

Copyright © 2018 Jon Sonntag; All rights reserved.