v7 issues on very old cruncher

Message boards : Number crunching : v7 issues on very old cruncher
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 · Next

AuthorMessage
Profile j mercer
Avatar

Send message
Joined: 3 Jun 99
Posts: 2422
Credit: 12,323,733
RAC: 1
United States
Message 1373908 - Posted: 31 May 2013, 0:07:04 UTC

I'm having some issues with my original cruncher a 1995 200MHz Pentium MMx P55 128MB UW2SCSI. It error eight v7 wu before it grabbed a v6.

It has crunched everything up to this point. It is bare bones rig. I shut everything down even Explorer on it when crunching to get those few extra chunks of memory.

Not a major issue for now for it takes this machine three to fourteen days a wu. It's pulling 120 watts for that one wu so I only run it now when there is something new to test.

I double checked everything I believe and have read the forums with no luck yet. I would like to keep this machine crunching just for the nostalgic.


...
ID: 1373908 · Report as offensive
Wedge009
Volunteer tester
Avatar

Send message
Joined: 3 Apr 99
Posts: 451
Credit: 431,396,357
RAC: 553
Australia
Message 1373919 - Posted: 31 May 2013, 0:35:29 UTC

I don't know the details of the default CPU application, but I'm guessing they're assuming a minimum of SSE2 capability for CPUs nowadays. That's a possible reason for the tasks failing.
Soli Deo Gloria
ID: 1373919 · Report as offensive
spitfire_mk_2
Avatar

Send message
Joined: 14 Apr 00
Posts: 563
Credit: 27,306,885
RAC: 0
United States
Message 1373933 - Posted: 31 May 2013, 1:38:41 UTC

I would not run Xp on that, try Windows 2000.
ID: 1373933 · Report as offensive
Urs Echternacht
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 692
Credit: 135,197,781
RAC: 211
Germany
Message 1373944 - Posted: 31 May 2013, 2:36:31 UTC - in response to Message 1373908.  
Last modified: 31 May 2013, 2:44:57 UTC

Looking at the error message the problem seems to be related with the use of this new library : libfftw3f-3-3_upx.dll which is deployed with setiathome_7.00 on windows.

Checking the fftw release notes :
FFTW 3.1.1
March 18, 2006

Performance improvements for Intel EMT64.
Performance improvements for large-size transforms with SIMD.
Cycle counter support for Intel icc and Visual C++ on x86-64.
In fftw-wisdom tool, replaced obsolete --impatient with --measure.
Fixed compilation failure with AIX/xlc; thanks to Joseph Thomas.
Windows DLL support for Fortran API (added missing __declspec(dllexport)).
SSE/SSE2 code works properly (i.e. disables itself) on older 386 and 486 CPUs lacking a CPUID instruction; thanks to Eric Korpela.
i can only guess that Eric's fix for 3.1.1 is no longer active.

Maybe it works if you rename the old libfftw3f-1-1_upx.dll to the new name mentioned before ?

ps: maybe a more efficient psu could reduce the powerneed on that old museum piece host.
_\|/_
U r s
ID: 1373944 · Report as offensive
Profile betreger Project Donor
Avatar

Send message
Joined: 29 Jun 99
Posts: 11361
Credit: 29,581,041
RAC: 66
United States
Message 1373956 - Posted: 31 May 2013, 3:16:39 UTC - in response to Message 1373944.  
Last modified: 31 May 2013, 3:17:19 UTC

Where J Mercer lives electricity is relatively inexpensive, lots of hydro, so the cost to run an antique once in a while is small.
ID: 1373956 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 1373997 - Posted: 31 May 2013, 5:20:14 UTC - in response to Message 1373919.  

I don't know the details of the default CPU application, but I'm guessing they're assuming a minimum of SSE2 capability for CPUs nowadays. That's a possible reason for the tasks failing.

That is definitely not the issue. All functions have base versions which should run on any x86 CPU, checking capabilities and testing is done before any SIMD variants are used.

SETI@home v7 has been under test at SETI Beta for over a year, but of course there's a larger variety of hosts attached to this main project, so Murphy's Law almost guarantees there will be some that have problems.
                                                                   Joe
ID: 1373997 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22200
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1373999 - Posted: 31 May 2013, 5:32:51 UTC

A thought - Is JMercer reporting the estimated run time, or the actual run time for a task. I suggest this because the estimated run times on some v7 tasks on both my crunchers are wildly pessimistic, with estimates of an hour when the run time was about 15 minutes.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1373999 · Report as offensive
Profile j mercer
Avatar

Send message
Joined: 3 Jun 99
Posts: 2422
Credit: 12,323,733
RAC: 1
United States
Message 1374037 - Posted: 31 May 2013, 6:47:19 UTC

Great replies everyone. Thank you.

Wedge009, I was thinking same, possibly a minimum of SSE2 capability requirement.

spitfire_mk_2, I’ve tried most everything from Win95, NT4 and up with XP giving the best results on wu. Yes inter face is painfully slow but it crunches better than the others did.

Urs, when this wu finishes I’ll give that a try. My file’s name is ‘libfftw3f-1-1a_upx.dll’ with timestamp of 5/302013 3:27PM.

betreger, in 2010 it was $90US a year running at 100% for 24/7 with a RAC of 5. $.084786US per KWH in 2010.

Josef, is right, there will be glitches. LOL I’m not worried if it will still get v6 wu but would really like it to do v7. If it is found to be important enough it will get a fix. I’m surprised it is still crunching. I’ve only had to change the CMOS battery it's all original.

rob, it was showing about eight days a wu for each of the v7 and they error out in around 30 seconds. The last v6 wu were taking around 7-9 days and the estimated complete times were within minutes.

...
ID: 1374037 · Report as offensive
Wedge009
Volunteer tester
Avatar

Send message
Joined: 3 Apr 99
Posts: 451
Credit: 431,396,357
RAC: 553
Australia
Message 1374042 - Posted: 31 May 2013, 6:53:37 UTC - in response to Message 1373997.  

That is definitely not the issue. All functions have base versions which should run on any x86 CPU, checking capabilities and testing is done before any SIMD variants are used.

Good to know. (:

Soli Deo Gloria
ID: 1374042 · Report as offensive
Ianab
Volunteer tester

Send message
Joined: 11 Jun 08
Posts: 732
Credit: 20,635,586
RAC: 5
New Zealand
Message 1374062 - Posted: 31 May 2013, 7:27:55 UTC - in response to Message 1374037.  

Could it be running out of RAM?

I notice that it's only 128mb of RAM.. not very much by todays standards.

Ian
ID: 1374062 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1374167 - Posted: 31 May 2013, 10:29:59 UTC - in response to Message 1374062.  

Could it be running out of RAM?

I notice that it's only 128mb of RAM.. not very much by todays standards.

That would be a problem, even with an addon video card.
It's an eternity ago, but from what i can remember- Win98 was happy with 128MB, 256MB was about the most that would have any effect on performance.
For WinXP/2k 512MB was about the sweet spot, 256MB was generally considered the minimum for reasonable performance.
Grant
Darwin NT
ID: 1374167 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1374168 - Posted: 31 May 2013, 10:30:39 UTC - in response to Message 1373933.  

I would not run Xp on that, try Windows 2000.

WinXP was Win2k prettied up.
Grant
Darwin NT
ID: 1374168 · Report as offensive
Profile j mercer
Avatar

Send message
Joined: 3 Jun 99
Posts: 2422
Credit: 12,323,733
RAC: 1
United States
Message 1374373 - Posted: 31 May 2013, 16:00:40 UTC - in response to Message 1374062.  

Ianab, I believe the processor/ram is the issue too. It has been in the past and is why I run it bare boned. With anything extra running and Boinc will bog.

Grant, you are spot on for the sweet spot. XP loves 256 MB or more.

"The minimum hardware requirements for Windows XP HE/Pro include:
•Pentium 233-megahertz (MHz) processor or faster (300 MHz is recommended)
•At least 64 megabytes (MB) of RAM (128 MB is recommended)
•At least 1.5 gigabytes (GB) of available space on the hard disk"

Two and half more days and I can try it again. Wheee!
...
ID: 1374373 · Report as offensive
Profile Link
Avatar

Send message
Joined: 18 Sep 03
Posts: 834
Credit: 1,807,369
RAC: 0
Germany
Message 1374429 - Posted: 31 May 2013, 17:15:36 UTC - in response to Message 1374167.  

For WinXP/2k 512MB was about the sweet spot, 256MB was generally considered the minimum for reasonable performance.

That was called sweet spot when almost nobody could afford more than that, the actuall sweet spot for 32bit WinXP is somewhere between 2-3GB. Sure, it was less before, but 512MB wasn't enough for XP long time before Vista came out. I have 2GB in my Laptop, can't say that's more than I need, I'm often over 1.5GB and almost never below 1GB under normal operation. Of course, for a dedicated single core cruncher 512MB should be enough.
ID: 1374429 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1374472 - Posted: 31 May 2013, 18:37:25 UTC - in response to Message 1374429.  
Last modified: 31 May 2013, 18:38:13 UTC

For WinXP/2k 512MB was about the sweet spot, 256MB was generally considered the minimum for reasonable performance.

That was called sweet spot when almost nobody could afford more than that, the actuall sweet spot for 32bit WinXP is somewhere between 2-3GB. Sure, it was less before, but 512MB wasn't enough for XP long time before Vista came out. I have 2GB in my Laptop, can't say that's more than I need, I'm often over 1.5GB and almost never below 1GB under normal operation. Of course, for a dedicated single core cruncher 512MB should be enough.

An old PIII I use at work has 512MB, but just sitting there crunching it is at 112MB used. So 128MB seems to be just fine for a box to sit in a corner and chew on some work.
I do have an old Pentium 200 MMX machine I keep wanting to get going, but is a mini desktop with a nonstandard size AT PSU. :/
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1374472 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 19062
Credit: 40,757,560
RAC: 67
United Kingdom
Message 1374519 - Posted: 31 May 2013, 20:29:00 UTC

To me that 120W seems like a big power consumption for so small and RAC.

My q6600 with nvidia670 only uses 165W for an RAC that was ~17,000 used ~16hr/day.
ID: 1374519 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1374561 - Posted: 31 May 2013, 22:42:17 UTC - in response to Message 1374429.  

That was called sweet spot when almost nobody could afford more than that, the actuall sweet spot for 32bit WinXP is somewhere between 2-3GB.

Try 512MB.
You can give it 1GB, but it doesn't have much effect. Adding more than 1GB has no effect unless you're runing something like Photoshop.
XP was written to minimise it's RAM usage, no matter how much RAM it had- because back then RAM was still very expensive.
It's wasn't until Visa that Windows could make good use of more than 1GB of RAM (unless you were running an enterprise edition).


Grant
Darwin NT
ID: 1374561 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1374563 - Posted: 31 May 2013, 22:44:00 UTC - in response to Message 1374519.  

To me that 120W seems like a big power consumption for so small and RAC.

My q6600 with nvidia670 only uses 165W for an RAC that was ~17,000 used ~16hr/day.

And the current series of CPUs use even less power, for more work.
Grant
Darwin NT
ID: 1374563 · Report as offensive
Ianab
Volunteer tester

Send message
Joined: 11 Jun 08
Posts: 732
Credit: 20,635,586
RAC: 5
New Zealand
Message 1374568 - Posted: 31 May 2013, 23:02:16 UTC - in response to Message 1374561.  

What XP itself can run in, and what you want to actually use the machine are 2 different things.

My XP box is using about a gig of ram with S@H, virus scanner, a few Chrome Windows open etc. When I only had a one gig of RAM you really noticed the performance drop as the system used the page file. Printing a full colour A4 photo to an HP deskjet would bring it to it's knees. 2 gig of seem to be adequate for my needs.

Yes we used to sell XP machines with 256mb of ram back in the day, and they "worked", but over time the applications and background tasks have become bloatware. Yes you can stick a stripped down XP box in the corner with 256mb, and it will crunch just fine. But to actually use it for normal use is painful. Give the same box 1 or 2gb and it suddenly becomes a usable machine again.

Ian
ID: 1374568 · Report as offensive
Ianab
Volunteer tester

Send message
Joined: 11 Jun 08
Posts: 732
Credit: 20,635,586
RAC: 5
New Zealand
Message 1374569 - Posted: 31 May 2013, 23:07:01 UTC - in response to Message 1374519.  

To me that 120W seems like a big power consumption for so small and RAC.

My q6600 with nvidia670 only uses 165W for an RAC that was ~17,000 used ~16hr/day.


I'd hazard a guess that a Raspberry Pi would keep up with it, using about 2 watts? The power saved in 6 months would pay for the Pi !!!

Now I run some older and less efficient machines, but only in the cold weather. 100w is 100w, whether you put it though a heater or an old P4 server. Put a couple of them under your desk and you keep your feet warm on a frosty winter night, and crunch a few work units :-)

Ian
ID: 1374569 · Report as offensive
1 · 2 · 3 · 4 · Next

Message boards : Number crunching : v7 issues on very old cruncher


 
©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.