Bernd: Solution to the cross-platform issue?

Martin P.
Martin P.
Joined: 17 Feb 05
Posts: 162
Credit: 40156217
RAC: 0
Topic 192941

I currently lose appr. 20-25% of all completed work-units to the famous cross-platform validation issue. e.g. http://einsteinathome.org/workunit/34110776 and http://einsteinathome.org/workunit/34073730 .
Although these invalid work-units get deleted very quickly and therefore it is very complicated to monitor them I manually logged them for a week now and see appr. every 4th work-unit being INVALID although claimed credit is exactly the same as the others. Also my RAC went down from over 1,000 to 739 nowadays using the exact same machines and schedule as before. Seems to be mainly an issue between MacOS X and most flavours of Windows.

We heard for months that a new version for MacOS X is just about to be released but nothing happened so far. Please give us a timeline when we can expect a compatible version.

Ziran
Ziran
Joined: 26 Nov 04
Posts: 194
Credit: 356403
RAC: 1626

Bernd: Solution to the cross-platform issue?

6 Jul 2007

Quote:

Just to keep you updated of our plans, mainly regarding the cross-platform differences:
- Early next week (probably Monday) we'll issue a new validator that should make things easier for transition and probably fix some invalid results by itself
- After the new validator is in place, we'll issue a new set of Apps for public Beta Test (for all platforms) that incorporate the fixes accomplished so far. I'll keep on tracking problems and fixing bugs I find until the very last moment. The new Apps will also incorporate a new feature that we might need.
- If it turns out that we need this feature (using pre-calculated files instead of doing the calculations in the Apps to avoid platform differences there), we will issue new workunits (actually a new workunit generator) that will make use of this feature after the new Apps have been made "official".
- Once we got the validation working properly, I'll work on speeding up the computation in the Apps. The current code I plan to use for parts of the calculation btw. doesn't make use of neither modf() nor ftol() anymore but actually uses bit-operations to achieve something similar.

BM

http://einsteinathome.org/node/192905&nowrap=true#71592

Then you're really interested in a subject, there is no way to avoid it. You have to read the Manual.

Martin P.
Martin P.
Joined: 17 Feb 05
Posts: 162
Credit: 40156217
RAC: 0

RE: 6 Jul 2007RE: Just to

Message 69555 in response to message 69554

Quote:
6 Jul 2007
Quote:

Just to keep you updated of our plans, mainly regarding the cross-platform differences:
- Early next week (probably Monday) we'll issue a new validator that should make things easier for transition and probably fix some invalid results by itself
- After the new validator is in place, we'll issue a new set of Apps for public Beta Test (for all platforms) that incorporate the fixes accomplished so far. I'll keep on tracking problems and fixing bugs I find until the very last moment. The new Apps will also incorporate a new feature that we might need.
- If it turns out that we need this feature (using pre-calculated files instead of doing the calculations in the Apps to avoid platform differences there), we will issue new workunits (actually a new workunit generator) that will make use of this feature after the new Apps have been made "official".
- Once we got the validation working properly, I'll work on speeding up the computation in the Apps. The current code I plan to use for parts of the calculation btw. doesn't make use of neither modf() nor ftol() anymore but actually uses bit-operations to achieve something similar.

BM

http://einsteinathome.org/node/192905&nowrap=true#71592

Thanks Ziran,

the title of that thread "Windows Beta Test App 4.24 available" did not seem to be appropriate with my MacOS X problem.

Martin P.
Martin P.
Joined: 17 Feb 05
Posts: 162
Credit: 40156217
RAC: 0

Until they sort out the

Until they sort out the problem and an optimized version for PPC is released I'll reduce the share for E@H from 60% to 10%. Let's wait and see...

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686138627
RAC: 563921

RE: Until they sort out the

Message 69557 in response to message 69556

Quote:
Until they sort out the problem and an optimized version for PPC is released I'll reduce the share for E@H from 60% to 10%. Let's wait and see...

The cross-validation problems are possibly fixed this very week.

I guess we are both crunching for CPDN? Do they have at least plans for platform-optimized apps? I haven't followed discussions at CPDN for a long time.

CU

BRM

Martin P.
Martin P.
Joined: 17 Feb 05
Posts: 162
Credit: 40156217
RAC: 0

RE: RE: Until they sort

Message 69558 in response to message 69557

Quote:
Quote:
Until they sort out the problem and an optimized version for PPC is released I'll reduce the share for E@H from 60% to 10%. Let's wait and see...

The cross-validation problems are possibly fixed this very week.

I guess we are both crunching for CPDN? Do they have at least plans for platform-optimized apps? I haven't followed discussions at CPDN for a long time.

CU

BRM

There is no CPDN-client for MacOS X anymore, they quit support. However, there is an excellent, optimized client for SETI@Home!

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686138627
RAC: 563921

RE: RE: RE: Until they

Message 69559 in response to message 69558

Quote:
Quote:
Quote:
Until they sort out the problem and an optimized version for PPC is released I'll reduce the share for E@H from 60% to 10%. Let's wait and see...

The cross-validation problems are possibly fixed this very week.

I guess we are both crunching for CPDN? Do they have at least plans for platform-optimized apps? I haven't followed discussions at CPDN for a long time.

CU

BRM

There is no CPDN-client for MacOS X anymore, they quit support. However, there is an excellent, optimized client for SETI@Home!

There is one for Intel MacOS X (I'm using that one), so I guess they abandoned the Power PC platform?

The SETI@Home optimized clients are cool, but I left S@H because I fail to see any science being done with the tons of data they collect.

Upps, getting off-topic, I guess. Sorry

CU

BRM

Luis
Luis
Joined: 11 Feb 06
Posts: 15
Credit: 1968357
RAC: 0

I just remember somebody else

I just remember somebody else that had cross platform issues with linux before:extended precision Turns out in Linux x86 the math unit is always set to extended precision mode (80 bits) while most other platforms use double precision (64 bits). That is mostly irrelevant unless you have code like 'if (10*0.3==3)' in which case minute differences in the least significant digits show up as different integer results.

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

I found this tutorial useful

I found this tutorial useful for understanding floating point arithmetics:
CERN Computing Seminar
Tullio
Go to Archives, January 2005, Computing with floating point: it's not dark magic, it's science

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686138627
RAC: 563921

Interesting stuff indeed,

Interesting stuff indeed, thanks for posting the links.

BTW, just saw this in the news section of the E@H homepage:

Quote:

July 10, 2007
We have started a new validator. This change is one of the steps needed to get rid of the cross-platform validation problems.

CU

BRM

Martin P.
Martin P.
Joined: 17 Feb 05
Posts: 162
Credit: 40156217
RAC: 0

RE: Interesting stuff

Message 69563 in response to message 69562

Quote:

Interesting stuff indeed, thanks for posting the links.

BTW, just saw this in the news section of the E@H homepage:

Quote:

July 10, 2007
We have started a new validator. This change is one of the steps needed to get rid of the cross-platform validation problems.

CU

BRM

Does it - finally - work?

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.