Windows S5R3 App 4.15 available for Beta Test

Bernd Machenschalk
Bernd Machenschalk
Joined: 15 Oct 04
Posts: 2,684
Credit: 25,950,161
RAC: 34,820

RE: A quick check on

Quote:

A quick check on "ghost" in a Cruden's Condensed Concordance suggests that most times the word ghost was used (I assume this is in the Authorized Version, aka King James) was in that phrase, including:

Gen 49:33 (yielded up the ghost)
Job 10:18, 11:20, 14:10
Jer 15:9
Mat 27:50 (yielded up the ghost)
Acts 5:10 (said of Sapphira, the spouse of Ananias)


Quite interesting. So in Mathew and Acts this is already cited from the Old Testament. Gen 49 refers to Jacob, there should be a similar term referring to Abraham (Gen 25:8). Mathew speaks of Jesus, Acts 5:5 talks about Ananias, and 5:10 about his wife.

Ok, we get a little bit distracted here, we should probably move this to an own thread in the Cafe.

BM

BM

Bernd Machenschalk
Bernd Machenschalk
Joined: 15 Oct 04
Posts: 2,684
Credit: 25,950,161
RAC: 34,820

RE: OK, I just want to make

Quote:
OK, I just want to make sure I understand what to do. I should delete app_info.xml from the einstein.phys.uwm.edu folder, and then I should restart BOINC manager. Right?


This is right in principle. However with the current project setting the app will not be accepted without a signature, i.e. the client will terminate the tasks in progress with a client error and download the "official" 4.15 App again (not knowing that it is identical).

The convenient way without wasting to much CPU time is to do this when a new task has just started.

The clean way would be to set the project to "no new work", finish all the tasks in the cache, stop the client, remove the file, start the client again and "allow more work" for the project.

BM

BM

Bernd Machenschalk
Bernd Machenschalk
Joined: 15 Oct 04
Posts: 2,684
Credit: 25,950,161
RAC: 34,820

RE: Interesting detail in

Quote:

Interesting detail in the stderr.txt:

2007-11-13 01:07:06.6875 [CRITICAL]: Checksum error: -262144
2007-11-13 01:07:06.6875 [CRITICAL]: Could not resume from checkpoint (-2)
2007-11-13 01:07:06.6875 [CRITICAL]: ERROR: MAIN() returned with error '10'
	zip warning: name not matched: h1_0520.20_S5R2__70_S5R3a_1_0

zip error: Nothing to do! (../../projects/einstein.phys.uwm.edu/h1_0520.20_S5R2__70_S5R3a_1_0)
2007-11-13 01:07:06.7500 [normal]: WARNING: Can't zip output file 'h1_0520.20_S5R2__70_S5R3a_1_0'

I only just noticed that the earlier one also crashed with that error.

Will try a reset, see what that does. Perhaps that my datafile is corrupt?
I'll also check the disk BOINC runs from.


It looks like a disk error. It's not a corrupt data file, it's actually the checkpoint file that's broken. The zip messages are misleading, they have nothing to do with the actual error.

BM

BM

Bernd Machenschalk
Bernd Machenschalk
Joined: 15 Oct 04
Posts: 2,684
Credit: 25,950,161
RAC: 34,820

Not much news. I have been

Not much news. I have been busy hunting Bugs that lead to more serious client errors, and for some reason I don't fully understand yet the SSE code that works fine in the MacOS Intel App give wrong results when compiled with a diferent compiler (e.g. gcc on Linux).

I'm still working on this, but the next two weeks will be very busy for basically the whole project team (for resons / deadlines unrelated to this project), I don't know how much progress will be made during these.

BM

BM

Bernd Machenschalk
Bernd Machenschalk
Joined: 15 Oct 04
Posts: 2,684
Credit: 25,950,161
RAC: 34,820

RE: I hope the next Win app

Quote:
I hope the next Win app will use the linear sin/cos code. :-)


I just found a portable way to make this code work correctly. The reason why the 4.24 that features this code isn't showing any speedup is that the compiler is messing up the "hot loop" with a variable on the ordinary stack then, as he thinks he has only four registers left in the FPU. I really like to solve this problem on the source-code level, i.e. without any manually inserted assembler code, at least for generic Apps that should run on all CPUs.

BM

BM

Bernd Machenschalk
Bernd Machenschalk
Joined: 15 Oct 04
Posts: 2,684
Credit: 25,950,161
RAC: 34,820

RE: How do you linearise

Quote:
How do you linearise sin & cos?
Given power series representations, wouldn't you have to go quadratic or higher?
.... unless you chopped up the domain to smaller intervals then linearly approximated within each them ( from some pre-calculated mid-points say ). You'd have to use more & smaller intervals to retain accuracy when in the extremes ( high rate of change of slope ie. 2nd derivative ). Or some other method - a ( huge ) lookup table for instance?


This is what we do. The nice thing about the linear approximation is that we can perform the float -> int conversion (argument to table index) with just bit-operations in integer registers, which is pretty fast.

Quote:
Sounds like you need something akin to a 'register' command but one which forces, rather than just requests/advises, the compiler to keep it's fingers off.


Precisely. Or, even better, something that assigns priorities to the "register" hints.

BM

BM

Bernd Machenschalk
Bernd Machenschalk
Joined: 15 Oct 04
Posts: 2,684
Credit: 25,950,161
RAC: 34,820

4.25 is available. The usual

4.25 is available. The usual suspects will probably want to dig through the assembler code ...

BM

BM

Comment viewing options

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