BETA 2.76

[Guru]

Fresh off the press GUI beta 2.76:

quote:
--------------------------------------------------------------------------------
The SETI@home beta client version 2.76 fixes the remaining
minor bugs that the beta testers have reported, as well as
tunes the final bits of scientific code.
NEW BUG FIXES:

There isn't much difference between 2.74 and 2.76, except for some
science tweaking (adjusting thresholds so the result files didn't
fill up with too much garbage) and some small bug fixes which were
mostly cosmetic. We did make one interim build (2.75) and found
that there was an actual bug in a Windows time library routine
that caused an invalid page fault when given a negative number.
This was not a visible problem for most users, but we added a
fix to make sure this routine was never called with negative values.

At any rate, we felt enough has changed to warrant one final beta.
If everything is working as expected, then this exact version
will be renamed 3.0 and released to the public.

By the way, a couple testers complained about "program file corrupted"
errors. We believe this is a fault on the server side. If you see this
happen, please send e-mail with information regarding what platform
you are using, and if you are using any third-party "add-on" software.


--------------------------------------------------------------------------------

Here is a link


------------------
Join the MURC SETI team! | SETI @ MURC

[Guru]

...."Lawrence Kirby chimed in with this:

>> 2.76 should be faster at gaussian searches.
>Why is that Lawrence?

An inefficiency in the gaussian search code has been eliminated.

>I found a healthy 17% speed increase for a 0.417 angle range, as reported on alt.sci.seti.

If it is down to the gaussian code then very large and small angle ranges should be unaffected...."

Still not as fast as 2.70 And Rat. I don't want to hear it!




------------------
Join the MURC SETI team! | SETI @ MURC

[Brian R.]

Oh No! You mentioned the 2.70 word...

Guru, just for curiosity, how does 2.70 compare with 2.76 as far as speed goes?

[This message has been edited by Brian R. (edited 21 September 2000).]

[Guru]

I have not tried 2.76 but from what I have heard 2.70 is abaut 15-20% faster!

------------------
Join the MURC SETI team! | SETI @ MURC

[Rattledagger]

The important thing for the vast majority of users is that it seems the times with beta-2.76 is the same as with cmd-2.04, and therefore faster than gui-2.04.

As always, since dual-machine not directly comparable, but with beta-2.76 I'm getting around 8 hours for units with gaussian, while cmd-2.04 running alongside uses 7 hours. Average for both, 7,391h/wu.

For comparison, I've run two instances of cmd-2.04 for some days, getting 7,747h/wu as average.
With the faster code for units without gaussian, beta-2.76 seems to be faster, but as always, no real answer before cmd-3 is released.

[SteveC]

seven thousand three hundred and ninety one hours per work unit?!??

You mean 7.391 h/wu surely?

------------------
Cheers,
Steve

"Life is what we make of it, yet most of us just fake"

[Sisyfos]

I do believe that, in all Scandinavian countries, comma is the decimal point and period is the number group delimitor.
At least it is in Denmark.

Kim

------------------
Join the MURC SETI team!
View the exciting stats