Wacky SuperPi run

[largon]

Decided to run some SuperPi 1M (1.5 XS version). Result was really nothing spectacular for a Opty146:

This (http://www.xtremesystems.org/forums/attachment.php?attachmentid=48142&stc=1&d=1149952120) would have been a personal record SPi 1M, but something went awry. The run had an extra loop (20th) - which naturally ruined the score. A normal run takes 19 loops and the Pi value output stage.
:confused: :shrug: :wth:
Any attempts to duplicate the issue were unsuccessful.
Anyone seen something like this?


And btw. the result is valid (http://www.xtremesystems.com/pi/).
c/p:
27.422s
FC35DE00 (last 2 letters are zeros)

[^don.k's^]

That has happened to me a bunch of times, it was more likely when OC wasnt stable enough, i dont know if thats the real cause....

[corvus_corax]

Hi! that has happened to me too, i think that happen when your really are in the thin line between stability and instability :p:
Check s7e9h3n's thread - Something's Wrong Here..... (http://www.xtremesystems.org/forums/showthread.php?t=83366)

[largon]

Strange...
It seems this is due to instability "of a sort". I'm surprised that SuperPi doesn't actually error out as a result of this glitch.

IMO, this makes SuperPi less reliable. :(

[wwwww]

You think that's wacky?

Explain how this happens:

http://users.tpg.com.au/weigner/wtf.PNG

[s7e9h3n]

Strange...
It seems this is due to instability "of a sort". I'm surprised that SuperPi doesn't actually error out as a result of this glitch.

IMO, this makes SuperPi less reliable. :(
No, it doesn't since you can't duplicate it @ will ;) I've seen the same problem as well:

http://img377.imageshack.us/img377/9008/holychit2mb.jpg

I'll even toss in a better one :D :

http://img97.imageshack.us/img97/1010/13x2758mwtf7py.jpg

[largon]

wwwww,
Now, that truly is pretty disturbed run!
And it's valid. :eek:
I wonder what caused it. At least it wasn't instability...

s7e9h3n,
You wouldn't happen to have a shot of a "succesful" 1M run at the exact same settings as on that 1M with 20 loops? Did the 20th loop increase the total computation time?

Something horrible must have happened between the 1st and 2nd loop during that 27m 8M run... The 1st is fine, 2nd loop took twice the normal time.

[corvus_corax]

The second one it seems like when you run SPI in a RAM Drive but you don't have memory space enough , i mean, for example, to run SPI 8M in 8MB RAM Drive or even a too tight maxmem :p: but i think that in that shot s7e9h3n was not using maxmem.

But it's just a theory :D

[s7e9h3n]

s7e9h3n,
You wouldn't happen to have a shot of a "succesful" 1M run at the exact same settings as on that 1M with 20 loops? Did the 20th loop increase the total computation time?

Something horrible must have happened between the 1st and 2nd loop during that 27m 8M run... The 1st is fine, 2nd loop took twice the normal time.
Of course the 20th loop increasese the total time ;) What I find intriguing about it is the fact that the 20th iteration scales accordingly to all the other prior iterations. I would have expected at least an unusual time between iterations. I tried to locate a screen of that cpu running the exact same timings, but I wasn't able to find one. This is the closest I could find:

http://img118.imageshack.us/img118/5670/14x2652za.jpg

That second run I just chalk up to a fluke.....I got the cpu eventually to do this ;)

http://cryo-laboratory.com/upload/userfiles/s7e9h3n/Opty%20154/14x2708m.jpg


@ Corvus: Yep, I never use maxmem - Probably because the ONE time I did try - I totally screwed up my OS :p:

[couppi]

It would be nice if it would drop a few loops sometimes...

OT:
Wwwww, what theme is that? It looks nice and clean.

[wwwww]

It would be nice if it would drop a few loops sometimes...

OT:
Wwwww, what theme is that? It looks nice and clean.

Don't remember, from about a year ago...Rhodium or something like that - came with the original styleXP.