AVWhitney parallel calculation large mesh crash

Clearly defined bug reports and their fixes
Post Reply
Takala
Posts: 186
Joined: 23 Aug 2009, 23:59

AVWhitney parallel calculation large mesh crash

Post by Takala »

Hello,

I run AV Whitney with 1.5 million tetra elements. I use a virtual machine. I get an error:
MAIN:
MAIN: -------------------------------------
MAIN: Time: 1/10 1.00000000000000002E-002
MAIN: -------------------------------------
MAIN:
[elmerbox:02069] *** Process received signal ***
[elmerbox:02069] Signal: Segmentation fault (11)
[elmerbox:02069] Signal code: Invalid permissions (2)
[elmerbox:02069] Failing at address: 0x7fff012c0430
[elmerbox:02069] [ 0] /lib/x86_64-linux-gnu/libc.so.6(+0x364c0) [0x7ff32f8d84c0]
[elmerbox:02069] [ 1] [0x7fff012c0430]
[elmerbox:02069] *** End of error message ***
[elmerbox:02067] *** Process received signal ***
[elmerbox:02067] Signal: Segmentation fault (11)
[elmerbox:02067] Signal code: Invalid permissions (2)
[elmerbox:02067] Failing at address: 0x7fffa283be80
[elmerbox:02067] [ 0] /lib/x86_64-linux-gnu/libc.so.6(+0x364c0) [0x7fda47a214c0]
[elmerbox:02067] [ 1] [0x7fffa283be80]
[elmerbox:02067] *** End of error message ***
What do you think gives a problem here?

Cheers,

Eelis
Takala
Posts: 186
Joined: 23 Aug 2009, 23:59

Re: AVWhitney parallel calculation large mesh crash

Post by Takala »

I found it. Mi own mistake (How shocking, I never make mistakes. :D just kidding)

-E
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: AVWhitney parallel calculation large mesh crash

Post by raback »

Hi Eelis

Pretty big case you have hear for this kind of equation. Could you comment a little bit how the number of iterations grow with the problem size. I guess you are using diagonal preconditioner and some Krylov method.

-Peter

PS. Happy holiday season. I expect it will be quite little traffic here and possible questions may ave to wait to get an answer.
Takala
Posts: 186
Joined: 23 Aug 2009, 23:59

Re: AVWhitney parallel calculation large mesh crash

Post by Takala »

raback wrote:Hi Eelis

Psetty big case you have hear for this kind of equation. Could you comment a little bit how the number of iterations grow with theproblem size. I guess you are using diagonal preconditioner and some Krylov method.

-Peter

PS. Happy holiday season. I expect it will be quite little traffic here and possible questions may ave to wait to get an answer.
Hi,

yes it is a big case (linear luckily) and you are right I use BiCGStab but without preconditioning. I didn't see any major increase in iteration count when increasing from 150k to 1.5M elements. Seems to be very robust solver.

Cheers,

Eelis

ps. Speaking of holiday season: I caught an influenza virus presumably on the 22nd of December and had a complication meaning that I have been sick from 24th of December until now... Luckily I'm feeling better now that I start working tomorrow :lol: Life is great sometimes!
Post Reply