KulMari / Parsek2D_MLMD

2 stars 2 forks source link

Weak scaling plots #10

Open KulMari opened 11 years ago

KulMari commented 11 years ago

plots 001

valsusa commented 11 years ago

Hi Indeed, the work of the colleagues at CS showed that our cg solver is very problematic, possibly buggy. Their libraries performed orders of magnitude better. Talk to Jan or directly to Kumar in CS to know more. I am not using Poisson anymore, and live a happy life. Cheers Gianni

Sent from my iPad

On 01/set/2013, at 04:25, KulMari notifications@github.com wrote:

— Reply to this email directly or view it on GitHub.

KulMari commented 11 years ago

Who's Kumar?

On Sep 1, 2013, at 3:32 PM, valsusa wrote:

Hi Indeed, the work of the colleagues at CS showed that our cg solver is very problematic, possibly buggy. Their libraries performed orders of magnitude better. Talk to Jan or directly to Kumar in CS to know more. I am not using Poisson anymore, and live a happy life. Cheers Gianni

Sent from my iPad

On 01/set/2013, at 04:25, KulMari notifications@github.com wrote:

— Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub.

valsusa commented 11 years ago

Ciao magari lo conosci con il primo nome: Pawan? Un postdoc di CS. Ciao Gianni

On Sep 1, 2013, at 9:50 AM, KulMari wrote:

Who's Kumar?

On Sep 1, 2013, at 3:32 PM, valsusa wrote:

Hi Indeed, the work of the colleagues at CS showed that our cg solver is very problematic, possibly buggy. Their libraries performed orders of magnitude better. Talk to Jan or directly to Kumar in CS to know more. I am not using Poisson anymore, and live a happy life. Cheers Gianni

Sent from my iPad

On 01/set/2013, at 04:25, KulMari notifications@github.com wrote:

— Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

KulMari commented 11 years ago

Lo conosco di nome. Lo contatto per chiedergli se ha bisogno di un caso di test a convergenza difficile (la griglia raffinata spesso addirittura non converge nei test a box grossa). Comunque ho smesso anche io di usare Poisson e per ora non vedo problemi, pero' non ho ancora simulazioni molto lunghe

On Sep 1, 2013, at 5:58 PM, valsusa wrote:

Ciao magari lo conosci con il primo nome: Pawan? Un postdoc di CS. Ciao Gianni

On Sep 1, 2013, at 9:50 AM, KulMari wrote:

Who's Kumar?

On Sep 1, 2013, at 3:32 PM, valsusa wrote:

Hi Indeed, the work of the colleagues at CS showed that our cg solver is very problematic, possibly buggy. Their libraries performed orders of magnitude better. Talk to Jan or directly to Kumar in CS to know more. I am not using Poisson anymore, and live a happy life. Cheers Gianni

Sent from my iPad

On 01/set/2013, at 04:25, KulMari notifications@github.com wrote:

— Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

KulMari commented 10 years ago

plots 007

v0= same results as before v1= commit f0500e33861f716ede5924ea7b361fc73d29ddf9

same inputfile, but smoothing (Smooth=0, Nvolte=5) and tol= -3, Nvolte =3 instead of no smoothing, tol= -7, Nvolte= 7. This is now possible because of fixed smoothing.

Much faster now, but it still does not scale