Re: [lno] 50% runtime performance regression since yesterday

[Sebastian Pop <sebastian dot pop at cri dot ensmp dot fr>]

On Wed, Jun 23, 2004 at 11:22:48PM +0000, Sebastian Pop wrote:
> Maybe I have to try on ia64... 

I have reproduced the regression on ia64. 

i = 1    t = 0.00407921  dt = 0.00407921 (0.213325s/it)
i = 2    t = 0.00679089  dt = 0.00271169 (4.84207s/it)
i = 3    t = 0.00937923  dt = 0.00258833 (4.84805s/it)
i = 4    t = 0.0120634   dt = 0.0026842 (4.84939s/it)
i = 5    t = 0.0150383   dt = 0.00297491 (4.85705s/it)
i = 6    t = 0.0183952   dt = 0.00335687 (4.84058s/it)
i = 7    t = 0.0217122   dt = 0.00331702 (4.84577s/it)
i = 8    t = 0.0250681   dt = 0.00335591 (4.85288s/it)
i = 9    t = 0.0284961   dt = 0.00342798 (4.846s/it)
i = 10   t = 0.032036    dt = 0.00353984 (4.85405s/it)

vs. before:

i = 1    t = 0.00407921  dt = 0.00407921 (0.166039s/it)
i = 2    t = 0.00679089  dt = 0.00271169 (2.78032s/it)
i = 3    t = 0.00937923  dt = 0.00258833 (2.79729s/it)
i = 4    t = 0.0120634   dt = 0.0026842 (2.78145s/it)
i = 5    t = 0.0150383   dt = 0.00297491 (2.79876s/it)
i = 6    t = 0.0183952   dt = 0.00335687 (2.7805s/it)
i = 7    t = 0.0217122   dt = 0.00331702 (2.79288s/it)
i = 8    t = 0.0250681   dt = 0.00335591 (2.77603s/it)
i = 9    t = 0.0284961   dt = 0.00342798 (2.79529s/it)
i = 10   t = 0.032036    dt = 0.00353984 (2.78375s/it)

Sebastian