Re: [isabelle-dev] Towards the release

2016-01-15 Thread Makarius
Current Isabelle/9ca00b65d36c and AFP/2c322507b8a6 appears to be a fairly stable state, so I will take this as starting point for Isabelle2016-RC1 later today. The main isabelle-dev remains open a few more days for fine-tuning and consolidation. Makarius

[isabelle-dev] Multicore timings

2016-01-15 Thread Makarius
Just for the record here are some timings on dual Xeon E5-2620v3 2.40GHz with 12 threads: Isabelle/9ca00b65d36c + AFP/2c322507b8a6 ML_OPTIONS="-H 2000 --gcthreads 12" isabelle build -o threads=12 Finished HOL-Bali (0:00:57 elapsed time, 0:05:26 cpu time, factor 5.71) Finished HOL-MicroJava

Re: [isabelle-dev] Multicore timings

2016-01-15 Thread Gerwin Klein
Nice! I distinctly remember the times when MicroJava was a very long session ;-) Cheers, Gerwin > On 15 Jan 2016, at 11:16 AM, Makarius wrote: > > Just for the record here are some timings on dual Xeon E5-2620v3 2.40GHz with > 12 threads: > > Isabelle/9ca00b65d36c +

Re: [isabelle-dev] Multicore timings

2016-01-15 Thread Lawrence Paulson
I do remember. It’s almost a shame how years of work can be executed in minutes. Larry > On 15 Jan 2016, at 19:16, Makarius wrote: > > The sessions from Bali to JinjaThreads are historically related; this used to > be the top-end in resource usage until 1-2 years ago.

Re: [isabelle-dev] Multicore timings

2016-01-15 Thread Makarius
On Fri, 15 Jan 2016, Lawrence Paulson wrote: It’s almost a shame how years of work can be executed in minutes. Lets say that's the struggle for survival of maintainers behind the scenes :-) Both Intel/AMD contributed a little, but David Matthews also did a great job. For the release, I