Re: [Python-Dev] configure produces a non-working Makefile in some scenarios, due to ASDLGEN

2012-10-29 Thread Vinay Sajip
Ned Deily  acm.org> writes:

> Makefile.  You now have the opportunity to override the behavior at 
> ./configure time by using the PYTHON variable to specify the path to the 
> Python executable you want to use for the asdl_c.py step, as in:
> 
>./configure ... PYTHON=/usr/bin/python2.7

Aha! Thanks for pointing that out.

Regards,

Vinay Sajip

___
Python-Dev mailing list
[email protected]
http://mail.python.org/mailman/listinfo/python-dev
Unsubscribe: 
http://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com


Re: [Python-Dev] configure produces a non-working Makefile in some scenarios, due to ASDLGEN

2012-10-29 Thread Vinay Sajip
Cameron Simpson  zip.com.au> writes:

> Hoping this is more helpful than I've probably made it sound,

Sure. I know my setup was somewhat unusual, which is why I posted here rather
than the tracker, but it was a surprise when a long-working configuration
suddenly broke because of an unbuilt Python 3.4. Ned's post showed me why.

Regards,

Vinay Sajip


___
Python-Dev mailing list
[email protected]
http://mail.python.org/mailman/listinfo/python-dev
Unsubscribe: 
http://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com


Re: [Python-Dev] Python 3.3 vs. Python 2.7 benchmark results (again, but this time more solid numbers)

2012-10-29 Thread Brett Cannon
On Sun, Oct 28, 2012 at 3:48 PM, Tim Delaney wrote:

> On 28 October 2012 18:22, Stefan Behnel  wrote:
>
>> How much of an
>>
>>> effect would it have on startup times and these benchmarks if
>>> Cython-compiled extensions were used?
>>>
>>
>> Depends on what and how much code you use. If you compile everything into
>> one big module that "imports" all of the stdlib when it gets loaded, you'd
>> likely loose a lot of time because it would take a while to initialise all
>> that useless code on startup. If you keep it separate, it would likely be a
>> lot faster because you avoid the interpreter for most of the module startup.
>>
>
> I was specifically thinking in terms of the tests Brett ran (that was the
> full set on speed.python.org, wasn't it?),
>

It's not the full set as not all of them can be run on Python 3, but it is
as many as can be run.

 -Brett
___
Python-Dev mailing list
[email protected]
http://mail.python.org/mailman/listinfo/python-dev
Unsubscribe: 
http://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com


Re: [Python-Dev] Python 3.3 vs. Python 2.7 benchmark results (again, but this time more solid numbers)

2012-10-29 Thread Brett Cannon
To see if the bad iterative_count and threaded_count results were
consistently bad, I ran the benchmark suite on my MacBook Pro to see how
"reliable" the benchmarks were. The output is below.

Basically 6 benchmarks (regex_effbot, queens, startup_nosite,
iterative_count, threaded_count, and telco) had a variance of more than 15%
performance between my 2 computers, although queens, iterative_count, and
threaded_count were the only ones that swung between neutral/good to bad
depending on the machine (the rest either want from bad to very bad, or
very good to more very good).

And before Antoine asks, I added a ``sys.modules['markupsafe'] = None` line
to the mako_v2 benchmark locally. =) Still need to either explicitly block
it or emit a warning in the code in the repo.


#


Report on Darwin Darwin Kernel Version 12.2.0: Sat Aug 25 00:48:52 PDT
2012; root:xnu-2050.18.24~1/RELEASE_X86_64 x86_64 i386
Total CPU cores: 8

### 2to3 ###
10.321463 -> 9.525119: 1.08x faster

### call_method ###
Min: 0.466812 -> 0.417812: 1.12x faster
Avg: 0.483324 -> 0.427158: 1.13x faster
Significant (t=28.77)
Stddev: 0.01876 -> 0.01483: 1.2644x smaller
Timeline: b'http://tinyurl.com/8al5lmm'

### call_method_slots ###
Min: 0.484923 -> 0.409452: 1.18x faster
Avg: 0.487877 -> 0.413054: 1.18x faster
Significant (t=131.11)
Stddev: 0.00395 -> 0.00577: 1.4589x larger
Timeline: b'http://tinyurl.com/9zhpg6z'

### call_method_unknown ###
Min: 0.547050 -> 0.406866: 1.34x faster
Avg: 0.550721 -> 0.409359: 1.35x faster
Significant (t=328.32)
Stddev: 0.00415 -> 0.00325: 1.2795x smaller
Timeline: b'http://tinyurl.com/9wxoddz'

### call_simple ###
Min: 0.391213 -> 0.332055: 1.18x faster
Avg: 0.393563 -> 0.335362: 1.17x faster
Significant (t=127.15)
Stddev: 0.00363 -> 0.00427: 1.1764x larger
Timeline: b'http://tinyurl.com/8mmepzw'

### chameleon ###
Min: 0.078505 -> 0.070175: 1.12x faster
Avg: 0.083754 -> 0.071500: 1.17x faster
Significant (t=2.95)
Stddev: 0.05086 -> 0.00119: 42.8425x smaller
Timeline: b'http://tinyurl.com/8bz9hpl'

### chaos ###
Min: 0.353739 -> 0.423587: 1.20x slower
Avg: 0.356297 -> 0.428197: 1.20x slower
Significant (t=-108.44)
Stddev: 0.00200 -> 0.00424: 2.1147x larger
Timeline: b'http://tinyurl.com/98e56le'

### django ###
Min: 0.824149 -> 0.862750: 1.05x slower
Avg: 0.831614 -> 0.869112: 1.05x slower
Significant (t=-21.47)
Stddev: 0.01020 -> 0.00697: 1.4634x smaller
Timeline: b'http://tinyurl.com/8kz8owv'

### fannkuch ###
Min: 1.776913 -> 1.832973: 1.03x slower
Avg: 1.793116 -> 1.915348: 1.07x slower
Significant (t=-11.57)
Stddev: 0.01436 -> 0.07329: 5.1030x larger
Timeline: b'http://tinyurl.com/9ptae4z'

### fastpickle ###
Min: 0.810968 -> 0.739322: 1.10x faster
Avg: 0.818099 -> 0.745148: 1.10x faster
Significant (t=58.02)
Stddev: 0.00577 -> 0.00677: 1.1731x larger
Timeline: b'http://tinyurl.com/8l769dd'

### fastunpickle ###
Min: 0.644198 -> 0.659345: 1.02x slower
Avg: 0.647976 -> 0.666154: 1.03x slower
Significant (t=-18.96)
Stddev: 0.00343 -> 0.00584: 1.7020x larger
Timeline: b'http://tinyurl.com/93xn7el'

### float ###
Min: 0.420888 -> 0.363410: 1.16x faster
Avg: 0.432285 -> 0.376179: 1.15x faster
Significant (t=38.14)
Stddev: 0.00762 -> 0.00708: 1.0766x smaller
Timeline: b'http://tinyurl.com/8bjwka9'

### formatted_logging ###
Min: 0.325707 -> 0.413196: 1.27x slower
Avg: 0.329846 -> 0.418099: 1.27x slower
Significant (t=-119.89)
Stddev: 0.00397 -> 0.00337: 1.1787x smaller
Timeline: b'http://tinyurl.com/8ktbs49'

### genshi ###
Min: 0.254604 -> 0.269696: 1.06x slower
Avg: 0.258585 -> 0.275615: 1.07x slower
Significant (t=-33.39)
Stddev: 0.00283 -> 0.00557: 1.9704x larger
Timeline: b'http://tinyurl.com/8bqvcwl'

### go ###
Min: 0.676453 -> 0.745504: 1.10x slower
Avg: 0.681833 -> 0.752170: 1.10x slower
Significant (t=-48.67)
Stddev: 0.00520 -> 0.00880: 1.6917x larger
Timeline: b'http://tinyurl.com/9d6qj3y'

### hexiom2 ###
Min: 186.378727 -> 172.939507: 1.08x faster
Avg: 186.679821 -> 173.103242: 1.08x faster
Significant (t=39.61)
Stddev: 0.42581 -> 0.23156: 1.8389x smaller
Timeline: b'http://tinyurl.com/9mc3pmg'

### html5lib ###
Min: 11.827770 -> 11.239556: 1.05x faster
Avg: 11.858253 -> 11.370960: 1.04x faster
Significant (t=6.93)
Stddev: 0.02825 -> 0.15466: 5.4746x larger
Timeline: b'http://tinyurl.com/8vl952y'

### iterative_count ###
Min: 0.168182 -> 0.154105: 1.09x faster
Avg: 0.169512 -> 0.155952: 1.09x faster
Significant (t=50.77)
Stddev: 0.00139 -> 0.00128: 1.0899x smaller
Timeline: b'http://tinyurl.com/9eymjtf'

### json_dump_v2 ###
Min: 3.350528 -> 3.795307: 1.13x slower
Avg: 3.369661 -> 3.825400: 1.14x slower
Significant (t=-125.93)
Stddev: 0.01470 -> 0.02095: 1.4250x larger
Timeline: b'http://tinyurl.com/8wyn9qa'

### json_load ###
Min: 0.999717 -> 0.607549: 1.65x faster
Avg: 1.007319 -> 0.613016: 1.64x faster
Significant (t=289.24)
Stddev: 0.00673 -> 0.00690: 1.0240x larger
Timeline: b'http://tinyurl.com/8qxakdw'

### mako_v2 ###
Min: 0.094817 -> 0.279593: 2.95x sl

Re: [Python-Dev] Python 3.3 vs. Python 2.7 benchmark results (again, but this time more solid numbers)

2012-10-29 Thread Antoine Pitrou
On Mon, 29 Oct 2012 09:56:57 -0400
Brett Cannon  wrote:

> To see if the bad iterative_count and threaded_count results were
> consistently bad, I ran the benchmark suite on my MacBook Pro to see how
> "reliable" the benchmarks were. The output is below.
> 
> Basically 6 benchmarks (regex_effbot, queens, startup_nosite,
> iterative_count, threaded_count, and telco) had a variance of more than 15%
> performance between my 2 computers, although queens, iterative_count, and
> threaded_count were the only ones that swung between neutral/good to bad
> depending on the machine (the rest either want from bad to very bad, or
> very good to more very good).

This is using different compilers on the 2 computers, right?

Regards

Antoine.


___
Python-Dev mailing list
[email protected]
http://mail.python.org/mailman/listinfo/python-dev
Unsubscribe: 
http://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com


Re: [Python-Dev] Python 3.3 vs. Python 2.7 benchmark results (again, but this time more solid numbers)

2012-10-29 Thread Brett Cannon
On Mon, Oct 29, 2012 at 3:22 PM, Antoine Pitrou  wrote:

> On Mon, 29 Oct 2012 09:56:57 -0400
> Brett Cannon  wrote:
>
> > To see if the bad iterative_count and threaded_count results were
> > consistently bad, I ran the benchmark suite on my MacBook Pro to see how
> > "reliable" the benchmarks were. The output is below.
> >
> > Basically 6 benchmarks (regex_effbot, queens, startup_nosite,
> > iterative_count, threaded_count, and telco) had a variance of more than
> 15%
> > performance between my 2 computers, although queens, iterative_count, and
> > threaded_count were the only ones that swung between neutral/good to bad
> > depending on the machine (the rest either want from bad to very bad, or
> > very good to more very good).
>
> This is using different compilers on the 2 computers, right?


Yes: gcc 4.6.3 on Linux and Clang 3.1 on OS X.
___
Python-Dev mailing list
[email protected]
http://mail.python.org/mailman/listinfo/python-dev
Unsubscribe: 
http://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com


[Python-Dev] Python Bug Day this Saturday announced

2012-10-29 Thread Éric Araujo
Hi everybody,

I just sent the announcement for the bug day to python-list (apparently
pending approval), core-mentorship and montrealpython.  Core developers
who plan on being on IRC can add themselves to the list on
http://wiki.python.org/moin/PythonBugDay so that people can connect
nicknames with people.

The list by Petri at http://piratepad.net/pyconfi-sprint-issues can
still be updated.  Otherwise we’ll fall back to the usual roundup query
for easy bugs.

Cheers!
___
Python-Dev mailing list
[email protected]
http://mail.python.org/mailman/listinfo/python-dev
Unsubscribe: 
http://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com