tag:blogger.com,1999:blog-3971202189709462152.post6048076549081013253..comments2015-03-26T13:18:37.960+01:00Comments on PyPy Status Blog: NumPy on PyPy progress reportCarl Friedrich Bolzhttp://www.blogger.com/profile/00518922641059511014noreply@blogger.comBlogger19125tag:blogger.com,1999:blog-3971202189709462152.post-88639300186858650342012-08-06T17:53:12.643+02:002012-08-06T17:53:12.643+02:00I think the numpy.linalg module is pretty importan...I think the numpy.linalg module is pretty important.<br />How to move efforts into this?Raul Durandhttp://www.blogger.com/profile/03417426673915503070noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-12163895949995119942012-08-06T17:52:22.556+02:002012-08-06T17:52:22.556+02:00I think the numpy.linalg module is pretty importan...I think the numpy.linalg module is pretty important.<br />How to move efforts into this?Raul Durandhttp://www.blogger.com/profile/03417426673915503070noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-17554315877719920122012-07-30T22:02:10.420+02:002012-07-30T22:02:10.420+02:00Hello,
This is a very promising result, thank you...Hello,<br /><br />This is a very promising result, thank you for sharing it.<br />Could you give a few more details about the differences wrt to numpy?<br /><br />What would people have to do to use numpypy with scipy?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-81333623362552979552012-04-22T21:07:55.418+02:002012-04-22T21:07:55.418+02:00What about pickling? I'd love to experiment wi...What about pickling? I'd love to experiment with hybrid CPython/PyPy execution using some magic from the multiprocessing module or a similar parallel computation framework.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-28552337547119678792012-04-20T16:26:56.094+02:002012-04-20T16:26:56.094+02:00Nice...but what is the next step?
Numpy alone is n...Nice...but what is the next step?<br />Numpy alone is not that useful.<br /><br />"We" need at least scipy and matplotlib.<br /><br />Are you going to port all these modules? I don't think so.<br /><br />One way forward could be to have numpy in pypy and at least scipy and matplotlib working with the pypy C api at a decent speed.<br /> <br />What do you think?Unknownhttp://www.blogger.com/profile/00591993364491869287noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-53282756919300009072012-04-19T18:19:11.406+02:002012-04-19T18:19:11.406+02:00Any chance comparing with Fortran? There are assum...Any chance comparing with Fortran? There are assumptions about pointers and alignment that Fortran compiler can make.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-74467518403801144132012-04-19T14:47:11.885+02:002012-04-19T14:47:11.885+02:00This sounds really cool!
And it would be awesome ...This sounds really cool!<br /><br />And it would be awesome if you’d manage to coordinate with numpy, so the projects merge to a single python codebase with two separate backends: One C-Extension based for CPython and one Pure-Python based for pypy.ArneBabhttp://www.blogger.com/profile/16449390422848764481noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-18058337785655439882012-04-19T11:17:38.373+02:002012-04-19T11:17:38.373+02:00Support for objects is actually quite useful: plea...Support for objects is actually quite useful: please reconsider adding it.<br /><br />Here is a very useful case: the manipulation of arrays of numbers with uncertainties (special uncertainties.UFloat objects). Numbers with uncertainties behave very much like regular numbers: it is very useful to be able to use the regular NumPy syntax for array operations, for calculating matrix inverses when the matrices contain number with uncertainties, etc. I know many people use these features.<br /><br />It would be *great* (read: irreplaceable :) to have support for the object NumPy dtype.EOL (Eric O LEBIGOT)http://www.blogger.com/profile/08843567382834026704noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-65393581392830393362012-04-19T11:16:14.482+02:002012-04-19T11:16:14.482+02:00This comment has been removed by the author.EOL (Eric O LEBIGOT)http://www.blogger.com/profile/08843567382834026704noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-426816871316519122012-04-19T11:15:33.439+02:002012-04-19T11:15:33.439+02:00This comment has been removed by the author.EOL (Eric O LEBIGOT)http://www.blogger.com/profile/08843567382834026704noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-67321073777085237512012-04-18T23:20:26.606+02:002012-04-18T23:20:26.606+02:00Ralf, we don't have to implement the pure-pyth...Ralf, we don't have to implement the pure-python stuff, so much as we need to make sure the features of NumPy's core that they depend on are implemented.Alexhttp://www.blogger.com/profile/14054821112394577330noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-42632503311556527792012-04-18T21:45:15.329+02:002012-04-18T21:45:15.329+02:00Hi, Numpy masked arrays, matrices and the testing ...Hi, Numpy masked arrays, matrices and the testing framework are pure Python, so why do you need to implement them?Ralf Gommershttp://www.blogger.com/profile/10189904547449375563noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-55237139966528753612012-04-18T11:07:28.564+02:002012-04-18T11:07:28.564+02:00"busy doing other things (pictured)". P..."busy doing other things (pictured)". Pictured where? :-)Armin Rigohttp://www.blogger.com/profile/06300515270104686574noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-5585659618709439472012-04-17T23:55:04.987+02:002012-04-17T23:55:04.987+02:00Very cool!Very cool!xhttp://www.blogger.com/profile/06529725872860738749noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-81440340952673017432012-04-17T22:17:15.510+02:002012-04-17T22:17:15.510+02:00You should compare to numpy with a JIT, such as nu...You should compare to numpy with a JIT, such as numexpr, it would be interesting to see whether PyPy is able to beat the numexpr JIT.Gaëlhttp://www.blogger.com/profile/12853093900244667247noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-14750982177484187212012-04-17T21:04:33.904+02:002012-04-17T21:04:33.904+02:00Sorry, didn't RTFA completely. I just had a lo...Sorry, didn't RTFA completely. I just had a look at the C code.<br /><br />Still, a question: is PyPy doing the optimization of combining operations in one step ?<br /><br />A "good" Fortran compiler should be able to do those optimizations, for instance.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-37943499088636837632012-04-17T21:03:11.107+02:002012-04-17T21:03:11.107+02:00One thing I'll note is that I do from time to ...One thing I'll note is that I do from time to time use the object dtype. Occasionally, I've got multidimensional arrays of objects, and the array operations from numpy are useful. I don't really get a speed advantage there, but the interface from numpy is useful. But its not super necessary and certainly not a priority.Winston Ewerthttp://www.blogger.com/profile/09662140751753841349noreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-92129866454337356992012-04-17T20:57:31.260+02:002012-04-17T20:57:31.260+02:00Hello,
May you get a bit more precise on the GCC ...Hello,<br /><br />May you get a bit more precise on the GCC test ?<br /><br />For instance, is the GCC code using SSE too ? Is it written in a single loop (x[i] = a[i] + b[i] + c[i]) or in several consecutive loops first a+b then (a+b) + c ?<br /><br />Just to know :-)Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3971202189709462152.post-30738594095400478282012-04-17T19:53:08.256+02:002012-04-17T19:53:08.256+02:00I think the string dtype is missing too?I think the string dtype is missing too?Jeffhttp://www.blogger.com/profile/05494402793921207666noreply@blogger.com