|Re: [eigen] row major issues|
[ Thread Index |
| More lists.tuxfamily.org/eigen Archives
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] row major issues
- From: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>
- Date: Thu, 24 Jun 2010 20:05:04 -0400
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:received:in-reply-to :references:date:message-id:subject:from:to:content-type :content-transfer-encoding; bh=Hwit7wiMt1aCG+z/gjUK9yFPFh3kU0udloBy23fFcug=; b=lXsnCol88oIubuydeQ5DDV3grEdZQO/rjtXwQXyXA3Y/fpgNobEKTLGsRm2fFZPEQw MNiYwXWOgW8eoImYsyI3Y8MUKHCJ3T2ibXBypbCWHgK++xPe2u6EPNS6Gbe7B/G2ltu6 PV6vOTJVf2Oq29Z+QUOTnnu+CnXCFwxiyxqlI=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type:content-transfer-encoding; b=Nz0mTf58F3hbDPkjD1/qfKxBLKkL+gq7zDk4kq/CPm3SBfcnV9q2T753HGM9fMVZN2 nSTbytPd1damzBNVZZnTdHoeSFe/2bb5K8T3uW14PA5/oyX4PYQHn7GDjliWlibo5j+Q U6sWjSyniUy7WV7pfS7rIct7uxf2t+X60Gw4Y=
This is awesome!!! Thank you so much.
2010/6/24 Gael Guennebaud <gael.guennebaud@xxxxxxxxx>:
> 100% tests passed, 0 tests failed out of 453 with EIGEN_DEFAULT_TO_ROW_MAJOR=on
> On Tue, Jun 22, 2010 at 1:51 PM, Benoit Jacob <jacob.benoit.1@xxxxxxxxx> wrote:
>> EIGEN_DEFAULT_TO_ROW_MAJOR has never passed all the tests in the
>> development branch because of some errors in products. The QR errors
>> could very well be caused by them.
>> Thanks for the INVALID_MATRIX_TEMPLATE_PARAMETERS fix. Indeed, since
>> TmpMatrixType is a column-vector type, this is needed. Why isn't it
>> called TmpVectorType ??
>> I'll let Gael handle the mixing_fix.patch...
>> 2010/6/22 Hauke Heibel <hauke.heibel@xxxxxxxxxxxxxx>:
>>> I was able to fix the
>>> related error but I am unable to fix this test qr_colpivoting_5
>>> because MSVC's error output is unusable. I attached them anyways.
>>> I assume the errors caused from qr_1 and qr_11 are related.
>>> In order to fix mixingtypes, I needed to introduce two
>>> reinterpret_casts. Since I have no clue about the restrict keyword, I
>>> just attached the patch.
>>> I will leave the rest of the errors untouched for now.
>>> - Hauke
>>> On Mon, Jun 21, 2010 at 11:47 PM, Hauke Heibel
>>> <hauke.heibel@xxxxxxxxxxxxxx> wrote:
>>>> the dashboards for x32 and x64 bit builds with column major matrices
>>>> are now green. After switching to row major matrices we are
>>>> unfortunately seeing a bunch of errors. See here:
>>>> I am right now too tired to try to understand the assertions - maybe
>>>> it is super easy for one of you guys. I am guessing it has something
>>>> to do with the vector logic and the required storage order for
>>>> - Hauke