| Re: [eigen] alignment bug - or where is our non sse inversion gone? |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] alignment bug - or where is our non sse inversion gone?
- From: Gael Guennebaud <gael.guennebaud@xxxxxxxxx>
- Date: Tue, 24 Aug 2010 13:01:34 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:mime-version:received:in-reply-to :references:from:date:message-id:subject:to:content-type; bh=vE8QdddVTChsafDT6CikiLwqLKmdqgMFDM1JQpBabz8=; b=Igw55P66WmL29nj7gcQNq90d/Ps+cg5IRhdROfquCd+f8PtdALhn12tilfU4jBzHfk i9OEIhbBp6oFYdZ86VK51ole41nYNu4bouDJ3vuoAbM4J10AuxgCYIqwBXIrX87n6euH ziZ6TEKbcL3sP8j8cB+ZoL2oKJvOHMVkiHjwg=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type; b=UKWz/ufM7GSQEDWG6UDJN++TRSr+dvmNZmDRoZ5whMaJEg/d/0RPOxisxIpYIYzakv a3OSHtCfcXmCu1zVeeYd6lZOyJSEBXbWt1BliUIFEjHrB1wqQJZb6vsAQsPHcOm/XT0I 8q/Vkxb6Qz1WvUlf9HFcMyDac+WvzOpKoJ4xE=
actually this piece of code was really bugged: B =
A.transpose().inverse() was wrong too.
both problems solved.
gael
On Tue, Aug 24, 2010 at 12:29 PM, Hauke Heibel
<hauke.heibel@xxxxxxxxxxxxxx> wrote:
> Ok, I see what you mean. I mixed up the architecture (SSE vs. General)
> and the alignment flag for this case. I would have fixed it in a "bad"
> way by completely going for the non-vectorized version when the data
> were not aligned.
>
> Anyways, you know what you are doing and thank you for taking care!
>
> - Hauke
>
> On Tue, Aug 24, 2010 at 12:23 PM, Gael Guennebaud
> <gael.guennebaud@xxxxxxxxx> wrote:
>> hm sorry but I don't see it. Here is what I have for ei_compute_inverse_size4:
>>
>> template<int Arch, typename Scalar, typename MatrixType, typename ResultType>
>> struct ei_compute_inverse_size4;
>>
>> Anyway, actually I'm doing a different fix which consist in doing
>> unaligned loads/stores if the input and/or output do not have the
>> AlignedBit flags. Here doing unaligned accesses is fine.
>
>
>