Re: [eigen] sse asin implementation |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] sse asin implementation
- From: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>
- Date: Tue, 31 Mar 2009 17:06:40 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:in-reply-to:references :date:message-id:subject:from:to:content-type :content-transfer-encoding; bh=d/OjCrZtDqtN7hjB10xbo7FwtoxvkDW4qiZCEM/KZK0=; b=UZD4DgfQW63PMumXdSf+piplVcj3FRrw+xou/g1lIHY7zz4q9cVzcP12AdZRGOe0RW 8XvzynX+ZXRkOpveWeSRSEMHymyv+HS1YO7SCr5ZD+XT8jv+mnPtye0pPGWIBpWGqxeZ TS8BtRk7gMexudIXBuijkCmd5TKP7fS9YRh5M=
- 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=qypkA7fdQJitBrLnL4uf/884U4Pv4HTJgOXHuKT7ytVsV6TASOQA6qQjmUWk4mYeJj tViV6stwFCho52pHxh1XgioAMKXW+yG07zsDxWV2GE+hLAAvCeM+pQBMMy/XC3QyQN0d aVVEG4ZlmzOB68kgw+Jo1ejvZRwXoOZVSOY9Y=
I forgot to mention, that "pow" is also definitely useful to vectorize
(think gamma correction on an image).
Benoit
2009/3/31 Benoit Jacob <jacob.benoit.1@xxxxxxxxx>:
> 2009/3/31 Rohit Garg <rpg.314@xxxxxxxxx>:
>> On Tue, Mar 31, 2009 at 8:18 PM, Benoit Jacob <jacob.benoit.1@xxxxxxxxx> wrote:
>>> What is the motivation for vectorizing asin in Eigen? Do you mean for
>>> people who want a "cwise asin" on a vector? Do you have a use case for
>>> this?
>> Yes. The page on TODO says that transendental functions could be
>> vectorized and since recently sin, cos, log, exp for floats were
>> committed, I thought that eigen could do with other transcendental
>> functions as well. I was planning to implement more transcendentals in
>> the future, but if this is not useful, please tell me.
>
> Well, the TODO only mentioned very common functions like sin, cos,
> log, exp, for which use cases abound, for example, not long ago on
> this list, Jos Vandenoever had a use case for vectorized cwise sin or
> cos on vectors.
>
> So vectorizing sin, cos, log, exp, as well as sqrt and perhaps
> inverse_sqrt, is definitely useful.
>
> But the inverse trigonometric functions like asin, acos, atan, are
> much less commonly used. What's more, they tend to be used most of the
> time on single isolated numbers, not on all coeffs of a vector. So
> it's not clear if a vectorized asin would be used. I would prefer to
> wait for use cases to appear.
>
> I guess I should have replied sooner but since I didn't take part in
> the earlier discussion I was hesitant.
>
> Cheers,
> Benoit
>