OK, I manage to compile R-1.9.0alpha (2004-03-16) with the ACML. It's
rather easy:
- Install ACML. I installed in the default place (/opt/acml-1.5).
- This didn't work: ./configure --with-blas="-L/opt/acml-1.5/gnu64/lib
-lacml"
- Make symlink for libacml.so in /usr/local/lib.
- This worked: ./configure --with-blas="-lacml".
It passed make check, whereas the build with Goto's BLAS failed on this box,
as I reported previously.
Using a couple of examples that Prof. Dalgaard used previously in his post:
> set.seed(1234)
> m <- matrix(rnorm(9e6),3e3)
[with ACML]
system> system.time(solve(m))
[1] 28.58 0.98 31.99 0.00 0.00> system.time(m %*% m)
[1] 18.73 0.19 20.25 0.00 0.00> system.time(crossprod(m))
[1] 13.40 3.02 17.66 0.00 0.00
[standard BLAS]> system.time(solve(m))
[1] 165.43 0.86 179.47 0.00 0.00> system.time(m %*% m)
[1] 164.35 0.08 164.43 0.00 0.00
system.time(cr> system.time(crossprod(m))
[1] 79.36 0.11 85.53 0.00 0.00
I have not check the docs, but my guess is that the library is not threaded.
BTW, this is on dual Opteron 244 with 16GB of PC2100 and ATA100 EIDE HDD.
Best,
Andy
> From: Liaw, Andy
>
> Dear R-devel,
>
> Has anyone played with this?
> http://www.amd.com/us-en/Processors/DevelopWithAMD/0,,30_2252_
> 2282,00.html
> <http://www.amd.com/us-en/Processors/DevelopWithAMD/0,,30_2252
> _2282,00.html>
> .
>
> I'll probably give it a shot...
>
> Best,
> Andy
>
> Andy Liaw, PhD
> Biometrics Research PO Box 2000, RY33-300
> Merck Research Labs Rahway, NJ 07065
> mailto:andy_liaw@merck.com <mailto:andy_liaw@merck.com>
> 732-594-0820
>
> ______________________________________________
> R-devel@stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-devel
>
> --------------------------------------------------------------
> ----------------
> Notice: This e-mail message, together with any attachments,
> contains information of Merck & Co., Inc. (One Merck Drive,
> Whitehouse Station, New Jersey, USA 08889), and/or its
> affiliates (which may be known outside the United States as
> Merck Frosst, Merck Sharp & Dohme or MSD and in Japan, as
> Banyu) that may be confidential, proprietary copyrighted
> and/or legally privileged. It is intended solely for the use
> of the individual or entity named on this message. If you
> are not the intended recipient, and have received this
> message in error, please notify us immediately by reply
> e-mail and then delete it from your system.
> --------------------------------------------------------------
> ----------------
>
------------------------------------------------------------------------------
Notice: This e-mail message, together with any attachments,...{{dropped}}