[R-sig-hpc] New AMD Bulldozer for Parallel Computing with R

Andreas Recktenwald a.recktenwald at mx.uni-saarland.de
Fri Mar 23 13:24:21 CET 2012

Am 23.03.2012 12:50, schrieb Brian G. Peterson:
> On Fri, 2012-03-23 at 12:14 +0100, Andreas Recktenwald wrote:
>> Dear R enthusiasts,
>> i`m planning to set up a new PC and consider to buy an AMD FX-8120. For
>> my estimations (with floating-point data) I use the genoud optimizer in
>> the package rgenoud which (at least I think) would be able to handle the
>> 8 cores. Is anybody using the AMD proc for computing with R and could
>> tell about the performance? What scares me a bit to just buy it are the
>> huge amount of benchmarks on the net which tell that AMD has really bad
>> performance ( especially for single-threaded apps). Would the Intel I7
>> 2600 be the better alternative? All the Intel 6-core processors are out
>> of budget for me.
> Tests we've run here with multi-threaded R had horrible performance on
> the Bulldozer processors.
> For 'budget' 6-core systems, we've stuck with the AMD Phenom II
> processors and overclocked them. We have several running happily at
> 4.2GHz.  For non-budget systems for now we're using i7's and Xeons.
> The Phenom II wins in performance per watt and performance per dollar at
> this point, but the i7 and Xeon systems are faster by 10% or so.

Thanks for your replies,
really doesn't look too good for Bulldozer. Which OS where you running 
the tests on? I'm using Lubuntu 11.10 (forgot the mention this in my 
initial post) and I'm recently read that Bulldozer performance increase 
when using gcc with march=bdver1 or the AMD Open64 compiler.
At the moment only point for bulldozer are the price ( FX-8120: 150 €, 
Xeon E3 - 1230:  200 €, i7 2600k: 250 €).

I think i should take the money get an i7-2600k and then overclock this 

Best regards

Dipl.-Kfm. Andreas Recktenwald

Universität des Saarlandes – Fakultät für Rechts- und

Lehrstuhl für Statistik und Ökonometrie

Campus C3 1, Zimmer 206

66123 Saarbrücken

Tel.: ++49681 – 302 3571

More information about the R-sig-hpc mailing list