depend on what the user is rendering/calculating in MATLAB, from my experience (few weeks of experience in MATLAB for image recognition assignment). but I'm gonna say that it'll cost quite a bit

here's the information for sifu to recommend system for you.

CPU: more core  = better (quad core or quad core with HT) *I've heard some pro MATLAB user use an array of PCs to calculate

RAM: MATLAB is one memory hungry application (memory trashing might affect the rendering)

HDD / SSD: *if user uses virtual RAM, it'll affect the performance as well. *faster application startup. MATLAB do startup slow. (I use Crucial M4 128GB also feel it starts very slow.)

GPU: CUDA cores will help in computation, but in the end, it still relies on user's project whether got utilize it or not. nvidia cards with CUDA 1.3 or later (http://developer.nvidia.com/cuda/cuda-gpus) *MATLAB do not support computation acceleration from AMD/ATi

there you go, recommend a machine to fit his needs.

hmm, perhaps

intel core i7 3770 rm909
msi h77ma-g43 rm319
kingston value 4gb x4 rm248
leadtek gtx 550 ti rm369
intel 330 120gb rm319
wd blue 1tb rm249
liteon dvd-rw rm59
fsp hexa 500w rm149
aerocool strike x one advance rm139

total rm2760 (viewnet)

recommended monitors

viewsonic vx2250vm 21.5' full hd led rm350 (jayacom)
dell u2312hm 23' full hd ips led rm559 (boonykun garage sales)

That budget were kinda tight for a full setup..

Intel Xeon E3-1230 V2 3.3GHz RM699
MSI B75MA-P45 mainboard RM224
Kingston 1333 DDR3 ram 4x4GB RM256
Palit GT640 2GB GDDR5 RM389
Corsair CX500 80+ PSU RM195
Intel 330 Series 120GB SSD RM339
WD Caviar Blue 1TB HDD RM240
Xigmatek Asgard Pro case RM159
Lite-On 22x DVDRW RM59
LG IPS2240T 21.5" monitor RM419

Total - RM2979.00

IINM,for MATLAB,you need a better GPU which implements as much of OpenGL on it's hardware as possible..read from some sos..
So,if u can afford more pricier card,the better..