Computer build - text mining / data analysis

someone16

Senior member
Dec 18, 2003
522
9
81
Looking for advice for my upcoming build. This will be mainly for data analysis including text mining and bayesian analysis in Python and R. Currently using a regular laptop it will sometimes say not enough memory and/or the program will crash alot.

CPU: i7-9700k -> anything similar on the AMD side?
CPU Cooler: Cooler Master Hyper 212 EVO
Mobo: Gigabyte Z390UD -> Not sure if I want to overclock, if so it will be mild. Would I be better just using the default turbo rather than trying to mess around with the settings?
RAM: G.SKILL Ripjaws V Series 32GB (2x16GB) DDR4 3000MHz -> Think there will be any advantage going to 64gb? Or will 32gb be plenty?
HDD: Samsung 860 Evo 256gb ssd + 1TB HDD -> Just HDD's I have lying around
PSU: Corsair CX Series CX650 -> May put in a 1060ti or 1070ti in there, so 650W should be enough? If no discrete GPU would I be better off using something in the 500W range?
Case: Corsair Carbide Series SPEC-05 -> Or whatever else is on sale

Thanks everyone in advance!
 

Flayed

Senior member
Nov 30, 2016
431
102
86
You could use a 2700X for AMD although the Intel chips are a bit faster.
For the memory it all depends on how the software is written as to whether it will take advantage of having 64GB over 32GB. 32GB is quite a lot of memory. Having said that if you wanted to use a ram drive for this software to run from that would likely speed up the program execution time.
The 650W for the system sounds like enough. It may be worth getting a better quality unit than the bronze rated one you listed for piece of mind.
 

someone16

Senior member
Dec 18, 2003
522
9
81
Thank you for the reply, upon further research I found the following:

-R is normally single-threaded, but you can run Microsoft R instead which can parallelize the work across multiple threads. Unfortunately it seems that it only works with Intel.
-R also seems do run its data all through RAM, so if you were to work in bigger data sets you'd either need a ton of ram or use packages like bigmemory to manage the data. Currently I am trying to run data in R but it shows "error: cannot allocate vector of size 14.5gb", so that's why I thought I needed 32gb of ram atleast.
 

Flayed

Senior member
Nov 30, 2016
431
102
86
I didn't know that there are languages that don't support threading. I don't know anything about R though. If it supports having a GUI that will run in its own thread. It doesn't sound like a good piece of software if it dies trying to allocate memory.