Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Macbook Pro 2014 15" Statistical Computing Performance

M

mendico

macrumors newbie
Original poster
Mar 30, 2015
14
5
I'm thinking to buy MGXA2LL/A (Base model of 15" Macbook Pro with quad core 2.2 cpu) since there are good discounts. I can also buy the new 13" for the same price.

I am using R software for some statistical computing in bioinformatics area for my PhD study. Which model I should take? Is there anybody using R statistical software with Macbook Pro 15" 2014?

It has I7-4770HQ CPU while 13" model has I5-5257U CPU with clock speeds 2.2 and 2.7 respectively. Does 13" model gives better performance than 15" model because of its higher single core speed?

I should buy the computer tomorrow. So your advices are very important for me.
 
I run R and SPSS on my 13" rMBP without any issues (16gb RAM, 1tb SSD, i7). The question will probably depend on the size of your datasets. For larger datasets I use my university's server SAS implementation via web browser. I've run SAS a couple times via VirtualBox on my macbook, but much prefer the server installation. Hope that helps.

EDIT. I missed your specific question, which was "does the number of cores matter for this type of software vs. clock speed". Unfortunately, I don't know the answer to that.
 
mendico said:
I'm thinking to buy MGXA2LL/A (Base model of 15" Macbook Pro with quad core 2.2 cpu) since there are good discounts. I can also buy the new 13" for the same price.

I am using R software for some statistical computing in bioinformatics area for my PhD study. Which model I should take? Is there anybody using R statistical software with Macbook Pro 15" 2014?

It has I7-4770HQ CPU while 13" model has I5-5257U CPU with clock speeds 2.2 and 2.7 respectively. Does 13" model gives better performance than 15" model because of its higher single core speed?

I should buy the computer tomorrow. So your advices are very important for me.
Click to expand...

I used to use R, but now when I do non-proprietary statistics I use SAS. Back then it was very complicated to use multi-core CPUs fully in R, but since parallel is now included in most installs I would assume you can take advantage of it. If your project is threaded, the i7 quad will be hugely faster (2x or possibly more) than the ULV i7 dual, depending on workload. If your project is complex and not threaded, consider optimizing it to be threaded as it will help you in the long run.

The most important part is how large your data sets are. IIRC R loads the data sets into RAM, so plan how much RAM you need accordingly. I'm sure you know you can't upgrade the RAM after purchase.

Another important factor is how long you will need your data sets to run. For most shorter, single threaded computations, a 13" rMBP will be fine. But the longer and more threaded it is, the more you will likely need the larger system because the 13" is less suited to running at 100% without throttling. If your data sets are even more complex (like my old protein and fluid dynamics sets) then you need to get a real mobile workstation, like a Dell Precision or HP Zbook.
 
If you can use utilise the parallel package, go for the quad core. If not, there will not be not much difference between the new 13" and the 15". As far as performance goes, R is not well designed and has a very inefficient interpreter. Higher clocks do help, but the speedup is not linear — I guess branch mispredictions and cache behaviour play a significant role here.

P.S. I would really encourage you to inquire whether your university can give you access to a supercomputing cluster. Especially for complex analyses on large datasets, the time savings are considerable. And you need to spend less time optimising your scripts.
 
Thank you to everyone for your valuable opinions. Currently, I'm using Macbook Air 2011 with 4 GB ram. Since R software loads biological datasets to the RAM, Macbook Air becomes useless.

As you know there are 2 base models with 2.2 ghz and 2.5 ghz cpu. I don't play games so does 2.2 model enough for me? The best prices I found in my country are (I converted the prices to USD currency):

The base model with 2.2 cpu: 2150 USD (gives 330 USD gift card) =~ 1820 USD

The model with 2.5 cpu: 3000 USD (gives 470 USD gift card) =~ 2530 USD
or in another store
The model with 2.5 cpu: 2700 USD (gives 40 USD gift card) =~ 2660 USD

If you were me, which model you would buy? The base model is affordable for my budget. Does the second model make significant advantage for me? (No games, maybe little video tasks for hobby)

My university doesn't give an access to a supercomputing cluster since it doesn't have. But my advisor plans to buy a powerful workstation using project fund for heavy tasks. I'll use this Macbook to develop algorithms for gene network inference.

Note: Apple Store Turkey prices for these models:
The model with 2.2 cpu: 2575 USD (education: 2370 USD)
The model with 2.5 cpu: 3230 USD (education: 2970 USD)

I believe that the new Broadwell model will be ~2370 USD. So there will be 550 USD difference for a bit fast cpu, fast ssd, 1 hour extra battery and force touch pad (According to the posts in the forum). Are they worth 550 USD?
 
Last edited:
leman said:
If you can use utilise the parallel package, go for the quad core. If not, there will not be not much difference between the new 13" and the 15". As far as performance goes, R is not well designed and has a very inefficient interpreter. Higher clocks do help, but the speedup is not linear — I guess branch mispredictions and cache behaviour play a significant role here.

P.S. I would really encourage you to inquire whether your university can give you access to a supercomputing cluster. Especially for complex analyses on large datasets, the time savings are considerable. And you need to spend less time optimising your scripts.
Click to expand...

My thesis advisor wants me to develop scripts in R. What is your opinion for my previous message in the thread?
 
The base retina 15" will be more than sufficient for your needs. I do some fairly crazy stuff with R and I still have the 2012 model :)

When running longer sequences, don't forget to use the parallel package! Usually its a simple as converting

X <- lapply(...)

to

X <- mclapply(..., mc.cores=4)
 
Since your software is forced to run in RAM and your MBA's 4gb is already limiting your use, the 8gb of ram on the rMBP 13 (unless you upgrade it - which will make it already more expensive than the rMBP 15) will limit you sooner than the standard 16gb of ram on the rMBP 15. I'd go for the rMBP 15 unless battery life & portability are more important.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back