What does JMP need

[#1]

Specially with 600k rows and working with complex 30x30 tabulates JMP starts bleeding with my actual laptop (Mid 2010 MacbookPro) and it seems the time has come to purchase a new one.

But what does JMP need really to run fast and smooth?

What is the most important spec of a laptop which JMP really takes advantage?

Is it Processors Cores?

Is it Processor Speed?

Is it L2 Cache?

Is it RAM Speed?

Is it Amount of RAM?

Hence I am unable to buy the best, I must find a good balance of those factors.

Many thanks!

[#2]

I would rank, in priority (and this is just an opinion)

Amount of RAM

Processor Cores (if doing a lot of statistical analysis)

Processor Speed

L2 Cache

RAM Speed

The amount of RAM is the biggest factor.  8 GB is good for 64-bit, 16 GB is better.

Brian Corcoran

JMP Development

[#3]

I've never seen JMP threading so processor cores are pretty useless, unless you are running other high performance demanding programs at the same time.  Otherwise you should be good with 2 cores.

What is important for you depend on what you will do with JMP:

• For calculations you need processor speed
• For massive datatables, you need RAM and L2 (or better L3) cache (faster memory-access)
• For operations on datatables, both CPU and RAM is important
• If your datatables are really massive (millions of records), I would recommend to put a SSD in your laptop, preferably one with fast write-access.
• RAM Speed is less important for PC's.

Jan

[#4]

Related Question:

Mac versus PC speed.  I am working on a Mac (2.6Ghz i5 with 16 Gb of Ram) to support some clients who run Dell laptops (unknown specs, but 1-2 years old).

The script is large and involves incorporating between 2-5 data files and generating about 100 plots.  The table size is ~45 columns and between 10,000 and 50,000 rows.

The speed difference is on the order of 3-5x faster on the PC. I have not worked to define the difference because it is not relevant.  But I would like to know if there are some objective speed specs comparing Mac and PCs?

[#5]

There are lots of places where JMP is multithreaded, mostly in the platforms. For example, try Distribution with lots of columns.

You'll see all your processors light up as JMP sends each column to a different thread.

While JSL itself isn't multithreaded, anything that it calls that is multithreaded will use all available cores.

So, multiple cores can make a difference.

As BrianCorcoran points out, the most important factor for performance is RAM. We typically recommend that you have at least twice as much physical memory as the largest data table that you will use. So, a 4 GB data table would require at least 8GB of physical memory.

After that, number of processors and processor speed is next.

-Jeff