Computer build for Physics and Math

[Dan J]

Hey folks!

I'm a proud poppa of a dedicated physics and math student. He's outgrown his laptop and is interested in putting together a system exclusively for crunching numbers and doing physics and math related output. He does not require a gaming build (which seems to dominate the "build" community out there...!), but would like to assemble a powerful system that will last.

Our budget is around $1500. Room for expansion would be good, and any advice about operating system preference - and anything else - would be greatly appreciated! Thanks in advance for your help!

-DJ

 

[phinds]

A reasonable-speed processor should be OK but your budget probably would allow a fast one. Big thing is to get lots of memory. 8 gigs minimum for going forward but your budget would probably allow for 16. It would also be a good idea to get a big hard drive (1 terabyte or more) and set it up as a dual-boot system with windows and linux that way it should be easy to handle any software needs.

 

[Greg Bernhardt]

Will he be running specialized software using big data sets?

 

[Dan J]

Great - thanks, phinds... This is a first build, so I'm thankful for your advice!

 

[Dan J]

Will he be running specialized software using big data sets?

Yeah, "big data sets" sounds like exactly what he's wanting to tackle...!

 

[Greg Bernhardt]

Yeah, "big data sets" sounds like exactly what he's wanting to tackle...!

For $1500 you should be able to get the latest processor and 16GB of RAM.

 

[baudrunner]

From Wikipedia, "A CPU core can execute 4 32-bit instructions per clock (using a 128-bit SSE instruction) or 8 via AVX (256-Bit), whereas a GPU like the Radeon HD 5970 can execute 3200 32-bit instructions per clock (using its 3200 ALUs or shaders)." Some software packages like MATLAB are GPU compatible.

 

[phinds]

From Wikipedia, "A CPU core can execute 4 32-bit instructions per clock (using a 128-bit SSE instruction) or 8 via AVX (256-Bit), whereas a GPU like the Radeon HD 5970 can execute 3200 32-bit instructions per clock (using its 3200 ALUs or shaders)." Some software packages like MATLAB are GPU compatible.

that doesn't seem to make sense. WHAT wikipedia article?

And what does it have to do with the OP's question?

 

[nsaspook]

You might start with a low-end server board system like this. https://www.amazon.com/dp/B00FE29IWK/?tag=pfamazon01-20

 

[Mark44]

From Wikipedia, "A CPU core can execute 4 32-bit instructions per clock (using a 128-bit SSE instruction) or 8 via AVX (256-Bit), whereas a GPU like the Radeon HD 5970 can execute 3200 32-bit instructions per clock (using its 3200 ALUs or shaders)." Some software packages like MATLAB are GPU compatible.

that doesn't seem to make sense. WHAT wikipedia article?

It makes sense to me. Current CPUs contain a handful of arithmetic logic units (ALUs), but some of the graphics processing units (GPUs) contain hundreds of cores that can run concurrently.

And what does it have to do with the OP's question?

I believe the connection is that MATLAB calculations can take advantage of a GPU like the Radeon mentioned. nVidia also is a big player in this market.

 

[phinds]

It makes sense to me. Current CPUs contain a handful of arithmetic logic units (ALUs), but some of the graphics processing units (GPUs) contain hundreds of cores that can run concurrently.

I believe the connection is that MATLAB calculations can take advantage of a GPU like the Radeon mentioned. nVidia also is a big player in this market.

Ah ... that makes sense. Thanks.

 

[Mark44]

As an aside, I just got a new nVidia video card so that I can write some massively parallel code. I don't care at all about games and whatnot, but the programming capabilities are a draw to me.