Sign in with
Sign up | Sign in
Your question

Upgrade to SSD?

Last response: in Storage
Share
March 31, 2010 5:22:14 PM

Hi. I am currently using some single threaded software for some mathematical simulations that is very CPU intensive. I have already upgraded my motherboard, memory and cpu. This speeded up the time my simulations takes to run but I am wondering what kind of further boost can be obtained if I make the upgrade from my regular hard drive to an SSD, specifically the intel x-25V. Can anyone help and offer any advice?

More about : upgrade ssd

a b à CPUs
a b G Storage
March 31, 2010 5:28:37 PM

Doubt an SSD would help in this case. List full specs.
m
0
l
March 31, 2010 5:35:15 PM

Shadow703793 said:
Doubt an SSD would help in this case. List full specs.


AMD Phenom II X4 Quad Core 965 Black Edition 3.40GHz
Gigabyte GA-MA770T-UD3P AM3 DDR3 ATX Motherboard
G.Skill RipJaw 4GB (2x2GB) DDR3 PC3-10666C8 1333MHz Memory
Maxtor 200 Gb 7200rpm Hard Drive
m
0
l
Related resources
a b à CPUs
a b G Storage
March 31, 2010 8:11:47 PM

Ahh..... seems like you have an old Maxtor drive. You probably want to get one of the new 500GB (single platter) drives like the Samsung F3.

Also, download HD Tune Pro trial and monitor the disk when you'r running the sim. If the disk is barely used (low I/Os,etc) then a SSD won't help.
m
0
l
a b à CPUs
a c 415 G Storage
April 1, 2010 1:40:35 AM

If your program is really CPU bound then a faster hard drive isn't going to do anything at all. Even if your program writes out it's results to a disk files, Windows will cache the output and do lazy writing to flush it to disk. Unless you're writing data at a rate that's faster then the hard drive can accept it and so much of it that you run out of memory for write caching, a different drive isn't going to change your performance.

There's only three ways to speed up a CPU-bound program:

1) Use a different algorithm that's more efficient.

2) Get a faster CPU.

3) If you have a multi-core CPU, rewrite the algorithm to use multiple threads so that you can put all the cores to work.
m
0
l
Anonymous
a b à CPUs
a b G Storage
April 3, 2010 6:22:02 PM

See if CUDA could help you out (NVidia Graphics cards helping out the CPU).
Stanford's folding projects support CUDA, resulting in a serious performance increase!
m
0
l
a b à CPUs
a b G Storage
April 3, 2010 8:05:51 PM

^ Yes, but only if the code is massively parallel.
m
0
l
!