Upgrade Path Suggestions -> Please
 

I have a a budget of a about $250 and would like to give my system a shot in the arm because I know I wont be able to replace for 2-3 years.

:scratch:

I am interested in opinion of bang for the buck and any insight to any compatibility issues. I currently am using LM 9 and plan on sticking with linux but may experiment with new distro if I upgrade / drive.

Thanks in advance for your input. :cool:

I was considering(I will consider other options)->

CPU's

$160-Under Budget but leaves room for other parts
http://www.newegg.com/Product/Produc...82E16819103727

$265 -> Kills Budget +
http://www.newegg.com/Product/Produc...82E16819103849


Hard Drive->
$72 - SSD I thought it would work as a good boot drive giving a system speed boost--> NOT SURE if it works well with Current Linux Distros.
http://www.newegg.com/Product/Produc...82E16820227393

$60 -> Just another 1T of storage. Don't need today but this is cheap. I would perfer a speed boost of the SSD but not sure if it is there.
http://www.newegg.com/Product/Produc...82E16822152185


I would look at a video card but I believe the CPU and or HD would impact my usage more significantly.

Computer Use

Web Surfing
Learning Linux
Media Collection (Video,Pictures,Music)
--Editing, Ripping, Storage

Occasional Light Games

You haven't mentioned RAM - if your motherboard can take any more then add some. As with Windows, adding more RAM will make your Linux PC run faster.

Oversight -> 4 GIG DDR 1066

I doubt if 2.9G - 3.2G is worth $159 in your situation. It rarely translates to a significant increase.
Buy enough ram. Extra cores aren't a huge deal in a home user's box.
Think about a motherboard - faster bus speeds, etc.

business_kid,

Thanks or the input.

So in your experience Dual to Quad core is minima's impact ?

Hi,

Your still using a Desktop system and utilized as single user. If you are wanting to VM then more cores would be a wise choice. How extensive is the video editing?

The increase in memory is a gimme that will show gain.

Sure, storage media choices will speed things up but how you setup the SSD & spinning disks will dictate the speed of the system as related to storage. Bumping memory and setting up RAMDISKS may show a better increase at a lower cost.
:hattip:

If you want your hardware upgrade to be as cost-effective as possible then you need to find out what your current bottlenecks are and improve in that/those area(s). When doing the sort of things that make the system slower than you would like, is it CPU-bound (CPU(s) running at 100%), GPU-bound (CPU(s) running at 100% ?), I/O-bound (processes in I/O wait, long I/O queues) or memory-bound (processes in memory wait, swapping activity).

It's a long time since I did this sort of exercise (and never regards GPU performance) so I am not familiar with Linux tools to do the job but the principles don't change. Hopefully other LQ contributors with current performance analysis tools knowledge will comment. systat looks promising.

Thank You for everyones input. I decided to go with ->


http://www.microcenter.com/single_pr...uct_id=0344221



I am going to play with this for a awhile and see if I want to invest more into a CPU. I also want to see if they come down a bit in price as I decided to invest a few more dollars into a better boot drive than I originally thought.


My next question would be to decide on OS install. I like LM 9 but thought about experimenting since I will need to reinstall anyway.

I was thinking of another Debian based spin but some only come in 32 bit variations and I don't want a performance impact.

Any thoughts ?

Debian Based (Where I have been spending my time)

Mint 9
LM Debian (32 Bit -Only Have)
LM 9 Isadore(64 Bit-Current System)
Dreamlinux (32 Bit - Only Have)
Ultimate 2.7 (64 Bit)
Debian (64 BIT)

New Territory -> Not Sure if I want to take the leap, still learning)

Slackware
CentOS

I agree! It's an utter waste to shell out cash for something if it doesn't improve performance significantly. Some tools for logging system statistics next to sysstat are psacct (coarse), Atsar or Dstat or collectl, Atop (allows you to replay process table later on) but it makes sense if one designed the system with respect to its intended use beforehand. (For instance a network proxy has different requirements than a video editing workstation, the common newbie partitioning scheme of "/, (/home, )swap" doesn't allow one to split writes over disks in a fine-grained way and different file systems can be used for different reasons, for instance one may want to trade in journaling overhead for fractionally lesser writes and more speed.) Also a baseline should be set up before making changes and afterwards testing changes with tools like stress, bonny et cetera makes sense too.

No, I wasn't saying that really. I don'y have that experience.
I was trying to say that multicore systems outperform others only on cpu-intensive stuff. I have a single, and a dual core box here. According to computer sales people the single core is 'underpowered.'
I am not so sure. Compiles are obviously faster on dual core, but otherwise, life is much the same. I don't do gaming. That uses all the ram, cores, disks and video performance it can find as a rule. Once you go multicore you gain that advantage only while running from cpu cache, so cache is important. So don't do multicore on the cheap.

It's different on a server where the load is heavy, and each core can have it's own process and the output is awesome.

Run top at any time. What's your cpu usage? The highest I've seen is 150% on 2 cores.

Thanks for the clarification & insight :cool: