Sign in with
Sign up | Sign in
Your question

How to get best BANG/BUCK when building a comp

Last response: in Systems
Share
December 3, 2010 3:11:31 PM

I’m relatively new to computer building, but I’ve learned a lot already and I wanted to give a quick summary of advice to new builders

My assumption with this is that you want the most bang for the buck, while meeting a certain performance level. If you build to just have bragging rights / benchmark queen, then all this is NA

Some general notes and stuff ive learned:


CPU: INTEL VS AMD
Don’t be a fanboy, look at CPU benchmarks for programs you will actually be using. Some AMD chips are a better bang/buck than Intel for certain applications, and some Intel are better bang/buck than AMD for certain applications. And it changes from year to year.
The question is NOT "which is better, AMD or Intel"
The question is "what is the chip that gives me the best bang/buck this year while meeting my goals"
If it’s AMD, then great. If it’s Intel, then great. It will all probably change next year.
By not sticking with one or the other, you allow a wider range of options to always get the best bang/buck


CPU COMPARISON – COMPARE OC’d CPU’S
Try to compare similar chips at their max overclock ratings (on similar air cooling) to make a fair comparison.
For example, some AMD chips come out with a higher factory “overclock” than Intel.
So for example an AMD chip might be set at 3.2Ghz, but it can only reach a max overclock of 4.0Ghz. So it is only .8ghz from its max OC
An Intel chip, like the i7 series, might be set at 2.66, but can OC to 4.2Ghz. You are 1.5Ghz below the max OC. So the net effect is that the AMD chip came “pre overclocked”, and you need to take this into account when comparing the chips.
You do this by comparing them at their peak OC settings (even if you aren’t going to run it OC’d), to find the best bang/buck

Compare CPU's on benchmarks that represent what you will do. For example, few games use more than 2 cores. Few applications will make use of more than 4 cores. But a CPU can look like great if it is running a benchmark that is utilizing 6/8/12 cores, etc


OVERCLOCKING = GOOD
Regardless of how you feel about overclocking, it seems to be a consensus that a small overclock of 10-20% (along with a prim95 pass, and cool cpu temps) will not hurt the CPU. It might take 1 year off its 10 year lifespan, but who keeps a cpu for 9 or 10 years anyway?
However, overclocking does go a long way toward reaching a great bang/buck.


BUY PARTS ACCORDING TO YOUR INTENDED USE
CPU / GPU
You don’t need an i7 for most gaming because >60fps is pointless since that the refresh limit of the monitor anyway. You can get away with an i3 or even less. Use the money you save to buy a better GPU.
If you are running CPU intensive applications, look up benchmarks for the specific apps you are going to use. Sometimes an Intel chip will work better, sometimes an AMD will be the best.

MEMORY
Figure out if you need fast memory or not (> 1333), and buy accordingly
For gaming, you don’t need fast memory
For data crunching with tons of memory access, you can benefit from running faster memory and an overclocked bus to utilize it

MOBO
If you aren’t going to use 99% of the features on your mobo, odds are you don’t need the best one out. Get a cheaper one that is ranked high on Newegg for reliability, and be happy with it. Most mobo’s don’t show much if any of a speed difference anyway.
If you are OCing, you probably want to look into what boards are best at OCing your particular CPU.


FUTUREPROOFING = WASTE OF TIME AND MONEY
Always get best current price/performance ratio for what you need right now - Forget out futurizing it. That is a total waste of time and money
You will usually upgrade the system as a unit in the future
The processors will have changed, sockets will have changed, supported memory types will change, supported mobos will change, etc.

Computers depreciate at something around 40%/year.
So let’s say that a $500 computer is the baseline for what you need right now
Let’s compare buying a new baseline machine every year with buying a new kickass machine every 4 years.

Case 1: New $500 machine every year
In this case, you buy it for $500, sell it on ebay for $300, and lose $200 each year

Case 2:
If instead, you try to build a machine NOW that will still do what a (4 years from now) new baseline $500 machine will do.

Here is what happens

The extra performance is more than you need or can use, and you won’t notice much of a difference in the first 2 years that it is actually significantly faster than the new yearly $500 baseline machine

At 40% depreciation, a $2300 machine will depreciate to $300 4 years, so lets use $2300 as the comparison number.

2300-300 = 2000
2000/4 = $500/yr

You spend $500/yr instead of $200/yr in your attempt at futureproofing

If you want the best bank/buck, get what you need right now, and upgrade every year or so.

The other advantage of building/buying once a year is that you can constantly upgraded based on your new uses/goals. Its hard to predict what you will need a comp for in 4 years. but its not that hard to do it for one year.



QUALITY PARTS = GOOD
Quality is king. You will probably sell your computer in 1-3 years on eBay. But you want to sell it WORKING and not because it broke.
If the parts break, they aren’t worth anything. While an older working computer is still work something.
If you do your research, you can usually buy quality at a small premium. The key is to figure out what is actually quality and what is just overpriced junk

I like the newegg reviews and ratings for this purpose

Being able to sell your computer in working condition every 1 or 2 years is elemental to getting your best overall bang/buck


FINDING THE BEST PARTS
Use newegg reviews/ratings and benchmarks on reputable websites
benchmarkreviews.com
tomshardware obviously
Others?

Making your money stretch the best is elemental to getting the best bang/buck


Thoughts?
a b B Homebuilt system
December 3, 2010 4:14:00 PM

Thats a nice bit of research and information,
I vote for this to be included in sticky list for new builders to peruse.
Moto
m
0
l
a b B Homebuilt system
a b å Intel
December 3, 2010 4:39:25 PM

Sorry, I don't. It has already been covered in some of the other stickies, such as by Tecmo34 and Proximon. Some of it is philosophy, which is very subjective. Some of it is too general. For example, I agree that "future-proofing" is often futile (as there's really no such thing) and can certainly be wasteful, but adding some "future-resistance," e.g. by choosing a Crossfire or SLI capable mobo even though you're starting with only one card, or buying a PSU 50W-100W stronger than needed (even for a second card), can be a cheap way of avoiding an expensive rebuild in a year or two.
m
0
l
Related resources
December 3, 2010 4:57:56 PM

i recommend selling the whole computer at the end of the 1-1.5 year period, since it is worth more when it is together and working, and is easier to sell, instead of selling 8 different parts, etc.

this means that buying more PSU than you need would just be a waste of money
m
0
l
a b B Homebuilt system
a b å Intel
December 3, 2010 5:04:03 PM

No, you're not selling anything off, you are adding a second graphics card to a system that was designed in anticipation of it (or of a single more powerful card). A slightly oversized PSU will negate the effects of capacitor aging, allowing it to last a long time.
m
0
l
a b B Homebuilt system
December 3, 2010 5:33:08 PM

I split the difference in some ways: I get a decent $800 computer, then replace it in 2 years.
m
0
l
a b B Homebuilt system
December 3, 2010 6:33:08 PM

Onus said:
Sorry, I don't. It has already been covered in some of the other stickies, such as by Tecmo34 and Proximon. Some of it is philosophy, which is very subjective. Some of it is too general. For example, I agree that "future-proofing" is often futile (as there's really no such thing) and can certainly be wasteful, but adding some "future-resistance," e.g. by choosing a Crossfire or SLI capable mobo even though you're starting with only one card, or buying a PSU 50W-100W stronger than needed (even for a second card), can be a cheap way of avoiding an expensive rebuild in a year or two.



I agree with you tbh,
I should have posted the recommendation for inclusion after some tweaking to make it a more objective and less opinionated piece, but the general idea was sound was my thinking, sorry for jumping the gun a bit :) 
Moto
m
0
l
December 3, 2010 7:35:21 PM

"No, you're not selling anything off, you are adding a second graphics card to a system that was designed in anticipation of it (or of a single more powerful card). A slightly oversized PSU will negate the effects of capacitor aging, allowing it to last a long time.
"

right, i agree,
but it doesnt matter if the psu lasts a long time because it gets sold along with the rest of the system on ebay in 1-2 years!

possibly you could purchase a average psu to sell with your old system and keep the good one, but i dont think thats as efficient


planning for 2 GPUs might be a good idea, especially for a gamer, where he probably just needs a card upgrade in a year. however, when i tried to figure this out, it seemed that most of the time, depending on how efficient the 2 GPUs worked together (often not that efficient at all!), it was still better bang/buck to just sell the old GPU and buy a new one. instead of having the higher overhead of the initially more expensive mobo to support 2 cards, and the cost of the additional card. that seemed to be the general consensus of what ive read also. im not sure on this though.
m
0
l
a b B Homebuilt system
a b å Intel
December 3, 2010 7:45:49 PM

makesumwake said:
...im not sure on this though.

And neither am I, which is the point. This topic is just too subjective to make a good sticky. Prices are always in flux, so the +$30 today for a Crossfire mobo may be +$10 or +$50 tomorrow. The HD2900XT which needed a beefy PSU a few years ago is beaten by the HD5670 today which doesn't.

m
0
l
!