Over the next few months, a lot of you will face some tough upgrade choices. Do I go with DDR? Should I buy a 133A board and keep my RAM instead? Should I just leave well enough alone for a while?
To help you, let me share a simple little formula I use to assess whether an improvement is worth it or not. I call it the Bang for the Buck Index (BBI)
First, I figure out my net cost in making the change. By that I mean how much additional cash will I end up laying out after buying the new equipment and (if applicable) selling the old equipment.
Second, I figure out how much performance improvement I’ll get out of it.
Then I take the first number, and divide it by the second. I look at the answer. If the answer is less than $1,000; that tells me I’m getting a reasonably good bang for the buck. The lower the number is, the better bang for the buck.
If the number is above $1,000, I’m not getting such a good bang for the buck. The higher the number, the worse the bang for the buck is.
Example 1: Upgrade
I have a KT133 motherboard. A few weeks from now, I start thinking about a KT133A motherboard. I look around a bit and figure I’ll get about a 4% improvement over what I have now.
Let’s assume the motherboard costs $140. Not considering what I’ll do with the old motherboard; the math would be:
$140/.04 = BBI of $3500.
At first, this doesn’t look like a good upgrade. However, I’m not likely to just let the old motherboard rot. I’ll sell it, or put it in somebody else’s system.
What I should do is figure out what I’ll get (or would have gotten) by selling it. Let’s say I could sell the board for $100. The math then becomes:
$140-$100 = $40/.04 = BBI of $1,000.
That’s not too bad.
Example 2: New System
I have to choose between a KT133 motherboard and a KT133A motherboard. Same 4% difference. The KT133 motherboard costs $120; the KT133A costs $140. Doing the math:
$140-$120 = $20/.04 = BBI of $800.
Pretty good.
Processor Upgrades:
As a rough conservative rule of thumb, you should assume that the real improvement you’ll get is about 60% of the percentage increase in Mhz.
Example 3: Processor Upgrade
Let’s say you have a processor you’re running at 800Mhz. You want a processor you reasonably expect will get you 1100Mhz. Here’s the math:
1100Mhz-800Mhz = 300Mhz difference. 300Mhz/800Mhz = .375. .375 X 60% = .225.
Let’s say the new processor will cost you $230, and you can sell the old one for $80. The math would be:
$230-$80= $150. $150/.225 = BBI of $667. That’s pretty good bang for the buck.
Exception for gamers: Any improvement you’ll get is limited by the headroom your video card has. If your video card won’t put out a single frame per second more after 900Mhz, then you either have to calculate based on 900Mhz, or also include
in the calculation the net cost of a video card that will give you the full range of improvement.
Example 4: Gamer Upgrade
Assume the 800/1100Mhz upgrade described above, but the video card won’t do anymore past 900Mhz.
No video card upgrade
900Mhz-800Mhz = 100Mhz. 100Mhz/800Mhz = .125. .125 X 60% = .075
$230-80 = $150. $150/.075 = BBI of $2,000. Not too good bang for the buck.
Video card upgrade
Assume you can buy a video card that will give you the full benefit of the processor increase for $250, and you can sell the old one for $100.
Net cost of processor upgrade = $150
Net cost of video card upgrade = $150
Total net cost of upgrade = $300.
$300/.225= BBI of $1,333. Not so hot bang for the buck, but better than not upgrading the video card.
Full real world example:
You have a 900Mhz TBird system with a KT133 motherboard using a GTS2 card and 128Mb of PC133 CAS2 RAM.
Next spring, you start to plan for an overclocked 1.5Ghz Palomino system with a DDR board, DDR RAM and an NV20 video card. Assume the NV20 has enough headroom to take full advantage of the Palomino.
Your alternative is to keep the RAM and buy a 133A motherboard instead.
Assume the Palomino costs $200, the DDR board costs $150, the 133A board costs $130, the DDR RAM costs $120, and the video card costs $300. Assume a 4% improvement from the 133A board, and 10% from the DDR board.
Also assume that at the time, your processor is worth $50, the KT133 board is worth $70, the video card is worth $80, and the RAM is worth $60.
Amount of Performance Improvement
133A board (you keep your RAM)
Net costs:
CPU: $200 – $50 = $150.
133A motherboard: $130 – $70 = $60.
Video card: $300 – 80 = $220.
Total net cost = $150 + $60 + $220 = $430.
Performance improvement:
1500Mhz – 900Mhz = 600Mhz. 600Mhz/900Mhz = .667. .667 X 60% = .40 improvement.
Improvement of 133A board over KT133 board = .04 improvement.
.40 + .04 = .44.
$430/.44 = BBI of $977.
DDR board: (you replace the RAM)
Net costs:
CPU: $200 – $50 = $150.
133A motherboard: $130 – $70 = $60.
Memory: $120 – $60 = $60.
Video card: $300 – 80 = $220.
Total net cost = $150 + $60 + $60 + $220 = $490.
Performance improvement:
1500Mhz – 900Mhz = 600Mhz. 600Mhz/900Mhz = .667. .667 X 60% = .40 improvement.
Improvement of DDR board over KT133 board = .10 improvement.
.40 + .10 = .50.
$490/.50 = BBI of $980.
In this particular example, it works out to be the same. If there were more RAM, DDR would come out somewhat worse.
This Is A Tool, Not A Rule
The purpose of this is to give you a fairly simple tool to judge whether or not you’re getting real value for your money. It is not eternal law. It just lets you
judge one of the factors in determine whether an upgrade is worth it or not. Don’t take it literally, and automatically upgrade if the figure is $980, but reject any
upgrade with a BBI of $1,100. A difference of 10% in BBI is probably not significant, a difference of 50% or more probably is, but again, it’s only one factor.
The figure of $1,000 is not written in stone; that just happens to be the figure I happen to use. I figure it’s a pretty decent deal if it works out to somewhat less than the cost of what I would consider a high quality budget system. If you’re more comfortable using a higher or lower number, go right ahead.
There are plenty of other reasonable modifications to this. For instance, you might not be interested in any upgrade until you get a 50% improvement out of it. You might set a dollar cap on any improvement. Whatever works for you. I’m not Moses coming down from the mountain with this; modify it to make you happy.
The numbers I came up with are educated guesses; the point of them is not to tell you what the BBI is likely to be down the road, but to give you a roadmap that you can use to plug in the real numbers when you’re deciding.
Hope this helps.
Be the first to comment