Dual-core Which Processor? Apple vs PC Wireless? Scanning
The views expressed here are ours alone.
We feel they're supported by both our 25+ years of experience and our on-going monitoring of this technology.
Still, in the end, you should take everything here with a grain of salt.
This is our news and op-ed -- our world-according-to-us -- page.
Now, a few timely items; then some ideas on what may or may not be good in this industry:
 |
 |
Processors are the "chips" that are the heart (well, more like the brain) of your computer. Everything your computer does is driven by the processor. For years, high-end "workstations" (computers designed for intensive uses such as graphic design) have been available in multi-processor configurations, so they can spread the workload over more than one processor -- kind of like having a team rather than a single person. Dual-core processors are similar and even better in some ways. Essentially, they incorporate two processors on a single "chip", sharing some resources but capable handling two "threads" of work simultaneously.
How much practical benefit this is depends on the situation. Usually, graphical work -- such as working with digital photos and other images -- is much faster. Simple database work, such as entering an order, is a single task and normally doesn't use multi-processor capabilities. Running another application at the same time, however, such as Inventory Control while entering orders, is two tasks and should benefit from dual-core technology.
We recommend dual-core processors, because the costs are relatively low compared to the potential performance advantages, especially as users increase their demands on their systems.
If performance, alone, matters, and cost does not, then the newest and fastest is the "best" -- keeping in mind that there will be another "best" very soon. If, as is more commonly the case, "best" is a balance between cost and what it gets you, then there's a very interesting pattern in the processor-pricing structure: Processor speed and cost do not increase proportionally!
Take a look at the following table. It lists Intel desktop and laptop processors, in ascending order of speed, with unit costs, and shows how both increase with each model:
| Model | Speed (GHz) | % Chg | Cost* | $ Chg | % Chg | Index** |
| Intel® Core 2 Duo processors for desktops (Conroe) | ||||||
| E4300 | 1.80 | -- | $177.99 | -- | -- | -- |
| E6300 | 1.86 | 3.33 | 187.00 | 9.01 | 5.06 | 0.66 |
| E6400 | 2.13 | 14.52 | 222.00 | 35.00 | 18.72 | 0.78 |
| E6600 | 2.40 | 12.68 | 316.00 | 94.00 | 42.34 | 0.30 |
| E6700 | 2.66 | 10.83 | 512.00 | 196.00 | 62.03 | 0.17 |
| Intel® Core 2 Duo mobile processors (Merom) | ||||||
| T5500 | 1.66 | -- | $206.99 | -- | -- | -- |
| T5600 | 1.83 | 10.24 | 249.99 | 43.00 | 20.77 | 0.49 |
| T7200 | 2.00 | 9.29 | 298.99 | 49.00 | 19.60 | 0.47 |
| T7400 | 2.16 | 8.00 | 425.99 | 127.00 | 42.48 | 0.19 |
| T7600 | 2.33 | 7.87 | 651.99 | 226.00 | 53.05 | 0.15 |
| * Retail cost ** Percentage increase in speed divided by percentage increase in cost; higher value indicates better cost/performance. Source: Newegg.com, February 2007 | ||||||
Note the lines in bold. Intel's E6400/2.13GHz (desktop) processor offers a 14.5% performance increase over the E6300 for an additional $35.00 or 18.7% increase in cost. The next step up gives less than a 10% performance gain at more than a 42% increase in cost. (The "Index" column compares the performance gain to the cost increase; higher values represent a better cost/performance relationship.)
Among laptop processors, the performance improvements of the T5600 and T7200 processors, over the next lower level, are matched with comparable (about 20%) increases in cost, while the next step up, the T7400, gives only an 8% performance advantage with more than a 42% cost premium.
The cumulative values are even more dramatic. The E6400 is 18.33% faster than the E4300 and costs 24.73% more; the E6700 is 47.78% faster but costs a whopping 187.66% more!
Note: The changes in the second digit of the model numbers, from E4xxx to E6xxx or T5xxx to T7xxx represent processor configurations (on-processor memory size and speed) which directly affect performance. So, for example, using the T7200 moble processor offers additional performance advantages, over the T5600, beyond just the 9+% speed gain -- though they are less quantifiable than simple processor speed.
We go through this kind of analysis whenever we configure for or recommend systems to our customers.
Have you seen Apple's adds in which their MacIntosh computers are played by a cool dude from TV's Ed and "PC" is played by a pudgy nerd who has trouble with complete sentences? We don't have any problem with these characterizations. They're kind of fun, and, since it was announced, the Mac has always been the hands-down style leader. Some of them come in colors! Clearly, if you purchase a computer as a fashion statement, the Mac is the one for you.
Our problem is with the ads' "factual" statements. They claim advantages as inherent to the Apple platform that just aren't. Yes, there are more viruses designed to attack WinTel machines... because there are ten times as many of those computers! The copy doesn't say that, with propertly maintained virus protection -- a requirement for both -- both are equally protected. Then the ads turn around and say that Windows' firewall (the software that protects the machine from intrusions over the Internet) intrudes too much, when the reality is that you can configure all firewalls to intrude as much or as little as you like. Virtually ever key marketing point in every ad is exactly like this.
What they definitely don't tell you is that, finally, Mac's use the same processors as WinTel machines (should we call them "AppTels"?) and cost around 50% more for the same configuration. Style does come at a premium after all.
Here's the bottom line: Mac owners are a fevered cult visualizing themselves as "industry rebels". If that or sexy packaging appeals to you, buy an Apple. If you're mostly driven by cost-benefit analysis -- as most successful businesses are -- then it's WinTel and a trip to Old Navy for a clothing upgrade.
Here's the bottom line: The one-time cost savings of "going wireless" do not offset the other on-going "costs" that come with that technology.
It's true that the cost of very high-quality desktop units has come down to virtually negligible. But, the problems come with actual implementation. In other words, what is your end product?
It's important to realize that when you scan a document, you're making a digital image of it, and detailed digital images are very, very large. In practical terms, few businesses can justify the disk space required to convert their paper documents into digital files on disks, so they must compromise quality. If you can live with digital images equal to low-quality faxes, you may not have a problem, but most businesses will. Of course, there's also the issue of maintaining a database that gives you access to where all these files are stored. It's all not as simple as it sounds.
Now, that's all based on bit-mapped images, that is, essentially digital photographs of the papers at 100, 200, or 300 dots per inch. Again, that's a lot of dots, but there's another way. The technology exists to convert the dots (actually, they're called "pixels") into vectors, actually computer equations that recreate lines and text without actually saving huge images of them. The most popular of these is the PDF or Portable Document Format from Adobe Acrobat. The good news is that this works pretty well indeed; the bad news is that you must have the Adobe Acrobat application (not just the free Reader download) and this is a fairly expensive package.
OCR -- Optical Character Recognition -- is a variation of vector conversion. In this case, you scan a document directly (or indirectly) into your word-processing application, and this works surprisingly well. It's not perfect. You will need to edit some, and the more special formatting there is, such as tables and margin changes, the more you'll have to edit. And, graphics come over very poorly if at all. Still, in environments such as law offices, the time saved retyping lengthy documents can be dramatic! Just keep realistic expectations.Here's the bottom line: Scanning works pretty well for specific properly implemented, usually fairly limited purposes. It works poorly from a general "scan everything" "eliminate paper" perspective.
