Archive for October, 2005

CPU happenings

Tuesday, October 25th, 2005

In the news recently: the Dual Core Xeons suck. The AMD Opterons outperform them on every benchmark, and in many cases the slowest Opteron is faster than the fastest Xeon. The Xeons also use almost twice as much power as the Opterons. Why would you buy a Dual Core Xeon today? I’m not sure.

In other news, AMD has a round of price cuts at the end of October. The S754 chips are getting phased out soon, with the M2 socket slated to be introduced at the beginning of 2006. What does that mean today? It means that the high-end S754 and low-end S939 chips offer the best value on the desktop front. An OEM S754 A64 3700+ can be had for just $165 from ewiz.com. Pair that with an nVidia 6100-based motherboard and you have a workstation that easily outperforms a Dell Optiplex GX620 for roughly half the price!

Everyone agrees that Intel does have a winner on their hands with the Pentium M, but they can’t introduce it for all purposes without cannibalizing their other CPU lines. So that’s my summary: buy Opteron for the servers, A64 for the desktops and Pentium M for the laptops. That advice will probably be true for the next six months.

Computer Buying Guide

Thursday, October 20th, 2005

People are always asking me for my advice when they build a computer, so I’ll write down the advice once and for all.

1) The Display

The monitor is the most important part of your system. It is one of the few things you interact with directly and the component most directly linked with your productivity. You should buy the largest display you can afford. It should probably cost at least half as much as the rest of your computer. It should be an LCD, unless you’re not paying for your electricity and have a very low budget. It should have at least VGA and DVI inputs. As of this writing, I would recommend the Dell Ultrasharp line of displays.

The only other trick here is that two 19″ LCDs can cost less than a single 20″ LCD. More screens makes you as productive as a single large screen, so the choice depends on your specific applications.

Dell 2005FPW

2) The Keyboard

The keyboard is the next most important part of your computer. It is one of the two components which can damage your body if used improperly, so you should choose the most comfortable keyboard. I prefer the ergonomic layout (as does EVERYONE after using it for a while), so my recommendation here is the Microsoft Natural 4000 Keyboard.

Microsoft Natural 4000

3) The Mouse

As you might have guessed, I consider the mouse the next most important component. It is also the most difficult to make a recommendation on. There are many design variations, and you should try several out and choose the one you find most comfortable. The mouse should definitely be an optical one, and not a ball-mouse. The mouse should definitely have a scroll wheel. Features like extra buttons, wireless capability cost extra and are worth it if you use them.

4) The Case

The case is the next component you should choose. It is highly dependent on the parts that follow later in the guide, and they, in turn are dependent on the case. As one of my professors used to say, “it’s a whole-system design issue.” In general, you will want to decide on the size of the case, and then the aesthetics. The current recommendations are the Antec Performance I P180 or P150 for a full-size ATX tower, or an In-Win Z720 for a MicroATX tower.

5) The Optical Drive

The rest of the components are not in order of importance. You should buy a DVD burner because the price difference between optical drives is very small. As of this writing, a regular CD-ROM is ~$15, a DVD-ROM is ~$20, a CD-RW is ~$25, a CD-RW/DVD-ROM combo drive is ~$30, and a DVD-burner of all formats is ~$40. Spend the extra $10 just in case you want to burn a DVD sometime in the next few years.

6) The Hard Drive(s)

You should purchase the fastest drive that you can stand. Generally, the faster the drive, the louder it is. While SCSI U320 drives are wicked fast, I don’t want to be within a few feet of one. The Western Digital Raptors are the fastest SATA drives, and are quiet enough for your machine. Standard SATA drives are nice and quiet. Your drive selection will depend on your budget, your storage space needs and your noise sensitivity.

7) The Rest

Buy at least 512MB of RAM and leave room for more. This depends very much on your application load. Unlike most guides, I leave the Motherboard/ CPU/ RAM/ Video combination until the end. This is mostly because the recommendations for these change every few months, and each time there are different recommendations for different budgets. In general, as of this writing in October 2005, an Athlon64 X2 processor should be the core of your system. The rest depends on your exact needs.

There are already a number of guides out there that keep track of the latest recommendations. After you read this guide, you should look at the latest Ars Technica guides and the latest Anandtech Price Guides.

Fall 2005 Distro Wars

Wednesday, October 19th, 2005

OpenSUSE 10.0 wa released on October 6.
Mandriva Linux 2006 was released on October 13.
Ubuntu 5.10 was released on October 13.

I’m a big fan of the Ubuntu distribution because of the underlying philosophy and the fact that Mark Shuttleworth’s personal agenda is better than any corporate agenda of Mandriva or SUSE. Now that Debian/Ubuntu has Mark’s deep pockets on their side, we’re sure to see excellent results.

Selling Linux desktops

Monday, October 10th, 2005

Neil McAllister over at Infoworld suggests that Linux vendors copy Apple:

Suppose a single vendor took the big gamble and offered the whole ball of wax: a complete desktop Linux distribution pre-installed on certified custom hardware. I don’t mean a budget white box, either, but a well-crafted system designed with usability in mind, fully supported, but with the built-in cost savings of open source.

This is an interesting idea, but undoubtedly not a new one. Why hasn’t anyone done it? I think the answer is that people have done it and failed. You or I have not seen a nice shiny Linux box because nobody wants to buy one when they can buy a Mac mini instead.

One big difference between the Mac mini or the iMac and any Linux machine: the software. I claim that while the latest Linux distributions are usable, they’re not as nice to use as OS X. OS X has a lot of usability research behind it, and an army of well-paid software engineers. However, the point i’m trying to make is that the Linux movement is catching up, slowly but surely. Perhaps Ubuntu Dapper Drake will be as great to use as OS X. If not Dapper Drake, then the next one. And perhaps then we’ll be able to sell a shiny Linux box with great Free software in it.

Speaking of the Mac mini, there’s new competition: the Voodoo Idol.

Better Desktop initiative

Monday, October 10th, 2005

In the news today: Novell has announced their Better Desktop project, which provides the Linux community with the real-world usability information that has always been lacking in the FOSS movement. Microsoft and Apple have put in a lot of effort into usability studies in order to design their interfaces, but most Linux projects have traditionally been developer-centric. With the backing of the big boys like Novell, this will soon change.

OpenSolaris + Wine = the new hotness?

Thursday, October 6th, 2005

I’ve just become aware that Sun has a nice Wine tutorial for JDS.

Perhaps OpenSolaris, JDS, and Wine is the new killer platform? Discuss.

PS: there’s also a nice Wine guide at Frank’s Corner.

“Office” and Linux

Thursday, October 6th, 2005

In the news today: Microsoft: No Office on Linux ‘at this time’.
Isn’t that obvious? Why would they bother porting a really big application to a new platform when that platform already has an excellent alternative? Especially when the people on that platform are likely to prefer the free alternative even if MS Office was an option?
To go even further, why do we need an “office” suite anyway? With today’s AJAX and DHTML tricks, it should be possible to provide all of the functionality of MS Word in a browser, to run on any platform. Here’s an example. Here’s another example.
Next, Google and Sun are in the news for possibly trying to do something similar, though there is no credible source from either company saying so explicitly. But the possibility that Google would launch some kind of an initiative to tie the OpenOffice.org suite into a browser is very exciting. Keep in mind, however, that this is not exactly a new idea. The overall goal is to improve the process of collaborating on a document, the exact thing Microsoft has tried with its Sharepoint Services.

Linux support

Thursday, October 6th, 2005

Paul Murphy says Linux tech support is badly needed. His readers’ comments agree. There is a serious lack of support for the Linux OS for the average Joe. Most of their friends have only dealt with Windows, all the support people at their place of employment only know Windows, and online sources of help are not as helpful as they could be. Paul says local entrepreneurs need to step in and offer this service.

While Paul may be on to something, my posts in my local Craigslist computer services classified offering Linux support hasn’t received any responses.

YALB competition

Thursday, October 6th, 2005

I didn’t realize that there already is a “Yet Another Linux Blog”. I guess we’ll just have to compete.

For some background, I’m a graduate student in Electrical Engineering at the University of Rochester. The guy running the other blog seems to be a “project manager” at a Fortune 500 company. He gets his wife to test out Linux distributions and post to his blog about them.

Why I think Linux is the future.

Wednesday, October 5th, 2005

Linux is the future? Why? Because there are many millions of dollars being poured into it by people with very deep pockets.

IBM and Novell want everyone to be able to buy and use their software. Therefore, it should run on an open platform, a platform that everyone can use for free. This is a perfect example of a “complementary good” from basic economics. If they put in some effort in improving the Linux environment, more people will use it, and more people will be able to buy their products.

Mark Shuttleworth is an even simpler case. He’s just a rich man who decided to devote a lot of money to developing a good Linux desktop environment and improving the open source development process. If you’re starting a Linux-based company today, it’s almost as if Mark is paying for someone to do some of your work. You can’t lose (unless you’re a direct competitor like Microsoft or Apple).

My point is, if you use Linux, there are many people on your side. If you use a proprietary OS, you probably have only the vendor to rely on.