Cool Tech Reviews

Charlie Demerjian at The Inquirer:

Every so often there comes a genre-bending product, and Tyan has one of those on its stand at CeBIT this year. It is called the Typhoon PSC or Person Super Computer, and is aimed at the scientific and HPC set. Typhoon PSCs come either Opteron or P4/PD flavours, and brushed aluminium or black finish.

The point of this box is not to make an ultimate gaming rig, it is meant to take what used to be the domain of a data centre and move it to under the desktop.
…
There are two PSCs, the Opteron one is the B2881YDS4T and the P4/D model is the B5160YDS4T. Both are basically four dual socket blades in a box with all the necessary cables and attachments in a nice aluminium case. It also has little wheels and two carry handles which I have personally tested to get the rear shot here.

The specs vary between the models a bit, but there are eight sockets and up to 16 cores on either one, depending on how you want to set up the system. The Opteron can handle 64GB of memory, 32 for the Pentium. There is one S-ATA drive per blade, and they are coupled through GigE on the back. Each blade has an independent 350W PSU for a total of 1400W, which fits nicely in the 1500W most circuits can provide.

Yikes, that’s still a pretty heavy load! If you click through to the article, you’ll see the photo mentioned and it looks like a box of fans, although Demerjian says it is very quiet. There are more details in the Tyan press release and the B2881YDS4T and B5160YDS4T product pages but the most startling thing is that the whole package is 14″ x 12.6″ x 26.7″.

No prices are mentioned, but if I really had a project in hand that required heavy computing I’d certainly give this a look.

One of my pet theories is that if a PC manufacturer would only turn out units with slightly “out of date” technology instead of competing on specs with everyone else, they could offer some amazingly cheap but still very functional PC hardware. It looks like Fry’s Electronics is doing just that as Loyd Case reports at PC Magazine in Your Next PC Will Cost $159:

Holy tightwads, Batman! A better PC than what you’re running costs less than a pair of designer jeans? What’s happened to the computer industry?

Were GQ magazine to design a computer, it would sport a Gucci leather jacket and stroll in slick Prada loafers. It would also cost eight, maybe nine thousand dollars. But when Fry’s Electronics designed the GQ system, it wasn’t thinking of luxury linens and leather. It wanted something cheap. The surprising thing is that the GQ (short for “Great Quality,” by the way, not Gentleman’s Quarterly) turns out to be a powerful PC. It’s low-cost, in other words, not high crap.

Hit the link for the full review but here’s what you get for $159:

What $159 Buys You…

• COMPUTER In addition to the 1.67-GHz AMD Sempron chip (and integrated graphics and 6 channel audio), you’ll get four USB ports, Ethernet, and an AGP slot.
• MEMORY Only 128MB of RAM, which is barely adequate, really. Replace it with 512MB for around $40.
• 40 GB hard disk.
• SPEAKERS They’re included, but they sound awful. Really, these speakers are just terrible.
• Generic modem
• KEYBOARD AND MOUSE A surprisingly responsive keyboard and generic ball mouse round it all out.

…& What It Doesn’t

• NEW TECH You’ve heard of PCI Express, SATA, and dual-core, but you won’t get them here.
• LCD MONITOR Nope. Not a chance. In fact, there’s no screen included, period. (Fry’s offers a companion 17-inch CRT for $119.)
• DVD BURNER There’s no DVD recorder, not even a CD burner. But there is a fast CD-ROM drive.
• SOFTWARE Nothing from Microsoft here, but to be fair, the Linspire OS is pretty decent.

As far as the hardware goes, the memory is obviously short, even for Linux, but while you can argue about some of the other features it seems to be a pretty capable machine. Case was impressed with Linspire (which comes with OpenOffice) but obviously that’s going to be a matter of taste. If you have to buy Windows XP, it’s likely at least another $70 on top which rather spoils the deal but gives you an idea of where the money on a PC is going.

However, it turns out that there’s another alternative. Microsoft’s Chris Sells saw the PC Magazine review and decided to give the bargain PC a try. Apparently the $159 PC was a limited quantity offer, but he came up with an equivalent one for $171 and added $100 for 1GB of RAM and $95 for an ATI Radeon 9550 AGP card. The net was a $366 PC and he installed a beta of Windows Vista on it!

At 12:04am, I started the Vista Feb ‘06 CTP installation. At 12:44am, I was running Vista, it having recognized all of hardware (except the sound device) from my $366 PC, including enabling those cool “glass” effects and the nifty animations, integrated search and all the neat things you’ve read about in the Vista reviews.

I know I work for “the man,” but even so, I’m seriously impressed. The install was fast and seamless. The performance is way better than I thought it would be. And the little UI tricks are fabulous. I can’t do any media stuff ‘cuz my audio device wasn’t recognized, but it was cool when I tried to play video and a DVD, that the Vista Media Center UI came up (my complete home entertainment needs are served with a coupla TVs, a Media Center PC and an XBox).

I know, I know, I got the OS for free, but come on! It’s still beta and it runs great on my cheapo PC!

Hit the link for all the details including a certain amount of disbelief from some commenters who have had performance problems with the Vista beta. I’m not recommending buying an inexpensive PC in anticipation of Vista, but it looks like there truly is some capable bargain hardware out there if you are willing to shop around.

The Pentium Chronicles : The People, Passion, and Politics Behind Intel’s Landmark Chips by Robert P. Colwell. Published 2006 by John Wiley & Sons. ISBN: 0471736171.

Summary: Dr. Robert P. Colwell was Chief Architect of Intel’s wildly successful P6 processor and the The Pentium Chronicles is an anecdotal account of its development from someone who was in on it from the start. While portions of the book will only appeal to those in the chip industry, there is something there for everyone about life in a large organization.

Review: Dr. Colwell’s account of the development of the Intel’s “P6″ processor (which appeared to great success as Pentium Pro, Pentium II, Pentium III, and other names) will inevitably be compared to Tracy Kidder’s The Soul of a New Machine, but the comparison is imprecise. While Kidder provides a start-to-finish narrative of the development of a minicomputer at Data General in the 70’s, Colwell recounts a selection of high (and low) spots along the way to the completion of the P6. Both projects were huge bets for their respective companies that were ultimately successful and both teams seemed to have suffered the same sort of “post partum depression” when they were completed.

Colwell describes starting out with a small team and the perils of each step along the way from concept to final production of a large chip technology project. The technological and development process insights are interesting to me and likely anyone else with experience in the industry. One is the guidance of design with performance information based on real data which is often amazingly absent from many technical projects. While many designers tend to shoot from the hip based on their “intuitions,” Colwell describes how his team trained their intuitions in designing the unprecedented features of the P6 using performance analysis tools they wrote themselves and exploiting real instruction traces.

If that is too techie for you, there are plenty of items of more general interest about the computer industry (e.g. a meeting at Microsoft turning into a shouting match between the Windows 95 and Windows NT teams or an off-the-cuff remark at a press event being blazoned across the headlines) and about life in a large organization. What do you do when when rival teams make bogus performance claims; or when there’s a corporate directive replacing satisfactory tool systems with buggy “approved” versions; or when implementing the latest corporate slogan campaign has your engineers in revolt? Colwell has no magic answer other than to shoot straight and tell the truth which is the best advice anyone can give.

However, while it never seems like it at the time, all successful technical projects are a brief magic moment which passes all too swiftly. Organizations inevitably change which places a premium on another key skill which is knowing when to move on. Colwell eventually did exactly that and explains why. The changes that have taken place at Intel over the years are likely one of the reasons it is facing its current difficulties.