Slashdot Mirror

I've been looking at RAID setups over the last few days. I'm just yer average geek who doesn't have unlimited funds, but who likes speed, and really really wants to not lose data.

What I'm thinking of getting is 4x28GB 7200RPM IDE drives (4x$199), putting most of the first three drives into a software RAID 5 using a 2.2 kernel and the 0.90 md patches, and using the fourth disk as combination hot-spare, scratch space, and incremental backups for my home directory and other parts of the filesystem which a) change frequently and b) I want old revisions available in case I trash something important.

I have four IDE channels on my motherboard, so I can put each disk on a separate channel, and make the seldom-used CDROM and CDR drives slave drives. So contention between disks shouldn't be a problem.

I'm not sure about the backups. This still leaves me vulnerable to accidentally blowing away large quantities of un-backed-up data, but it's stuff I don't change much, and might have copies of on CDR. (Like the MP3 collection.) And I'm not sure what the alternative is.

Tape drives that have anywhere-near-enough native capacity seem ridiculously expensive. RAID with more devices and SCSI disks is also too damn expensive - 100GB of SCSI disk is like $2500, way out of my budget.

Setting up two drives in a RAID 0 and periodically copying the same data to another two drives is another possibility, and protects me against an "oops" that I notice immediately, but doesn't do anything about a bad mistake that I notice the next day, and ensures that I'll lose my day's work if one of the primary drives dies.

Thoughts? I've never set up something like this, and could use all the advice I could get.

Another question: What do people think of hard drive coolers, like the CoolerMaster? ( Ars Technica review) Useful addition to preserve drive life, or a waste of money and bays?

Thanks,

Alan

Let's take a look: Ars Technica

Headlines:

Instant Message Wars heating up
Making history? PALM (spinoff of 3com)
Not exactly nipping to buy... (SGI and Cray)
Poll: Monitor Size
Seagate Barracuda ATA II @ SR and RAID hack follow-up
No BeIA on Intel's WA
Sexiest Geek Alive
AMD price cuts
Trade your games
Tim O'Reilly on 1-Click patent

Wow, we finally found one story that was posted on Slashdot also. Thanks for playing. I've been reading Ars Technica for a long time, they certainly aren't a Slashdot rip off.

He obviously don't want a dock and is ready to drop the baby with the bath. Having used macintoshes since 1986 and NeXT/OPENSTEP/OSXS machines since 1991, I would not trade the dock for the MacOS alternatives. It is amusing to see someone defending the total failure that windowshades are.

Admitedly, the DP3 dock seems to suck, but nothing that can't be corrected. He seems to have a problem because the dock is not the grand-unifier-launcher that he turned the applemenu into, but forgot that such an app is easy to code. Doing a system-wide dock for miniaturised windows is not that easy because it needs to be tighly integrated with the system.

At the end of the article, he rants about the disparition of the apple menu because that was the place where he found most of his desk accessories. Well, I really prefer having them in an equivalent of the Preferences.app, and keeping it on the dock (On NeXT, Preferences.app icon was the system clock. Double-clicking on it put the preferences panel, with item lookable by _icons_, instead of an un-understandable alphabetical list of items in a menu)

Btw, I really miss the OPENSTEP4.0 alpha dock that never saw the light.

(Why do I post that ? this article is old news already...)

Oh, no. It's a Good Thing. MacOS X's display system is a generation beyond Quickdraw. There was an article on Ars Technica that summed it up nicely that you should check out if you're interested. The only thing missing from the Quartz graphics system (which may be possible to add later) is the network protocol aspects of NeXT's DPS system. (Meaning you can't currently export apps to another display.)

X is really a primitive graphics system. The only reason to continue with it is its networked capabilities and for backwards compatibility with older UNIX applications. The Mac and Windows graphics APIs kick X in the teeth both by itself and combined with Motif/GTK. Trust me on this.

This is according to Ars Technica:

It seems as though upon boot-up, the user is presented with a logon window not dissimilar to that in NT, xdm, etc.). As it turns out, if you press Ctrl-Alt-Del, a message pops up saying:

This is not DOS!

--

But Siracusa doubts that there will ever be a release of OS X for x86, and I'm inclined to agree - this would seriously undercut Apple's hardware sales, if nothing else. Apple, I'd think, would want to leverage the cool factor of OS X against the cool factor of its hardware for a pretty hefty marketing beast.

Chris Tembreull
Web Developer, NEC Systems, Inc.

My opinions are my own, and nobody else's.

But Siracusa doubts that there will ever be a release of OS X for x86, and I'm inclined to agree - this would seriously undercut Apple's hardware sales, if nothing else. Apple, I'd think, would want to leverage the cool factor of OS X against the cool factor of its hardware for a pretty hefty marketing beast.

Chris Tembreull
Web Developer, NEC Systems, Inc.

My opinions are my own, and nobody else's.

Apple's solution to the tiny-icon problem in OSX seems to be where many GUI designs will be heading in the next few years. Rather than using static bitmaps the screen is rendered in vectors, so all elements of the screen are scalable, and even dynamically deformable (lending to nifty tricks like a window appearing as if it is peeling forward off the desktop behind it like a post it note, with the contents transformed accuratley). Ars Technica did a great writeup on it, I reccomend the read if you are at all interested in GUI design.

Ar stechnica I never did post in there. But I like to read the Lounge, which just broken down :)

A bit of a Windows Pub, but the forum allow livelier html tags. I still don't understand why there aren't people start slashdotted.org yet, host bitter submitters' rejected articles and all.
I like /. keep the way it is, don't "deja" youself rob.

CY


/_____\
vvvvvvv../|__/|
...I../O,O....|
...I./. .......|
..J|/^.^.^ \..|.._//|
...|^.^.^.^.|W|./oo.|

Actually, I can see a few things:

• Transparency in the menus and dialogs
• Anti-Aliased text
• Sharp-looking icons because they are vector- and not pixmap-based, and will scale to any size and still look good.

This is because of Quartz, a new vector-based graphics layer upon which Aqua is built.

Read the Ars Technica article for more info.

Your point is moot And don't even bother responding if unless you read the artical

[ c h a d o k e r e ]

I worked at CompUSA in Pittsburgh as a Mac salesman one summer when I was in college. Even then, trying to explain how a 25 MHz 68040 was really faster than a 40 MHz 68030 was a tough point to get across (same goes for a 60 MHz Pentium and a 66 MHz 486). The best way to describe it is probably this: all MHz are not created equal. That is, the clock speed merely measures the number of processor cycles per second. It says nothing about what the processor does with them. To really compare two different processors, you have to run some kind of a benchmark on them (that is, do the same operation on both and see who is faster). I don't have the numbers in front of me, but it is entirely possible that a 500 MHz G4 is twice as fast as a 800 MHz Athlon. It all depends on how efficient the processor is in its execution.

As to the Macs rather small market share, there are other factors involved. The presence of PC clone makers forced all PC vendors to lower the prices on their PCs; PC hardware tended to be much cheaper than equivalent Mac hardware. On the flip side, the Mac was much easier to use: true plug-and-play (which Windows still doesn't quite do right), a graphical interface that was designed from the ground up to be that way, etc. If you wanted true ease of use and low maintenance, went Apple's theory, you could have it, but you had to pay; as there were no other Mac vendors, you were a slave to whatever hardware Apple felt like making, at the price Apple felt like selling it for.

Needless to say, most people took the checp PC route, rather than the expensive Mac route. Those that took the Mac route, however, generally became die-hard loyalists, not unlike the Linux zealots we have all seen post here (I am a Linux user too, so don't flame me about being a Linux hater). Mac users are very enthusiastic about their platform of choice, sometimes to a fault. They sometimes forget that there is a fine line between advocacy and harassment.

Just a few thoughts on why the Mac (and the Mac community) is the way it is. And, of course, why the G4 really is that good! :)

Mark Pfingstler

You can find that review on Ars Technica.
--
Whether you think that you can, or that you can't, you are usually right.

Ars Technica has an interesting analysis of the Crusoe chips. One of the ideas mentioned is that Crusoe represents Transmetas solutions to a specific set of problems. IMHO, Transmeta isn't saying the Crusoe chipset *couldn't* be used in a desktop machine, just that you haven't seen their desktop solution yet.

Read RISC vs. CISC: the Post-RISC Era at Ars Technica and gain True Wisdom. ;-)

dufke

-

Does KDE like the sound of vector-based icons?

The Ars Technica article says:

The dock takes advantage of Quartz's ability to apply vector transformations to bitmapped images.

...

Aqua appears to use (via Quartz) a form of bicubic interpolation to scale the icons.

(emphasis mine) which seems to suggest that the icons are bitmaps, not "vector-based", and that the icons are scaled up by interpolating between the points.

I really like the idea of a vector based GUI. I think that looking at more "organic" shapes is easier on the eyes than looking at all those squares and blocks.

If by "'organic' shapes" you mean the jelly blobs of Aqua, I don't think they're intrinsically tied in any way to a "vector based GUI". In fact, the Ars Technica article says as much - it says of the gelatinous buttons, "On the other hand, nothing we've discussed so far can't be duplicated with a second generation display layer."

Why use a vector based windowing system on a monitor that can only display squares and blocks efficiently.

Well, I'm not sure what's "vector-based" about Display PDF; PDF is a PostScript-like language (not entirely surprising, considering who invented PDF...), so it might be "vector-based" in that a path might be made out of lines - but the PDF spec says that a "path object" is "an arbitrary shape made of straight lines, rectangles, and cubic curves", so it's more than just "vectors" (in the sense of "lines").

The moment a vector based windowing system becomes usefull even unavoidable is when we have "analog" monitors as in "monitors that can actually display vectors" as opposed to the current "monitors that diplay bitmaps"

To put it bluntly, I would not, if I were you, hold my breath waiting for that to happen. I suspect it may be easier to make raster CRTs (you just have to make the beam scan left to right, and then scan the next line below it, and..., rather than being able to steer it arbitrarily), and the display on my desk isn't even a CRT - LCD displays (which, as far as I'm concerned, rule) are intrinsically digital monitors that display bitmaps.

I have the impression that, these days, rasterizing vectors is pretty much a solved problem.

IMHO this vector based windowing system is just a marketing buzz-word induced waste of cpu-time.

The reason for a "vector-based" (or, as I might be inclined to say, "path-based") windowing system, at least as presented by Ars Technica's article, is that "vector" transformations (which, I suspect, are transformations on vectors representing points, i.e. the vector from the origin of the coordinate system to the point, not vectors representing lines from one arbitrary point to another) can be applied to the PDF description of something being drawn.

Much of the other advantages that article ascribes to a "vector-based" windowing system, such as the stuff Aqua does with transparency, have, I suspect, little if anything to do with PDF being "vector-based" (or "path-based").

Besides, I didn't see any mention of "vector-based graphics" on the Graphics page of Apple's stuff on MacOS X; "vector-based graphics" may have been Ars Technica's term - as suggested above, I'm not sure I'd use it, and that may be why Apple doesn't appear to be using it there, either.

Since the kernel is based on BSD, will OS/X use a relatively standard Unix filesystem? In the past MacOS had that wacky system of a "data fork" and a "resource fork". Does anyone know how that will be bolted into a Unix environment?

The kernel is based on Mach, not BSD.

There's some information on the filesystem issues in an earlier Ars article. Check it out (Skip to the section on "meta-information" if you're in a hurry.)

I'm not sure I like the new UI terribly much myself, but there is a hell of a lot of impressive technology behind it. I'd have to actually use it to know for sure, but it looks like too much gaudy eye candy for my taste. I'm hoping there's a Platinum Appearance that one can switch to if one desires.

There's an article on ars technica that another poster provided a link to, which goes into all of the swank new technology behind the eye candy. It says that PDF is a superset of PostScript, which isn't exactly true. PDF is a subset of PostScript with some new onscreen features added like forms and hyperlinks. Eventually PostScript and PDF are going to be pretty close to merged - Adobe's PostScript Extreme engine is a PDF RIP (PDF to print, with no PostScript in between) and a PostScript to PDF converter.

There are a couple things about Display PDF that aren't mentioned in the article that are extremely cool. GDI and QuickDraw are the current systems for onscreen display on Windows and the Mac OS, respectively. On Windows or the Mac, if you copy anything other than text from one app to another, you are copying not the original file, but GDI or QuickDraw commands. And most non-desktop publishing apps use GDI or QuickDraw to print, which causes a couple of problems. GDI and QuickDraw are both RGB, which throws color off completely if you copy a CMYK TIFF from Photoshop into Quark or copy an EPS with spot colors from FreeHand into PageMaker. And GDI (and to a lesser extent, QuickDraw) is not at all friendly to PostScript printers.

PDF (as of version 1.2) understands CMYK and it understands spot color channels. PDF is friendly to non-PostScript and PostScript printers alike. Which means that non-desktop publishing apps will suddenly print much nicer to PostScript printers, and it means that copying and pasting from one desktop publishing app to another just may stop being the Extremely Bad Thing that it is now.

Oh, and because Mac OS X is based on NeXTStep is based on BSD, for the first time I'll be able to do my desktop publishing on a real OS. No more stopping to allocate more RAM to FreeHand or less to Quark; no more crash and reboot.

In the same way that I tolerate the bright gaudy blue of the G3 on my desk at work, I'll probably learn to tolerate the jelly bean buttons and the jewel-bright scrollbars of Mac OS X.

I'm not sure I like the new UI terribly much myself, but there is a hell of a lot of impressive technology behind it. I'd have to actually use it to know for sure, but it looks like too much gaudy eye candy for my taste. I'm hoping there's a Platinum Appearance that one can switch to if one desires.

There's an article on ars technica that another poster provided a link to, which goes into all of the swank new technology behind the eye candy. It says that PDF is a superset of PostScript, which isn't exactly true. PDF is a subset of PostScript with some new onscreen features added like forms and hyperlinks. Eventually PostScript and PDF are going to be pretty close to merged - Adobe's PostScript Extreme engine is a PDF RIP (PDF to print, with no PostScript in between) and a PostScript to PDF converter.

There are a couple things about Display PDF that aren't mentioned in the article that are extremely cool. GDI and QuickDraw are the current systems for onscreen display on Windows and the Mac OS, respectively. On Windows or the Mac, if you copy anything other than text from one app to another, you are copying not the original file, but GDI or QuickDraw commands. And most non-desktop publishing apps use GDI or QuickDraw to print, which causes a couple of problems. GDI and QuickDraw are both RGB, which throws color off completely if you copy a CMYK TIFF from Photoshop into Quark or copy an EPS with spot colors from FreeHand into PageMaker. And GDI (and to a lesser extent, QuickDraw) is not at all friendly to PostScript printers.

PDF (as of version 1.2) understands CMYK and it understands spot color channels. PDF is friendly to non-PostScript and PostScript printers alike. Which means that non-desktop publishing apps will suddenly print much nicer to PostScript printers, and it means that copying and pasting from one desktop publishing app to another just may stop being the Extremely Bad Thing that it is now.

Oh, and because Mac OS X is based on NeXTStep is based on BSD, for the first time I'll be able to do my desktop publishing on a real OS. No more stopping to allocate more RAM to FreeHand or less to Quark; no more crash and reboot.

In the same way that I tolerate the bright gaudy blue of the G3 on my desk at work, I'll probably learn to tolerate the jelly bean buttons and the jewel-bright scrollbars of Mac OS X.

I must say, i learned a lot from the report the Arsificial Intelligentia over at arstechnica.com put up.

Check it here.

Its got a great deal of info on how MacOS X and Quartz are a 3rd generation GUI, relying on vectors, and a great deal of pdf technology to speed things up. This decreases the amount of power needed to run a transformation like the genie effect by great amounts.

Good stuff.
/nutt

I must say, i learned a lot from the report the Arsificial Intelligentia over at arstechnica.com put up.

Check it here.

Its got a great deal of info on how MacOS X and Quartz are a 3rd generation GUI, relying on vectors, and a great deal of pdf technology to speed things up. This decreases the amount of power needed to run a transformation like the genie effect by great amounts.

Good stuff.
/nutt

Daniel.

The print press won, but plastics (woohoo!) is mentioned.

--

I read a review of a midtower that can be made into a desktop unit by changing the drive bay cage and cover plate(see link to story below). It is $110.00 online from the manufacturer. Review at arstechnica.

...saying that intel translates x86 instructions before executing them. This is true for the 16bit x86 instructions, but not for 32bit...

Native 32-bit instructions were executed ages ago, on x386/x486. Pentium already has RISC-like core and Pentium Pro/Pentium II added branch prediction.

BTW, I wonder if there will be Native Crusoe Support for GCC? Despite all this Code Morphing Coolness, native Crusoe VLIW instructions optimized by real compiler should run much faster. GCC++ knows all about program workflow and Code Morphing will tune (rearrange predictions?) it at run-time.

Furthermore, Transmeta can invent its own extentions like Intel's MMX and implement it real soon. Image tweaking the CPU to 3D card drivers! :-)

IMHO this technology will beat Intel in the long run.

Ars Technica is cooler than /. anyway. They're interested in technology, not politics.

--

Ars's resident CPU-meister, Hannibal, has taken an in-depth look

Bravo /. that is the kind of stuff I wanted to read about the chip.

Shouldn't that be bravo ars technica?

There article on the K7 was great, btw...

Ars Technica has an _excellent_ review of SCSI drive technologies. I found it extremely useful.

Emulating CISC based hardware is vastly easier than achieving acceptable performance emulating RISC based hardware. With CISC based hardware, each instruction does a lot, and might take several clock ticks to execute. The Transmeta emulator can come along and translate the instruction into equivalent Crusoe instructions, and achieve roughly comparable performance.

In principle, this is correct. In practice, Intel engineers made a deliberate design decision in the P5 and P6 generations to spend time making the RISC-like instructions fast (using deep pipelines, register renaming, and other magic), and the CISC-like instructions slow. Implementation of rarely-used, complex CISC instructions was simply too much of a performance drag.

As a result, modern optimizing compilers for x86 hardware generate very RISC-like code: mostly loads, stores, branches, and ALU ops. Look at the x86 assembly emitted by gcc -S sometime. Sensible compilers these days simply don't emit REP MOVSB or other behemoth CISC instructions left over from the 386 days.

Any putative performance gain from translating "complicated" instructions to many optimized Crusoe instructions is probably marginal. Hence, your argument about the relative "inefficiency" of Crusoe with regard to translating from RISC native code is incorrect.

BTW: for the interested, check out this Ars Technica article, which provides a fairly accessible discussion of why the RISC/CISC distinction isn't a very useful means of characterizing processors anymore.

~k.lee

If anyone reading this article hasn't already, check out some of the posts about this on the "Open Forum" at http://www.arstechnica.com. Some of the more interesting comments on the Apex player mentioned an inability to do low bitrates (less than 32kbps), an 8-character track name limitation on the unit's display, and weird problems with audio sync on certain DVDs.

Still, the overall consensus was that the unit was a bargain despite these limitations. Of course, I recommend you read and decide for yourself before you throw your money at Best Buy employees. :-)

The links are from an Ars Technica blurb

Some people have been able to find them, others haven't. Some have problems, others don't, etc. etc.

Go over there and scroll down on the first page a little. Some of the guys have gotten their hands on this little puppy and had good things to say about it.

It's Arsward. It comes from Ars Technica. Ars Technica is latin and translates roughly into "the technical art."

It was the Celeron. Whatever intel did to make such damn high quality C300A's was great. They used technologies from P2 production and yet left the Celeron at 1.5 times less Mhz than the P2.

This left us with only one thing to do. OVERCLOCK. They tried to lock the bus to make it hard to do, but that didnt matter. With the advent of really nice 100Mhz FSB boards out there like the BH6 which was rock solid for overclocking, things just took off.

Now that Celerons up to 533are now being produced, with Intel 'doing the OC for us' by having them at 100Mhz bus now, those halcyon days are over. The C300A was just the BEST at overclocking. No need for any special cooling devices for me, and among a dozen friends that tried it, 11 of us succeeded - the other guy went and got a replacement that worked. The power/$ because of this is something the Athlon wont even match til prices fall this summer.

Or ARE the halcyon days over? According to this slashdot note about this article, there may be more such days ahead during the early release of the intel flipchip 500 and 550e, before they start streaming into high quality high speed high price chips, and lo end ones which wont OC. GET THEM AT THE START of the cycle.

If you arent running a server, the leeway intel has given for overclocking is just too large to ignore. If you consider it, the Athlon, while a technological marvel and all for the elegant solutions it employs to be compatible with a stupid x86 design (see this ARSTechnica comparison G4 vs Athlon) just does NOT give the same bang for the buck if you check it all out. (And REALLY, the Alpha is still an amazing chip, 64 bits and all, and heavily underused - and its years old.)

So if price/performance and x86 compatibility are all that matters, the Celeron300A was the best CPU deal of the year.

Math

Since we are talking about Ars anyways, they have a good review of the mouse on their site that should answer your questions:

http://arstechnica.com/r eviews/4q99/msmouse/msmouse-1.html

Ars Technica had this link to an article about ram technologies. RAMBUS didn't look so good there.

Go check out atacom. They have some great full tower cases, I bought one from them to use on my BP6, the BP6 is definatly the best Socket 370 board so far. If you have that board you may be interested in the Socket 370 Celerons @ 533 (8.0x66) are comming out soon. They will be the last 66MHz Celerons, finally, Intel will have to make a Celeron with a 100MHz FSB. info about the Celeron 533.

I noticed that Woz referred to RISC designs as an advantage for the Mac..I would have too, if not for the ARStechnica article about RISC/CISC...which makes me believe that we are in a post RISC/CISC era. The G4 has a RISC history, but it has more instructions that some CISC chips! And I'm a Mac user to boot..

http://arstechnica.com/cpu/4q99/risc-cisc/rvc-1.ht ml
Risc vs Cisc

Now we just need the motherboard vendors to jump on the bandwagon. We're off to a decent start but more choices, and easier availability of systems and parts is needed.

If these are going to be used as phones, why not overclock the processor to 900MHz (or 2.4 GHz!) to match the frequency of cordless phones?

This would settle those occasional comments about CPU frequency interfering with radio frequency and vice versa, since they don't.

AnandTech, Sharky Extreme, Adrenaline Vault, and Ars Technica seem to have screenshots of a new overclocking record every week or two. I believe it currently stands at ~1300MHz -- is 2.4GHz unreasonable to think?

--

Does this guy know anything about architecture?

In a word, yes. He knows an awful lot. If you don't know what "Post-RISC" is talking about, why don't you read his article on RISC vs CISC like he suggested. Here, I'll make it easy for you: RISC vs. CISC: the Post-RISC Era

Of course, it also looks like you didn't even bother to fully read this article. Hannibal is hardly anti-mac, anti G4, or any of that. He concludes that he prefers the G4 over the Athlon (and the Alpha over both of them). Give me a break...


--------

Considering the common notion that the G4 is fast, multiple G4's are obviously the next logical step. The problem becomes: what would I do with G4-MP???

The MacOS seems infinitely capable -- why must it do all of them so poorly and/or primitively? MP support on a Macintosh works out to be a few specific Adobe apps crudely hacked to run two parallel threads on separate CPU's. The OS doesn't natively support any kind of MP, and 90% of Mac apps simply 'stay at home' on CPU 0. If there were more than ten apps used on a Mac (Adobe software, M$ Office, Quark, and Nutscrape), I'm sure that percentage would be higher.

Linux SMP is coming along (but admit it, it's not even up to pace with NT, let alone BeOS), but Yellow Dog only has half-assed support for the G4. Then there is the problem of getting it from kernel 2.2 to 2.4 in the next couple of months...

Apple's OS X is a way off, and given the hardware used in their systems since the introduction of the iMac, would you be able to find any other UN*X that supports bizarre foreign hardware like USB keyboards and mice?

Just imagine a bitchin' Beowulf cluster of MP G4's with a functional OS! =)
--

People keep talking about problems with heat dissipation. One possible solution to this might be to submerge the entire computer in mineral spirits. I saw a picture of a computer that was set up this way once, in order to provide cooling for overclocking (sorry, can't find the URL, but search Ars Technica for a link). Mineral spirits are nonconducting, and have been used in electrical transformers. If you put the computer in a ridged metal box, you should get okay heat transference from the box to the outside. You could even put heat sinks on the outside of the box if need be.

I don't knwo what this would do for other environmental issues like cold or moisture.

I have no idea if this would work, but I thought it might be helpful, and it shouldn't be too expensive.

Jon

Very true, the primary bottle neck in computers these days is memory and network latency. I think that the advances IBM is showcasing here will really pay off though in decreased power requirements which are becoming increasingly important as embedded devices appear. and the combination of Processor and Memory is an extremely attactive option as it relieves requirements on the bus.

intersting work is being done in this direction under the Processor-in-Memory (PIM) project.

Another mechanism to decrease the effect of this memory latency is to use large numbers of low-level threads (often automatically generated by the compiler) to mask latency. By decreasing the context switch penalty to a single cycle (or less with interwoven threads) and then switching on every cache miss substantial benifits can be made. One example of this is Tera computing MTA architechture. For certain common simulation tasks the 4-processor TERA machine blew away a multi-node Origin and Cray computer according to This NASA report.

Also, Sun's new MAJC architechture uses threads to mask latency.

Interwoven threads (where the processor switches thread every clock cycle) has the benifits of removing branch and data dependancies from a processor pipleine, thus removing the need for processor complexity like data forwarding, speculative execution, and the like. An example of this technique can be found a the TIPSI Project.

Second, arstechnica has a follow up article about a british venture capital company that's working on something even better: think multiple terabytes on a single cd!

"God does not play dice with the universe." -Albert Einstein

Maybe a PC manufacturer will build a custom web browser in order to use the BeOS?

If they are really looking to cut the cost of the machine, linux is probably the way to go...

Cool.

I mean, they're making CPUs so darn complex these days it's hard to appreciate the real work that the processor is doing. Example: I haven't got a clue what the PIII under my hood is doing, and I don't want to, because it would involve so many complex subjects that my mind would either explode or doze off. Now the PISC as described in that article is my kind of chip (or maybe it's not a chip if I try to build it, but same idea of CPU). So what if it runs at 5 MHz? That could be fixed--just speed up the underlying circuits. And at 5 MHz, my brain might actually have a chance at keeping up with it. Skimming the page was enough to give me the idea that "I could build this thing," even though I have very little experience in actually building electronic circuits. I'm not a chip engineer, obviously, or an electrical engineer, and neither is the average computer user. The average computer user, however, might be interested to know that they actually have a shot at figuring out what that magical black (or grey) box in that case is doing.

Demystify the processor. But don't call it Pathetic. I'd call it Understandable. Unfourtunately it's even harder to pronounce UISC than CISC or RISC, so make it Technically-understandable Instruction Set Computer, or TISC (cm) (Cool Mark).

Kenneth Arnold

PS - If you are going to have a shot at looking at some neat technology, I thought that the K7 Article also on Ars Technica was pretty cool. It was well written and provided a good metaphor. Still, it barely scratched the surface at telling me what is really going on down there.

PPS - I could have said more, but my hard drive is making funny noises and I better investigate.