Diagonal routing has been used to some effect on PC boards for a long time, but it's been hard to find enough wires going between the same sources and destinations to get a big group (river) together, so on a given layer they tend to cut off the vertical or horizontal routes that other pairs need. If you can arrange related groups of circuits to be adjacent ro on the same vertical or horizontal axis then diagonal routing isn't needed. It can certainly help time-critical signals in a sort of kludgey way if they're going to be wasting most of M4/M5 anyhow.
Me too. I couldn't believe that hacker humor could make it into Saturday morning cartoons. The concept of Phong being Phong shaded (hey, where's Goraud? He could be called Henri and have a French accent) was enough to get my lousy butt out of bed on onto the couch where it belonged. ReBoot and The Tick reignited my interest in cable TV and changed my life. I hate them.
Intel, Microsoft, Compaq and other heavyweights make many strategic investments in companies developing promising technology. Since they're in business they expect to eventually make money on their investments, which is normal for public companies. At the time of Intel's investment Rambus stock was trading much closer to $10/share than it is today, and Rambus had no customers in the CPU space, so it was a much riskier deal. Rambus' current price (and Intel's current paper profit) is due more to "irrational exuberance" in the stock market than any old-school valuation. The slimey business practice of getting competitive advantage by investing in partners is common in high-tech, but is far more common (in fact normal) in Germany; I'm sure that Dr Tom could also list the firms from his homeland, starting with all the banks, then working through the heavy industries on his way to media firms which he has similar complaints. Rambus will win because it produces higher bandwidth at lower cost (power, pincount, etc) than SDRAM, or it will fail because the RF engineering needed for 800+MHz traces can't be done economically by the time SDRAM busses can deliver the same data rates. The investments made in Rambus by its semiconductor partners will have little say in the long run, though they may have funded the firm for long enough to give the new idea a reasonable chance in the marketplace. elbee
At least one member of Transmeta's hardware team used to work at DEC WRL, which issued that famous technical note on electric pickles. Transmeta could have hired him for his pickle know-how, which when added to Linus' lutefisk background may give us another hint towards Transmeta's plan for world domination. ellbee
Swiss banks accounts come in many flavors; the most famous is the "numbered" account where only one bank official keeps a (non-compterized) record of who actually has power over the account. Most Swiss bank accounts are just like an account anywhere else, except they're covered by Swiss banking laws which make it more difficult for investigators to gain access to records. Over the past few years Swiss banks have become much more cooperative with international law enforcement (especially drug laundering), though they're much more private than American banks. Unless you make the transfer with cash US authorities should be able to follow Internet transactions even if they're into foreign accounts.
Last I heard AV had over 250 million non-duplicate web pages indexed, and knew of (but hadn't indexed) several hundred million more. One thing mentioned on the CNBC interview today is that CMGI will continue growing the index, and hopes to have "the entire web" indexed over the next few years, which they think will be over one billion pages.
I believe that the peak query rate is over 600/s, which means completing a query every 1.5ms. Good thing computers are getting faster, memory is getting denser, and 64-bit addressing is a reality.
There has been similar UI work before at Xerox PARC and notably the Fisheye View work by Marc H. Brown at Compaq's (formerly Digital's) Systems Research Center.
I arranged for three people (including myself) to get 384/384 service from Concentric/Covad in the Bay Area about eighteen months ago, and we've been pretty happy the result. My install was relatively smooth - PacBell (the ILEC) came out and provisioned the line, shorting just two of the six voice lines running into my apartment building which took only two extra hours to straighten out. Concentric emailed me IP adresses, DNS info, etc. Covad arrived with a preconfigured modem. We hooked it all up and it didn't work, but after only two modem reconfigs it came up and has been ticking ever since.
Most of the time I get close to the advertised bandwidth, but it's clear that they have to work on congestion. Packet loss to local, well connected Bay Area hosts regularly show 10% packet loss, and traceroute points the finger squarely into Concentric's IP network. Every few days the network freezes for a few minutes and all connections drop, but I've no idea where to point the finger.
Not everything has been smooth. My customer rep left witgh no notice and I haven't been assigned a new one. I couldn't find a 24 hour trouble line in the phone book or by calling Concentric HQ. Their accounting department is so screwed up that at one point I found they hadn't sent a bill out for five months. But the biggest problem is that they haven't kept their rates competitive - we're keeping the three Concentric lines, but the next dozen from work are going to PacBell which offers simlar service for 1/3 the price.
We currently have several dozen people with ISDN at home. The plan is to evaluate the next batch of DSL users and decide if we want to go further down that path.
Back in to old days (you young whipper snapper) reading Usenet meant reading ALL of Usenet. But as its popularity grew the volume grew and the news groups proliferated and the signal to noise ratio dropped - and that was before alt.* and AOL./. can continue to be truly worthwhile only as long as the posts are relevent and the volume stays below a moderateable threshold. Popularity of a forum like/. plants the seeds of its own demise.
One group at Los Alamos did just that with previous generation (21164) Alpha boxes and were quite happy with the results. I understand that they're now looking at building a cluster from 21264s, which have significantly better FP performance. Pretty zippy little buggers.
Intel has a history of supporting GCC for new chips. I was involved with a 1989 project where Intel gave money and machines to RMS to get GCC running well on the 386 as part of a plan to get Mach ported (when SystemV and Xenix were the only games in town). Intel's position was simple: all the 386 compilers suck - maybe with a small dose of cash (especially compared with what they paid Green Hills) these longhairs will come up with something.
The concept of "porting" GCC to something so different from current 32-bit machines as IA64 is going to be a major strain on the backend. Getting performance to even the current GCC standard (ie good but not best-in-class) will take at least a couple o years once hardware is available.
Optical mice were developed at Xerox (PARC maybe?) in the dark ages (before TCP/IP, somewhere around the late 70s) and used a set of three LED receivers to determine which axis movement occured on. I use to run them over my jeans, plaid shirts, and fake wood-grained desktops with no problems. We made mouse pads by laminating patterns we'd print on a standard laser printer.
Sun used to ship optical mice (from Mouse House or Mouse Systems) that used two wavelengths of LED for horizontal and vertical movement detection. Those mice required a special mirror-like aluminum pad with stripes that matched the LED's colors.
Sun's mice were shipped long after Xerox had a vastly superior and much simpler product; Xerox could have owned the mouse market along with everything else if they'ed had their act together.
I wouldn't be surprised if there was optical mouse development prior to Xerox; many people were working in the area at the time.
I still listen to the CDs I bought in 1983 and I haven't noticed any sound or material degredation. The better digital electronics of today's CD players combined with the overkill redundancy of the CD format will let me listen to them for years to come.
Slashdot Mirror
Diagonal routing has been used to some effect on PC boards for a long time, but it's been hard to find enough wires going between the same sources and destinations to get a big group (river) together, so on a given layer they tend to cut off the vertical or horizontal routes that other pairs need. If you can arrange related groups of circuits to be adjacent ro on the same vertical or horizontal axis then diagonal routing isn't needed. It can certainly help time-critical signals in a sort of kludgey way if they're going to be wasting most of M4/M5 anyhow.
Me too. I couldn't believe that hacker humor could make it into Saturday morning cartoons. The concept of Phong being Phong shaded (hey, where's Goraud? He could be called Henri and have a French accent) was enough to get my lousy butt out of bed on onto the couch where it belonged. ReBoot and The Tick reignited my interest in cable TV and changed my life. I hate them.
Intel, Microsoft, Compaq and other heavyweights make many strategic investments in companies developing promising technology. Since they're in business they expect to eventually make money on their investments, which is normal for public companies. At the time of Intel's investment Rambus stock was trading much closer to $10/share than it is today, and Rambus had no customers in the CPU space, so it was a much riskier deal. Rambus' current price (and Intel's current paper profit) is due more to "irrational exuberance" in the stock market than any old-school valuation. The slimey business practice of getting competitive advantage by investing in partners is common in high-tech, but is far more common (in fact normal) in Germany; I'm sure that Dr Tom could also list the firms from his homeland, starting with all the banks, then working through the heavy industries on his way to media firms which he has similar complaints. Rambus will win because it produces higher bandwidth at lower cost (power, pincount, etc) than SDRAM, or it will fail because the RF engineering needed for 800+MHz traces can't be done economically by the time SDRAM busses can deliver the same data rates. The investments made in Rambus by its semiconductor partners will have little say in the long run, though they may have funded the firm for long enough to give the new idea a reasonable chance in the marketplace. elbee
At least one member of Transmeta's hardware team used to work at DEC WRL, which issued that famous technical note on electric pickles. Transmeta could have hired him for his pickle know-how, which when added to Linus' lutefisk background may give us another hint towards Transmeta's plan for world domination. ellbee
Swiss banks accounts come in many flavors; the most famous is the "numbered" account where only one bank official keeps a (non-compterized) record of who actually has power over the account. Most Swiss bank accounts are just like an account anywhere else, except they're covered by Swiss banking laws which make it more difficult for investigators to gain access to records. Over the past few years Swiss banks have become much more cooperative with international law enforcement (especially drug laundering), though they're much more private than American banks. Unless you make the transfer with cash US authorities should be able to follow Internet transactions even if they're into foreign accounts.
Last I heard AV had over 250 million non-duplicate web pages indexed, and knew of (but hadn't indexed) several hundred million more. One thing mentioned on the CNBC interview today is that CMGI will continue growing the index, and hopes to have "the entire web" indexed over the next few years, which they think will be over one billion pages.
I believe that the peak query rate is over 600/s,
which means completing a query every 1.5ms. Good thing computers are getting faster, memory is getting denser, and 64-bit addressing is a reality.
Yes, I know people there too and those numbers seem to be correct. But are sex-related queries still the most popularity?
There has been similar UI work before at Xerox PARC and notably the Fisheye View work by Marc H. Brown at Compaq's (formerly Digital's) Systems Research Center.
r eports/abstracts/src-rr-084.html
http://gatekeeper.dec.com/pub/DEC/SRC/research-
A good path towards Dick Tracy watches.
I arranged for three people (including myself) to get 384/384 service from Concentric/Covad in the Bay Area about eighteen months ago, and we've been pretty happy the result. My install was relatively smooth - PacBell (the ILEC) came out and provisioned the line, shorting just two of the six voice lines running into my apartment building which took only two extra hours to straighten out. Concentric emailed me IP adresses, DNS info, etc. Covad arrived with a preconfigured modem. We hooked it all up and it didn't work, but after only two modem reconfigs it came up and has been ticking ever since.
Most of the time I get close to the advertised bandwidth, but it's clear that they have to work on congestion. Packet loss to local, well connected Bay Area hosts regularly show 10% packet loss, and traceroute points the finger squarely into Concentric's IP network. Every few days the network freezes for a few minutes and all connections drop, but I've no idea where to point the finger.
Not everything has been smooth. My customer rep left witgh no notice and I haven't been assigned a new one. I couldn't find a 24 hour trouble line in the phone book or by calling Concentric HQ. Their accounting department is so screwed up that at one point I found they hadn't sent a bill out for five months. But the biggest problem is that they haven't kept their rates competitive - we're keeping the three Concentric lines, but the next dozen from work are going to PacBell which offers simlar service for 1/3 the price.
We currently have several dozen people with ISDN at home. The plan is to evaluate the next batch of DSL users and decide if we want to go further down that path.
ellbee
Back in to old days (you young whipper snapper) reading Usenet meant reading ALL of Usenet. But as its popularity grew the volume grew and the news groups proliferated and the signal to noise ratio dropped - and that was before alt.* and AOL. /. can continue to be truly worthwhile only as long as the posts are relevent and the volume stays below a moderateable threshold. /. plants the seeds of its own demise.
Popularity of a forum like
ellbee
One group at Los Alamos did just that with previous generation (21164) Alpha boxes and were quite happy with the results. I understand that they're now looking at building a cluster from 21264s, which have significantly better FP performance. Pretty zippy little buggers.
http://cnls.lanl.gov/Internal/Computing/Avalon/
Intel has a history of supporting GCC for new chips. I was involved with a 1989 project where Intel gave money and machines to RMS to get GCC running well on the 386 as part of a plan to get Mach ported (when SystemV and Xenix were the only games in town). Intel's position was simple: all the 386 compilers suck - maybe with a small dose of cash (especially compared with what they paid Green Hills) these longhairs will come up with something.
The concept of "porting" GCC to something so different from current 32-bit machines as IA64 is going to be a major strain on the backend. Getting performance to even the current GCC standard (ie good but not best-in-class) will take at least a couple o years once hardware is available.
Optical mice were developed at Xerox (PARC maybe?) in the dark ages (before TCP/IP, somewhere around the late 70s) and used a set of three LED receivers to determine which axis movement occured on. I use to run them over my jeans, plaid shirts, and fake wood-grained desktops with no problems. We made mouse pads by laminating patterns we'd print on a standard laser printer.
Sun used to ship optical mice (from Mouse House or Mouse Systems) that used two wavelengths of LED for horizontal and vertical movement detection. Those mice required a special mirror-like aluminum pad with stripes that matched the LED's colors.
Sun's mice were shipped long after Xerox had a vastly superior and much simpler product; Xerox could have owned the mouse market along with everything else if they'ed had their act together.
I wouldn't be surprised if there was optical mouse development prior to Xerox; many people were working in the area at the time.
Stanford's CS Building - Gates Building
Camridge's CS Building (in design) - Gates Building
MIT's new LCS Building (just announced) - Gates Building
Random trend? I don't think so.
I still listen to the CDs I bought in 1983 and I haven't noticed any sound or material degredation. The better digital electronics of today's CD players combined with the overkill redundancy of the CD format will let me listen to them for years to come.