Put a cabinet in the middle of the room and replace the tiles around it with perforated ones and you get a lot of cooling for that cabinet.
Maybe this is the problem. Every industrial datacenter I have been in places racks over either empty spaces, or tiles with a large vent in them. The rack has fans in it to force air through vertically (bottom to top). A few perforated tiles get scattered about for the humans, but I have been in some datacenters without them to maximize airflow to the racks. But then again, I have worked with electronics that make current CPUs feel like popsicles in comparison.
Lots of people do DSP on general purpose CPUs. One reason is that some large DSP systems need the features in a full RTOS that only runs on a real CPU. Another is portability: the transition from desktop prototype to real system is often easier when the taret CPU is non a true DSP. Another reason has to do with memory and I/O. General purpose CPUs can have gobs of memory and have an easier interface to things like PCI and other standard busses. DSPs often have limited memory and require custom bridges to get to standard busses.
DSP on GP CPUs isn't always best, but I think you would be surprised at the number of people doing it.
Try to find Jacob Nielsen's "Why You Only Need to Test With 5 Users" or "A mathematical model of the finding of usability problems" for some more insight into population size for usability testing.
My copy is at home, but I believe Stevens' Unix Network Programming has an example with either listen() or accept() where using a goto is the only way to guarantee proper results.
Years ago, I had a collegue (a long-time RF engineer) tell me that the day they introduce wireless networks into the office will also be the same day he retires.
he killed her own child by signing up for a service that didn't fit her needs.
Ok, you cancel your landline and buy a mobile phone. The mobile phone doesn't work in your area. DO you sue T-Mobile?
I think the real problem is that VoIP is being marketed as a replacement as replacement for normal phone service. I don't recall any mobile phone commericals that say "Buy a wireless phone and drop your phone service," but I do recall ones that say "Buy VoIP from is and drop your normal phone service."
I think the average person here knows about the problems with VoIP, but I doubt the average person does. Also, what about people who dial 911 from a VoIP phone who don't know its a VoIP phone (eg, you have a heart attach and a friend calls 911).
Wild guess, but custom web-based apps are pretty popular in mega-corps. Mega-corps have to support a wide variety of users, including those with accessibility needs. Making Firefox more accessible in DHTML areas means they could potentially sell more servers and services to better support the needs of mega-corps.
My book with this info is at home, but IIRC, 60 GHz is one of the trouble spots for RF transmission because of absorption by atmospheric oxygen. This phenomena is exploited for some secure radios.
I have a hardcopy of a book called the Timing Reference Handbook. It is a fairly length tech note from a company called Austron, who got bought by Datum, who got bougth by someone else. At one point I know it was available as a PDF, but a quick search at the Datum website didn't reveal it, though. The interested party should be able to dig it up.
The book describes the difference between the various time bases (UT0, UT1, UT2, UTC, atamic time, etc) and gives some pretty good detail about why we have leap seconds.
This is what I was thinking. Most of my audio and projects would fit on a 4G drive (4G would be tight for video). A drive like this would get used while I was working on the project, and then finished projects would get moved to to my online and offline storage.
IIRC, Sun did this with the early versions of Solaris. The transtion from SunOS to Solaris was really painful, especially wrt. SunOS binary compatibility. Now that I think about it, it could have just been a bounty on compatibility problems.
When our interns or junior staff start complaining about mundane work, I show them pictures of wire-wrapping and tell them that used to be what the interns and junior staff did when I was learning the ropes. That ususally shuts them up for a while...
Hearing about REXX brings up memories about sitting on an IBM 3178, finishing an assignment, going to print it, only to realize that some prof decide to sent about 3000 pages of data to the line printer ahead of me.
It also brings up the nightmares of having to use XEDIT to do the work...
The mastering step of audio production involves the final EQ, compression/limiting, and output to the final media. When something is remastered, the mix tape is reprocessed. If you have the multitrack tape or stems, then you can remix to a higher bitdepth, saving some rounding/dithering degredation, master from this, and then dither to 16-bit as a final step. If done right, the end result is often better than the original, especially since dithering algorithms and other processing has improved greatly in the last few years.
As for difficult to test... not really, or at least I don't see why it would be worse than clocked chips. Have a test signal, put a stream of data into the chip, then see what comes out. Async should work the same as sync in that case...
The problem has to due with the variations in timing due to temperature and voltage changes. In general, synchronous logic is simulated at the speced max temperature and lowest voltage (this is typically the worst case (*)). With asynchronous logic you have to test over a range because you have to worry about things happening too soon rather than just too late, and as you add more parallel and sequential circuits things get really complicated.
(*) I'm not positive this is the worst case combination...
Slashdot Mirror
Put a cabinet in the middle of the room and replace the tiles around it with perforated ones and you get a lot of cooling for that cabinet.
Maybe this is the problem. Every industrial datacenter I have been in places racks over either empty spaces, or tiles with a large vent in them. The rack has fans in it to force air through vertically (bottom to top). A few perforated tiles get scattered about for the humans, but I have been in some datacenters without them to maximize airflow to the racks. But then again, I have worked with electronics that make current CPUs feel like popsicles in comparison.
Lots of people do DSP on general purpose CPUs. One reason is that some large DSP systems need the features in a full RTOS that only runs on a real CPU. Another is portability: the transition from desktop prototype to real system is often easier when the taret CPU is non a true DSP. Another reason has to do with memory and I/O. General purpose CPUs can have gobs of memory and have an easier interface to things like PCI and other standard busses. DSPs often have limited memory and require custom bridges to get to standard busses.
DSP on GP CPUs isn't always best, but I think you would be surprised at the number of people doing it.
Try to find Jacob Nielsen's "Why You Only Need to Test With 5 Users" or "A mathematical model of the finding of usability problems" for some more insight into population size for usability testing.
My copy is at home, but I believe Stevens' Unix Network Programming has an example with either listen() or accept() where using a goto is the only way to guarantee proper results.
Years ago, I had a collegue (a long-time RF engineer) tell me that the day they introduce wireless networks into the office will also be the same day he retires.
Does this count?
Reak geeks still use shoelace...
he killed her own child by signing up for a service that didn't fit her needs.
Ok, you cancel your landline and buy a mobile phone. The mobile phone doesn't work in your area. DO you sue T-Mobile?
I think the real problem is that VoIP is being marketed as a replacement as replacement for normal phone service. I don't recall any mobile phone commericals that say "Buy a wireless phone and drop your phone service," but I do recall ones that say "Buy VoIP from is and drop your normal phone service."
I think the average person here knows about the problems with VoIP, but I doubt the average person does. Also, what about people who dial 911 from a VoIP phone who don't know its a VoIP phone (eg, you have a heart attach and a friend calls 911).
It may help them sell more servers and services?
Wild guess, but custom web-based apps are pretty popular in mega-corps. Mega-corps have to support a wide variety of users, including those with accessibility needs. Making Firefox more accessible in DHTML areas means they could potentially sell more servers and services to better support the needs of mega-corps.
?
Betamax may have been a failure, but Betacam SP was a big hit and is a defacto standard for professionals.
My book with this info is at home, but IIRC, 60 GHz is one of the trouble spots for RF transmission because of absorption by atmospheric oxygen. This phenomena is exploited for some secure radios.
Two words: Literate Programming
I used to work with an RF engineer who used to say he would retire once they made the office wireless...
I have a hardcopy of a book called the Timing Reference Handbook. It is a fairly length tech note from a company called Austron, who got bought by Datum, who got bougth by someone else. At one point I know it was available as a PDF, but a quick search at the Datum website didn't reveal it, though. The interested party should be able to dig it up.
The book describes the difference between the various time bases (UT0, UT1, UT2, UTC, atamic time, etc) and gives some pretty good detail about why we have leap seconds.
This is what I was thinking. Most of my audio and projects would fit on a 4G drive (4G would be tight for video). A drive like this would get used while I was working on the project, and then finished projects would get moved to to my online and offline storage.
IIRC, Sun did this with the early versions of Solaris. The transtion from SunOS to Solaris was really painful, especially wrt. SunOS binary compatibility. Now that I think about it, it could have just been a bounty on compatibility problems.
I'm sorry, Fry, but astronomers renamed Uranus in 2620 to end that stupid joke once and for all.
When our interns or junior staff start complaining about mundane work, I show them pictures of wire-wrapping and tell them that used to be what the interns and junior staff did when I was learning the ropes. That ususally shuts them up for a while...
Hearing about REXX brings up memories about sitting on an IBM 3178, finishing an assignment, going to print it, only to realize that some prof decide to sent about 3000 pages of data to the line printer ahead of me.
It also brings up the nightmares of having to use XEDIT to do the work...
IBM made it available for nearly all of its operating systems. In particular, it was extremely handy if you were on a VM/CMS system.
Google on "flash wear leveling algorithm", and you are bound to turn up some info.
The mastering step of audio production involves the final EQ, compression/limiting, and output to the final media. When something is remastered, the mix tape is reprocessed. If you have the multitrack tape or stems, then you can remix to a higher bitdepth, saving some rounding/dithering degredation, master from this, and then dither to 16-bit as a final step. If done right, the end result is often better than the original, especially since dithering algorithms and other processing has improved greatly in the last few years.
Damn, I am a goodly speller...
Flourescent tubes will also glow if you hold them while standing under high-tension power lines.
As for difficult to test... not really, or at least I don't see why it would be worse than clocked chips. Have a test signal, put a stream of data into the chip, then see what comes out. Async should work the same as sync in that case...
The problem has to due with the variations in timing due to temperature and voltage changes. In general, synchronous logic is simulated at the speced max temperature and lowest voltage (this is typically the worst case (*)). With asynchronous logic you have to test over a range because you have to worry about things happening too soon rather than just too late, and as you add more parallel and sequential circuits things get really complicated.
(*) I'm not positive this is the worst case combination...