I used to do high-milage vehicle chalanges. There were two major driving styles used.. constant engine RPM, and coast/burn.
All the top competitors were coast/burn drivers.
The theory goes like this:
There are two points of engine efficiency.. at full throttle, and when it's off. No matter what engine speed you are at, you are always using a specific ammount of gas to overcome the engine friction, and other running requirements. this ammount is mostly the same across the RPM spectrum.
At full throttle, you have the largest % of engine power going to movement, and the smallest % being used for engine upkeep.
This is why I think the hybrids made so far are stupid.. they use the engine for coasting, and maintaining speed.. they should use the electric motor for cruising, and then bring the engine up to a good speed to re-charge ever 10-20min or so.. depending on how much battery is there.
Unfortunately, the charing process is not efficient enough for this to be very usefull.
The president of the U of MN ham club (well.. former pres now) would listen to CD's while walking around campus with CW encoded at 45+wpm. The guy is amazing for field day.. we were listening to a pileup of 3 CW signals, and he could easily pick out the one he wanted and just write it as it came in. This isn't a crusty old guy either.. just your average EE grad student age.
This is an IBM BlueGene system. I went down to Rochester, MN to see one of these systems.. very interesting architecture.. each "node" is a dual core system on a chip.. the compute node OS is a simple non-multi-tasking kernel with a simple linux-ish libc.
You cross-compile your application on a power4/5 linux box, and then submit it to the system.. they reduced the computational aspect to it's most basic components.. CPU/FPU, memory, and MPI interconnect..
compute nodes don't even have ethernet or drive controlers... all I/O is handled by another specialized I/O node, which provides data over one of the two MPI interconnects.
I would say that you're totaly wrong. Most of the CPU hours I see are good ol fortran. I see some requests for matlab runs, but it's dificult to parallelize, so it ends up being used for small projects, or quick tests of stuff. The bulk of the actual CPU hours is in C/C++/Fortran.
Re:Compatibility? Linux testing?
on
GCC 4.0.0 Released
·
· Score: 4, Informative
actualy, I saw some comments about optimization by the Fedora guys, they compile Fedora to be OPTIMIZED for i686, but still instruction compatable with i386.
specialized applications have code to auto-use different instructions if support is detected.
I think things will continue to be i686 optimized, but still i386 compatable until amd64 arch is more prevelent.
I don't know if this is changing for GCC 4.0/Fedora Core 4.. but I'm going to guess it is not.
Yep.. A few years ago now, I helped setup a computer lab system for a small charter school.. We got the server, a decent cheap box with a couple of 36gig scsi disks. I setup NFS/NIS/autofs. We took an old P100 to setup as a firewall, and a resurected what we could out of a stack of P166s. We installed RH6.2 (latest and greatest release at the time) with abiword, and a few other apps. This was the first time the school had more than 5 computers running at one time.. previous to that, the P166 win98 just crashed and caused problems. We now had a network authenticated login system for every student, who each had their own home directory to store files.. no more floppies needed to save stuff.
A couple of the teachers whined noisily about the fact that they couldn't run FOO application.. but really, all they really needed was word processing and web access. The "tech" head, who was just one of the teachers was a friend of mine, and stuck to his guns and prevented any hostile takeovers of our network.. it worked well.. everything ran as smoothly as a bunch of crappy old PCs could run with no budget.
Eventualy we got around to building some Duron 500s, and installed RH7.x and things were feeling better..
A couple years ago, my friend quit, and moved to another city.. about 6 months after that, their firewall stop responding to ssh, and they never called me back. I just wrote the whole thing off as not worth my time.. oh well.. I bet they had to spend tens thousands of dollars buying new machines and windows licenses, instead of paying me $300 or so to come down and upkeep their network once a year.
Actualy, I have met with the people in charge of this project. They want the whole system rolled out at once, as fast as possible to cover all areas, without gaps.
This isn't municipal broadband either, it's commercial wireless, but the city wants "shared governance" to keep the wireless company in check, and so they have a say in the coverage (to prevent the problems you talk about) Basicaly they are trying to avoid another ricochet, network hardware all over town rusting because they went under.
I live in Saint Paul, and we're trying to do something similar, although we're about 6 months behind minneapolis.
I attended a meeting with the people in charge of this project.
The proposed RFP will be for "shared governance" where the city will have a say in how the network is run, but the service provider (qwest/comcast/timewarner _could_ in theory bid for this) will do all the build out.
They will also provide city backed loans to help with the finantial burden. basicaly better financing terms, because the city is behind them.
The city will pay a certin ammount to have priority access to the network for use with police/fire/municipal departments.
It's a well thought out system, but is potentialy handing another monopoly over a big company. It is un-certin how badly they will step on local hotspots, educational institution wireless, and projects like the Twin Cities Wireless User Group.
(we have a hotspot network covering a large park near downtown)
I don't think I've ever had a problem giving dollar coins as payment.. mostly because when I'm paying cash, I pay close to exact ammount, so they have to look twice thinking I'm not paying enough. I'm also carefuly that i get correct change back.
It's one of those things that I would love to do if I were working in retail, give dollar coins as change.
most of the RPM limitations are due to the fact that the chips responsable for decoding the bit signals off the plattes can't keep up anymore. The reason they use 2.5" platters in the 15k RPM drives is because the bits on the outside of the platter are moving WAY too fast to make it worth the platter space.
Also for the reasons you mentioned.. but signal processing needs to get faster.
AFAIK, there were rumors going around about someone (maybe it was hitachi) working on 20k RPM drives.
Even funnier is I'm from Minneapolis where the geek squad started.. When I droped out of school back in 98 I interviewed there. (long before they were part of best buy) They were not impressive back then either.
Exactly, I have some mission-critical style server that are happily churning away in the corner with woody, I don't want 6 month release cycles, with 18 month drops in support.. it pushes packets, it doesn't surf.
My laptop is Ubuntu, my SO's laptop is Ubuntu.
My home file server is sarge (and maybe MythTV soon)
The router is Bering uClibc
All debian/based.
I liked the solid 2 year release cycle before woody, bo, slink, potato, woody.. all nicely timed... sarge is just getting too fat to get through the damn door. It's about time to put the whole thing on a diet or something.
Harder to start in the cold: False They fixed this with anti-gelling additives in the early 90s, and fuel pump heat-resurculation also helps even at extremely low temps.
Engines need to be larger: False the 1.9L TDI engine has a lower horsepower rating than the 2.0L gas, but if you've ever driven one, you would know that HP isn't the only thing to look at. It has the tourq of a much larger gas engine, so it is MUCH faster off the line, and it can tow stuff.
With the reliability of quality turbochargers these days, there isn't really any disadvantage to diesel over gas.
I am really looking forward to the 2006 ULSD regulations kicking in, so we can get better quality emission control devices on diesel, eliminiating the EPA concerns about NOx emissions and other things.
According to an article by a developer of the Rio Karma, they needed something that could be decoded by the iPod's CPU.
The Rev 1-3 iPods have a smaller CPU cache than the Rio Karma, or the iPod Rev 4 and the iPod Mini. The preformance hit for accessing memory while decoding is too great, so you must fit the decoder in the cache.
ALAC was designed for the simple reason that it has a smaller decoder on the iPod than FLAC.
Same reason why OGG can't be used on iPod Rev 1-3. (and for consistancy, not on the other two) Not enough CPU cache to decode.
I knew there were ways to patch the bios flash to fix this.. but I was not aware of a CMOS flag to disable the check.. can y you post more information?
of course, if you look around, there are a lot cheaper solutions than dynamat.. it's the monstor cable of sound deadening solutions.
http://www.soundprooffoam.com/vinyl_barrier.html
this is basicaly the same stuff as dynamat.. a dense stick-on sound deadening material.. but at 1/2 the price. I've heard of cheaper.. but I don't remember the specifics.
I agree with you about the municipal network.. with data being about as necessary as water for businesses, especialy in my town (minneapolis, mn) which was just rated "#1 tech city"
your quote about the governement and The Bell System is funny, but unfortunately breaking up Ma Bell was probably one of the better moves.
Slashdot Mirror
I used to do high-milage vehicle chalanges. There were two major driving styles used.. constant engine RPM, and coast/burn.
All the top competitors were coast/burn drivers.
The theory goes like this:
There are two points of engine efficiency.. at full throttle, and when it's off. No matter what engine speed you are at, you are always using a specific ammount of gas to overcome the engine friction, and other running requirements. this ammount is mostly the same across the RPM spectrum.
At full throttle, you have the largest % of engine power going to movement, and the smallest % being used for engine upkeep.
This is why I think the hybrids made so far are stupid.. they use the engine for coasting, and maintaining speed.. they should use the electric motor for cruising, and then bring the engine up to a good speed to re-charge ever 10-20min or so.. depending on how much battery is there.
Unfortunately, the charing process is not efficient enough for this to be very usefull.
The president of the U of MN ham club (well.. former pres now) would listen to CD's while walking around campus with CW encoded at 45+wpm. The guy is amazing for field day.. we were listening to a pileup of 3 CW signals, and he could easily pick out the one he wanted and just write it as it came in. This isn't a crusty old guy either.. just your average EE grad student age.
-KC0NBY
This is an IBM BlueGene system. I went down to Rochester, MN to see one of these systems.. very interesting architecture.. each "node" is a dual core system on a chip.. the compute node OS is a simple non-multi-tasking kernel with a simple linux-ish libc.
You cross-compile your application on a power4/5 linux box, and then submit it to the system.. they reduced the computational aspect to it's most basic components.. CPU/FPU, memory, and MPI interconnect..
compute nodes don't even have ethernet or drive controlers... all I/O is handled by another specialized I/O node, which provides data over one of the two MPI interconnects.
flash drives are great.. what does bitmicro have for price/gb?
I would say that you're totaly wrong. Most of the CPU hours I see are good ol fortran. I see some requests for matlab runs, but it's dificult to parallelize, so it ends up being used for small projects, or quick tests of stuff. The bulk of the actual CPU hours is in C/C++/Fortran.
Fedora Core 4 is GCC 4.0
actualy, I saw some comments about optimization by the Fedora guys, they compile Fedora to be OPTIMIZED for i686, but still instruction compatable with i386.
specialized applications have code to auto-use different instructions if support is detected.
I think things will continue to be i686 optimized, but still i386 compatable until amd64 arch is more prevelent.
I don't know if this is changing for GCC 4.0/Fedora Core 4.. but I'm going to guess it is not.
The average user is supposed to buy a Sempron.. DUH
The average user doesn't even need 1/4 the computation power of even athlon64 machines.
Hell.. I do most of my work on a 1.4ghz Pentium M, and a Duron 800.
Are you slow or something?
Dual core != Dual Processor
Besides, there have been dual core Power4 systems, dual core mips chips for a while now.. this is only new for x86 desktop systems.
Yep.. A few years ago now, I helped setup a computer lab system for a small charter school.. We got the server, a decent cheap box with a couple of 36gig scsi disks. I setup NFS/NIS/autofs. We took an old P100 to setup as a firewall, and a resurected what we could out of a stack of P166s. We installed RH6.2 (latest and greatest release at the time) with abiword, and a few other apps. This was the first time the school had more than 5 computers running at one time.. previous to that, the P166 win98 just crashed and caused problems. We now had a network authenticated login system for every student, who each had their own home directory to store files.. no more floppies needed to save stuff.
A couple of the teachers whined noisily about the fact that they couldn't run FOO application.. but really, all they really needed was word processing and web access. The "tech" head, who was just one of the teachers was a friend of mine, and stuck to his guns and prevented any hostile takeovers of our network.. it worked well.. everything ran as smoothly as a bunch of crappy old PCs could run with no budget.
Eventualy we got around to building some Duron 500s, and installed RH7.x and things were feeling better..
A couple years ago, my friend quit, and moved to another city.. about 6 months after that, their firewall stop responding to ssh, and they never called me back. I just wrote the whole thing off as not worth my time.. oh well.. I bet they had to spend tens thousands of dollars buying new machines and windows licenses, instead of paying me $300 or so to come down and upkeep their network once a year.
Actualy, I have met with the people in charge of this project. They want the whole system rolled out at once, as fast as possible to cover all areas, without gaps.
This isn't municipal broadband either, it's commercial wireless, but the city wants "shared governance" to keep the wireless company in check, and so they have a say in the coverage (to prevent the problems you talk about) Basicaly they are trying to avoid another ricochet, network hardware all over town rusting because they went under.
I live in Saint Paul, and we're trying to do something similar, although we're about 6 months behind minneapolis.
RTFA, this is a RFP for a company to build out an infrastructure, similar to ricochet, but with no proprietary spectrum.
I attended a meeting with the people in charge of this project.
The proposed RFP will be for "shared governance" where the city will have a say in how the network is run, but the service provider (qwest/comcast/timewarner _could_ in theory bid for this) will do all the build out.
They will also provide city backed loans to help with the finantial burden. basicaly better financing terms, because the city is behind them.
The city will pay a certin ammount to have priority access to the network for use with police/fire/municipal departments.
It's a well thought out system, but is potentialy handing another monopoly over a big company. It is un-certin how badly they will step on local hotspots, educational institution wireless, and projects like the Twin Cities Wireless User Group.
(we have a hotspot network covering a large park near downtown)
I don't think I've ever had a problem giving dollar coins as payment.. mostly because when I'm paying cash, I pay close to exact ammount, so they have to look twice thinking I'm not paying enough. I'm also carefuly that i get correct change back.
It's one of those things that I would love to do if I were working in retail, give dollar coins as change.
most of the RPM limitations are due to the fact that the chips responsable for decoding the bit signals off the plattes can't keep up anymore. The reason they use 2.5" platters in the 15k RPM drives is because the bits on the outside of the platter are moving WAY too fast to make it worth the platter space.
Also for the reasons you mentioned.. but signal processing needs to get faster.
AFAIK, there were rumors going around about someone (maybe it was hitachi) working on 20k RPM drives.
EnsignTaco? Back in my day, we just called him TacoBoy ;)
Even funnier is I'm from Minneapolis where the geek squad started.. When I droped out of school back in 98 I interviewed there. (long before they were part of best buy) They were not impressive back then either.
Exactly, I have some mission-critical style server that are happily churning away in the corner with woody, I don't want 6 month release cycles, with 18 month drops in support.. it pushes packets, it doesn't surf.
My laptop is Ubuntu, my SO's laptop is Ubuntu.
My home file server is sarge (and maybe MythTV soon)
The router is Bering uClibc
All debian/based.
I liked the solid 2 year release cycle before woody, bo, slink, potato, woody.. all nicely timed... sarge is just getting too fat to get through the damn door. It's about time to put the whole thing on a diet or something.
disadvantages:
Harder to start in the cold: False
They fixed this with anti-gelling additives in the early 90s, and fuel pump heat-resurculation also helps even at extremely low temps.
Engines need to be larger: False
the 1.9L TDI engine has a lower horsepower rating than the 2.0L gas, but if you've ever driven one, you would know that HP isn't the only thing to look at. It has the tourq of a much larger gas engine, so it is MUCH faster off the line, and it can tow stuff.
With the reliability of quality turbochargers these days, there isn't really any disadvantage to diesel over gas.
I am really looking forward to the 2006 ULSD regulations kicking in, so we can get better quality emission control devices on diesel, eliminiating the EPA concerns about NOx emissions and other things.
According to an article by a developer of the Rio Karma, they needed something that could be decoded by the iPod's CPU.
The Rev 1-3 iPods have a smaller CPU cache than the Rio Karma, or the iPod Rev 4 and the iPod Mini. The preformance hit for accessing memory while decoding is too great, so you must fit the decoder in the cache.
ALAC was designed for the simple reason that it has a smaller decoder on the iPod than FLAC.
Same reason why OGG can't be used on iPod Rev 1-3. (and for consistancy, not on the other two) Not enough CPU cache to decode.
Unfortunately, there is one problem with that.. and that is Dell.
The first x86-64 machines we had at my work were Intel. Simply for the fact that we are an EDU and get good prices on Dell machines.
Now.. if we do pick x86-64 for a cluster solution, Intel will probably not happen.
If Dell shiped AMD procs.. Intel would die overnight.
the farther left you go, the closer it comes to wraping around to the far-right.
Both ends are just plain crazy.
I knew there were ways to patch the bios flash to fix this.. but I was not aware of a CMOS flag to disable the check.. can y you post more information?
of course, if you look around, there are a lot cheaper solutions than dynamat.. it's the monstor cable of sound deadening solutions.
l
http://www.soundprooffoam.com/vinyl_barrier.htm
this is basicaly the same stuff as dynamat.. a dense stick-on sound deadening material.. but at 1/2 the price. I've heard of cheaper.. but I don't remember the specifics.
I agree with you about the municipal network.. with data being about as necessary as water for businesses, especialy in my town (minneapolis, mn) which was just rated "#1 tech city"
your quote about the governement and The Bell System is funny, but unfortunately breaking up Ma Bell was probably one of the better moves.