If you ever see a car with 1 KNm torque, tell me...
How can people in the 21th century measure forces or torques with units like "a pound at the lenght of the feet of a old king"...
And i still cant realize the beauty of a system of lenght units that convert like a*12=b, b*3=c, c*something=d.
how many yards are 0.34 miles?:)
HAHAHA!
Unter linux, you DID. And you still do if you have a uncommon graphic card.
I remember using debian on a a2000+ with a kyro2.
Studdering windows. Areas waiting to be filled with color. yeah.
almost like win3.1
But thats an APPLE estimation This while working with larg array the locality of the data is very low. Mostly you do MACs (not apple Macs, but MultiplyAccumulates). It is true that you can get VERY high theoretical FLOP ratings using SIMD, but considere: In the end, you still have to write a result to memory and read a new input value. 30 Gfps would need 240 GB/s memory bandwith. This arent your wheatstone benchmarks where all code and data fits into l1 cache.
One high end hard disc delivers 50MB+/s. One gigabit Ethernet card can do >80MB/s Together they are limited by PCI. Now try Raid, TV-Card with PCI-OVerlay, GFX-Cards (Yes, they need a few 100MB/s)...
Plus remember that you NEVER EVER reach 133 MB/S with PCI. Even a single device can be happy to get 110MB with long bursts, and if you have many devices, effective total bandwith is more like 66 than 133 MB/s.
I know it is better that its left out, because it wont work in a movie. But in the book it has the real message: heroes cant just go an a quest and come back as if nothing had happened. And the final leaftaking of frodo and to last line of sam made me cry. But i know that other people found that way to end the book horrible.
Can you tell me where your "unlimited bandwith" comes from? Lets use following scenario: Every user has a 786/128 bkit ADSL line. User 0 shares a 300 MB file.
Without swarming: after 1h there is another source (one complete transfer)->256 kbit upload. after 2h 2 new sources ->512 Kbit/s upload and so on
With swarming: (Bittorrent) after one minute: 1 block transfert. Client uses upload to send to other users. after 2 minutes: client 1 has given client2 his block+plus recieved another from client 0 ->384kbit upload after 3 minutes: client 3 got a packet from server, client 4,5 got one from 1,2. ->640 kbit upload....
Your analyses assumes there is an unlimited number of users available for EVERY release using your methode. But even if users let their client run 24/7, they will have many diffrent thing on their share, all sharing the same upload.
How do you think you network can scale log(x)?!? Considering your example about ADSL, you need 4000 users with DO NOT participate in another download giving their full upload to sustain 1000 new downloaders. Do you think this is more realistic than assuming that people have their bittorent running half the night uploading after finishing download at 3am?
Well, they sell their engine for 250,000$, so if they give the licence away for free, this is a service woth 250,000$.
Or do you think all the developers of games based on the unreal engine got the engine because they asked nice?
I Could be wrong but i believe they use a telescope from arecibo, but not the main one but a 20(?)meter secondary telescope. Well, but of course it would make perfect sense to give all the data a second use....
They use the earth's rotation as one axis of movement, plus they can move the reciever around (it is hanging from cables). Naturally, the second is rather limited. So they basically sort the targets in the correct order and then watch evey one of them when it becomes visible
Hey, this isnt your neigbours dish antenna, they got a whole day on the largest radio-telescope in the world.
This thing has 300m diameter.
Compare this to the fact that the "normal" data they use is from a insignificant, tiny telescope.
It maybe time that you learn the difference between Hardware Raid 0/1 and Hardware Raid5. Not needing any real computating power ( no need for to XOR 4+datastreams) and only 2 channels Hardware Raid 0/1 is dirt-cheap. You MAY call it software-raid because the actual work is done by the driver. But I call it hardware-raid as soon as the raid is recognized at bios-level (in contrast to REAL Software Raid like the one in Win2000 Server or WinXp pro).
And you would be surprised (if you had a clue) how damn slow even "real" ide-hardware-Raid controllers can be. Yeah, the 600$ 3Coms are 100% faster than software stripeset, but in Raid0/1 they can be slower than your "i dont have processing power onboard to speed up datatransfers" ide-controllers with added raid functionality.
btw: How should a card use "additional Computing Power" to speed up a raid 0 or 1?
Have you seen any recent ide-board? A short list of motherboards with ide-raid onboard. (only socket A, else it would be too long) Abit AT7 Max2 Abit KD7G Abit KD7RAID ASus A7N8X Deluxe Chaintech 7KDD Chaintech 7NJS Zenith ECS K7VTA3 V5 (yes, even a ecs 64 MB) ECS L7VTA EPOX EP-8K9A3+ GIGABYTE 7NNXP GIGABYTE GA-7DPXDW+ LeadTek K7NCR18G-PRO-I MSI K7N2 Delta-ILSR MSI K7N2G-ILSR MSI K7T266 Pro2-RU MSI KT4 Ultra-SR Soyo SY-KT333 Dragon Ultra
All this MBs are currently selling. Heck, even 2 years ago ide raid was on most "premium" boards.
Have you seen any recent ide-board?
A short list of ide motherboards with raid onboard.
(only socket A, else it would be too long)
Abit AT7 Max2
Abit KD7G
Abit KD7RAID
ASus A7N8X Deluxe
Chaintech 7KDD
Chaintech 7NJS Zenith
ECS K7VTA3 V5 (yes, even a ecs 64 MB)
ECS L7VTA
EPOX EP-8K9A3+
GIGABYTE 7NNXP
GIGABYTE GA-7DPXDW+
LeadTek K7NCR18G-PRO-I
MSI K7N2 Delta-ILSR
MSI K7N2G-ILSR
MSI K7T266 Pro2-RU
MSI KT4 Ultra-SR
Soyo SY-KT333 Dragon Ultra
All this MBs are currently selling. Heck, even 2 years ago ide raid was on most "premium" boards.
Mods please moderate down
on
ClusterKnoppix
·
· Score: 1
The problem is long time accuracy. Perhaps the small blast will just move it right on target.
Hey, it is even numerically challanging to prodict the orbit for 800 years with such a high resolution.
Well, that aint that easy. You cant just "switch" from RAM to Flash or ASICS. YOu need at least 45+ days until production is running even with compatible processes.
Main reason for the huge fluctuations is the fact that fabs are that expensive. Every company has to continue production even with very low prices because nobody can afford not solling chips. Even if they are below production cost. Also they cant just "mothball" a fab until the demand is large again because to ramp production and yield to normal levels can take half a year. Its just a gamble: Every company continues making losses until someone has to give up (or they loose the nerves), and then they rise the prices again (because they would go broke if they didnt have a "high price phase" every 2 years or so.
Well, i dont complain about ram prices: Currently the prices are really low (again), and everybody knows they will rise again. You cant just wait until ram is sold at a loss and then demand that this price level has to remain...
Do you really need official awards saying that an actor or movie is good?. Hey, with a book it makes more sense. you need a lot more time to read a book than to watch a movie. If a movie sucks, you lost 2 hours. if you try to finish a novel for 3 weeks only to find out it doest get better, well then you lost a lot more time....
So you want some kind of review/award system to see the gems of the genere, but if "conventional" media either ignores or geekifies all the stuff, you need special awards. HEy, it only the nebular and the hugo. If you really wanna see people jerking off each other, watch the next emmy award.
This new law has a good paragraph: It allows scientific articals ect to be distributed in intranets without need for royality fees if it is for tutoring purposes. Of couese the puplishers are fuming, but they are even worse than the music industry. (in germany,for example), a great part of the research at university is funded by the gouvernment. The results are given to puplishers for free. Now the puplishers make a little bit of peer-review (of course everyone his happy if he is allowed to review an article, so it doest cost the company) and sells the result back to the government (for a lot of money). So far so good (or bad). Now welcome to the digital age. The last xxx years everybody went to the university library and xeroxed tons of stuff. Now they want to release the stuff digitally do a littly pay per view. Great.
Slashdot Mirror
If you ever see a car with 1 KNm torque, tell me... How can people in the 21th century measure forces or torques with units like "a pound at the lenght of the feet of a old king" ...
And i still cant realize the beauty of a system of lenght units that convert like a*12=b, b*3=c, c*something=d.
how many yards are 0.34 miles? :)
HAHAHA! Unter linux, you DID. And you still do if you have a uncommon graphic card. I remember using debian on a a2000+ with a kyro2. Studdering windows. Areas waiting to be filled with color. yeah. almost like win3.1
And not to forget that the Power4 has 128 MB level3 cache. That COULD also affect performance....
Not very well...
Mostly because of reduced memory bandwith...
But thats an APPLE estimation
This while working with larg array the locality of the data is very low. Mostly you do MACs (not apple Macs, but MultiplyAccumulates). It is true that you can get VERY high theoretical FLOP ratings using SIMD, but considere: In the end, you still have to write a result to memory and read a new input value.
30 Gfps would need 240 GB/s memory bandwith.
This arent your wheatstone benchmarks where all code and data fits into l1 cache.
No
One high end hard disc delivers 50MB+/s.
One gigabit Ethernet card can do >80MB/s
Together they are limited by PCI.
Now try Raid, TV-Card with PCI-OVerlay, GFX-Cards (Yes, they need a few 100MB/s)...
Plus remember that you NEVER EVER reach 133 MB/S with PCI. Even a single device can be happy to get 110MB with long bursts, and if you have many devices, effective total bandwith is more like 66 than 133 MB/s.
you reread again. the post said it has been and still does
You only need to post a link to the reply
Why should gollum jump? Neither good nor evil gollum have any reason...
I know it is better that its left out, because it wont work in a movie.
But in the book it has the real message: heroes cant just go an a quest and come back as if nothing had happened. And the final leaftaking of frodo and to last line of sam made me cry.
But i know that other people found that way to end the book horrible.
Can you tell me where your "unlimited bandwith" comes from?
Lets use following scenario:
Every user has a 786/128 bkit ADSL line.
User 0 shares a 300 MB file.
Without swarming:
after 1h there is another source (one complete transfer)->256 kbit upload.
after 2h 2 new sources ->512 Kbit/s upload
and so on
With swarming: (Bittorrent)
after one minute: 1 block transfert. Client uses upload to send to other users.
after 2 minutes: client 1 has given client2 his block+plus recieved another from client 0 ->384kbit upload
after 3 minutes: client 3 got a packet from server, client 4,5 got one from 1,2. ->640 kbit upload....
Your analyses assumes there is an unlimited number of users available for EVERY release using your methode. But even if users let their client run 24/7, they will have many diffrent thing on their share, all sharing the same upload.
How do you think you network can scale log(x)?!?
Considering your example about ADSL, you need 4000 users with DO NOT participate in another download giving their full upload to sustain 1000 new downloaders. Do you think this is more realistic than assuming that people have their bittorent running half the night uploading after finishing download at 3am?
Actually, if you set compression to low, its not much slower than winzip and still better at compressing
Well, they sell their engine for 250,000$, so if they give the licence away for free, this is a service woth 250,000$. Or do you think all the developers of games based on the unreal engine got the engine because they asked nice?
I Could be wrong but i believe they use a telescope from arecibo, but not the main one but a 20(?)meter secondary telescope.
Well, but of course it would make perfect sense to give all the data a second use....
They use the earth's rotation as one axis of movement, plus they can move the reciever around (it is hanging from cables).
Naturally, the second is rather limited.
So they basically sort the targets in the correct order and then watch evey one of them when it becomes visible
Hey, this isnt your neigbours dish antenna, they got a whole day on the largest radio-telescope in the world. This thing has 300m diameter. Compare this to the fact that the "normal" data they use is from a insignificant, tiny telescope.
It maybe time that you learn the difference between Hardware Raid 0/1 and Hardware Raid5.
Not needing any real computating power ( no need for to XOR 4+datastreams) and only 2 channels Hardware Raid 0/1 is dirt-cheap.
You MAY call it software-raid because the actual work is done by the driver. But I call it hardware-raid as soon as the raid is recognized at bios-level (in contrast to REAL Software Raid like the one in Win2000 Server or WinXp pro).
And you would be surprised (if you had a clue) how damn slow even "real" ide-hardware-Raid controllers can be. Yeah, the 600$ 3Coms are 100% faster than software stripeset, but in Raid0/1 they can be slower than your "i dont have processing power onboard to speed up datatransfers" ide-controllers with added raid functionality.
btw:
How should a card use "additional Computing Power" to speed up a raid 0 or 1?
Have you seen any recent ide-board?
A short list of motherboards with ide-raid onboard.
(only socket A, else it would be too long)
Abit AT7 Max2
Abit KD7G
Abit KD7RAID
ASus A7N8X Deluxe
Chaintech 7KDD
Chaintech 7NJS Zenith
ECS K7VTA3 V5 (yes, even a ecs 64 MB)
ECS L7VTA
EPOX EP-8K9A3+
GIGABYTE 7NNXP
GIGABYTE GA-7DPXDW+
LeadTek K7NCR18G-PRO-I
MSI K7N2 Delta-ILSR
MSI K7N2G-ILSR
MSI K7T266 Pro2-RU
MSI KT4 Ultra-SR
Soyo SY-KT333 Dragon Ultra
All this MBs are currently selling. Heck, even 2 years ago ide raid was on most "premium" boards.
Have you seen any recent ide-board? A short list of ide motherboards with raid onboard. (only socket A, else it would be too long) Abit AT7 Max2 Abit KD7G Abit KD7RAID ASus A7N8X Deluxe Chaintech 7KDD Chaintech 7NJS Zenith ECS K7VTA3 V5 (yes, even a ecs 64 MB) ECS L7VTA EPOX EP-8K9A3+ GIGABYTE 7NNXP GIGABYTE GA-7DPXDW+ LeadTek K7NCR18G-PRO-I MSI K7N2 Delta-ILSR MSI K7N2G-ILSR MSI K7T266 Pro2-RU MSI KT4 Ultra-SR Soyo SY-KT333 Dragon Ultra All this MBs are currently selling. Heck, even 2 years ago ide raid was on most "premium" boards.
Noone should read his shit
The problem is long time accuracy. Perhaps the small blast will just move it right on target. Hey, it is even numerically challanging to prodict the orbit for 800 years with such a high resolution.
Well, that aint that easy.
You cant just "switch" from RAM to Flash or ASICS.
YOu need at least 45+ days until production is running even with compatible processes.
Main reason for the huge fluctuations is the fact that fabs are that expensive. Every company has to continue production even with very low prices because nobody can afford not solling chips. Even if they are below production cost. Also they cant just "mothball" a fab until the demand is large again because to ramp production and yield to normal levels can take half a year.
Its just a gamble: Every company continues making losses until someone has to give up (or they loose the nerves), and then they rise the prices again (because they would go broke if they didnt have a "high price phase" every 2 years or so.
Well, i dont complain about ram prices: Currently the prices are really low (again), and everybody knows they will rise again. You cant just wait until ram is sold at a loss and then demand that this price level has to remain...
Do you really need official awards saying that an actor or movie is good?.
Hey, with a book it makes more sense. you need a lot more time to read a book than to watch a movie. If a movie sucks, you lost 2 hours. if you try to finish a novel for 3 weeks only to find out it doest get better, well then you lost a lot more time....
So you want some kind of review/award system to see the gems of the genere, but if "conventional" media either ignores or geekifies all the stuff, you need special awards.
HEy, it only the nebular and the hugo.
If you really wanna see people jerking off each other, watch the next emmy award.
This new law has a good paragraph:
It allows scientific articals ect to be distributed in intranets without need for royality fees if it is for tutoring purposes.
Of couese the puplishers are fuming, but they are even worse than the music industry.
(in germany,for example), a great part of the research at university is funded by the gouvernment. The results are given to puplishers for free. Now the puplishers make a little bit of peer-review (of course everyone his happy if he is allowed to review an article, so it doest cost the company) and sells the result back to the government (for a lot of money).
So far so good (or bad).
Now welcome to the digital age. The last xxx years everybody went to the university library and xeroxed tons of stuff. Now they want to release the stuff digitally do a littly pay per view. Great.