As far as I can see, this discussion is trying to figure out how to prove where a user is.
This could be done via GPS or some other position specification technology. On a different tack, one could use source defined routes if they were available. That way, when a user is first logged into a system, the server can figure out how to get to their subnet. Once this has been found once, all subsequent transmissions could be sent to the same location. This way, someone connecting from the wrong network will not be able to use the server.
I'm guessing that internet backbone routers don't like being told what to do, so this is wouldn't be a viable option.
There is one problem with the calculations presented in the article:
4MB of RAM would not store the display lists for an entire second. Rather, it would set them up every frame. Therefore, those 5 million triangles could be done each frame. This is not including textures or anything else that will chew up memory.
This reminds me of a PlayStation vs. Nintendo 64 article where they stated that the PS could do 2 mil. triangles per second when the N64 could only manage 1/2 mil.. What they failed to mention was that the PS figure was for 2D flat shaded triangles and the N64 number was for fully textured and lit 3D triangles.
Then, the PS was the older console. Now, the Dreamcast is. A company will always try and make it's older product look good by trying to shoot down a well hyped new-to-market machine.
(I'm not speaking from experience here, so take it with a grain of salt)
Most of the antennas they are working with would be uni-directional. i.e. You have to point them toward another antenna that is pointing at you. This is inherited from the satellite dishes that are used.
The omnidirectional antenna designs that the website links to would be what you need. Also, you would have to stay within range of your nearest network gateway to access the internet to get outside.
How long is it before some Bluetooth or WAP people start crying "patent infringement"?
This DVD recompression business has been going on for months. There are a host of tools that have been developed to help copy one part or another of a vob stream.
If you feel like it, take a look at go.to/dvdsoft. Here, new versions of many tools are posted every few days. Personally, I prefer FlaskMpeg coupled with DODSpeedRipper.
Normally Slashdot is at the front, bringing us news (almost) as soon as it's available. Albeit this is the first time it's been on a commercial / famous site, this is still old news.
This will probably get shot down by a passing astrophysicist, but anyway:
Could a large meteorite have impacted on Mars circa 200 million years ago and thrown pieces of rock clear of the planet?
I don't know much about the surface of the red planet, but is there a large enough crater to support this?
Alternatively, such a large impact could have made a serious hole in the crust, creating a large volcano. This could have been the origin of Olympus Mons (the monster volcano roughly on Mars' equator).
You will need an exceptional kick-arse interconnect structure to get decent performance out of the various CPUs. This would be especially true if code running on one CPU needed data from another processor.
A couple of extra points:
- What happens when you want to upgrade your CPU?
- Memory becomes more expensive as a processor is tightly coupled to it.
I shudder to think of the kiddies at primary school arguing over the benefits of a hypercube compared to a mesh interconnect structure.
Just how complex is SRAM to build? Please note first that I don't know much about SRAM.
As far as I know, SDRAM requires 1 capacitor and 1 transistor per cell minimum. SRAM requires about 4. With companies today manufacturing 256M SDRAM, could they easily build 64M SRAM modules for a similar price?
The latency of these SRAM (not reliant of capacitor discharge) could push memory bus clock rates up to several hundred MHz. Data from an SRAM is available in a nanosecond or less, not 3 to 5 ns.
The other part of my comment is on a RAM drive. Could a memory manufacturer revive obsolete memory technology (fastpage, 30-pin SIMMs) that is extremely cheap? If so, they could produce an inexpensive 1-2 GB memory module that sits on an IDE interface. You could easily use the full bandwidth of an UATA-66 channel.
Instead of resorting to (comparatively) super-slow hard drive for virtual memory paging, the OS could just use a slightly slower memory technology. Using RAM negates the seek times that are inherent to hard drives.
You could give them a lesson in "How the old hackers used to do it". Show them how to call crypt() and then let them loose on the school's passwd file.
Mr Kurtz started out at the wrong end of the business. He started out knowing almost nothing and worked up from there. After four years on the job, he seems to think that putting a PC together is a great achievement. For those of us that are used to using technical jargon all the time and having it instantly understood, the average user is incredibly tiresome. They are at the stage now where we started ten years ago. At that time, we had to learn it ourselves and were not spoon fed by any helpdesk.
I have had the unfortunate experience of using SLOaris on and off over the last two years. I am not impressed with any part of Sun's OS. Why on earth would anyone want to inflict this on an x86?
Slashdot Mirror
As far as I can see, this discussion is trying to figure out how to prove where a user is. This could be done via GPS or some other position specification technology. On a different tack, one could use source defined routes if they were available. That way, when a user is first logged into a system, the server can figure out how to get to their subnet. Once this has been found once, all subsequent transmissions could be sent to the same location. This way, someone connecting from the wrong network will not be able to use the server. I'm guessing that internet backbone routers don't like being told what to do, so this is wouldn't be a viable option.
There is one problem with the calculations presented in the article:
4MB of RAM would not store the display lists for an entire second. Rather, it would set them up every frame. Therefore, those 5 million triangles could be done each frame. This is not including textures or anything else that will chew up memory.
This reminds me of a PlayStation vs. Nintendo 64 article where they stated that the PS could do 2 mil. triangles per second when the N64 could only manage 1/2 mil.. What they failed to mention was that the PS figure was for 2D flat shaded triangles and the N64 number was for fully textured and lit 3D triangles.
Then, the PS was the older console. Now, the Dreamcast is. A company will always try and make it's older product look good by trying to shoot down a well hyped new-to-market machine.
I'm sure similar bias exists here.
(I'm not speaking from experience here, so take it with a grain of salt)
Most of the antennas they are working with would be uni-directional. i.e. You have to point them toward another antenna that is pointing at you. This is inherited from the satellite dishes that are used.
The omnidirectional antenna designs that the website links to would be what you need. Also, you would have to stay within range of your nearest network gateway to access the internet to get outside.
How long is it before some Bluetooth or WAP people start crying "patent infringement"?
This DVD recompression business has been going on for months. There are a host of tools that have been developed to help copy one part or another of a vob stream. If you feel like it, take a look at go.to/dvdsoft. Here, new versions of many tools are posted every few days. Personally, I prefer FlaskMpeg coupled with DODSpeedRipper. Normally Slashdot is at the front, bringing us news (almost) as soon as it's available. Albeit this is the first time it's been on a commercial / famous site, this is still old news.
This will probably get shot down by a passing astrophysicist, but anyway:
Could a large meteorite have impacted on Mars circa 200 million years ago and thrown pieces of rock clear of the planet?
I don't know much about the surface of the red planet, but is there a large enough crater to support this?
Alternatively, such a large impact could have made a serious hole in the crust, creating a large volcano. This could have been the origin of Olympus Mons (the monster volcano roughly on Mars' equator).
You will need an exceptional kick-arse interconnect structure to get decent performance out of the various CPUs. This would be especially true if code running on one CPU needed data from another processor.
A couple of extra points:
- What happens when you want to upgrade your CPU?
- Memory becomes more expensive as a processor is tightly coupled to it.
I shudder to think of the kiddies at primary school arguing over the benefits of a hypercube compared to a mesh interconnect structure.
Just how complex is SRAM to build? Please note first that I don't know much about SRAM.
As far as I know, SDRAM requires 1 capacitor and 1 transistor per cell minimum. SRAM requires about 4. With companies today manufacturing 256M SDRAM, could they easily build 64M SRAM modules for a similar price?
The latency of these SRAM (not reliant of capacitor discharge) could push memory bus clock rates up to several hundred MHz. Data from an SRAM is available in a nanosecond or less, not 3 to 5 ns.
The other part of my comment is on a RAM drive. Could a memory manufacturer revive obsolete memory technology (fastpage, 30-pin SIMMs) that is extremely cheap? If so, they could produce an inexpensive 1-2 GB memory module that sits on an IDE interface. You could easily use the full bandwidth of an UATA-66 channel.
Instead of resorting to (comparatively) super-slow hard drive for virtual memory paging, the OS could just use a slightly slower memory technology. Using RAM negates the seek times that are inherent to hard drives.
You could give them a lesson in "How the old hackers used to do it". Show them how to call crypt() and then let them loose on the school's passwd file.
Everyone gets upset when the government censors or manipulates the media. I find it funny that the media is finally manipulating the government.
Mr Kurtz started out at the wrong end of the business. He started out knowing almost nothing and worked up from there. After four years on the job, he seems to think that putting a PC together is a great achievement. For those of us that are used to using technical jargon all the time and having it instantly understood, the average user is incredibly tiresome. They are at the stage now where we started ten years ago. At that time, we had to learn it ourselves and were not spoon fed by any helpdesk.
I have had the unfortunate experience of using SLOaris on and off over the last two years. I am not impressed with any part of Sun's OS. Why on earth would anyone want to inflict this on an x86?
I've found Tcl to be such an *ecch* method of creating an interface. I can't wait until SOME government brings out a law agaisnt it.
All fraud / non-fraud argument aside,
has anyone thought that they might have made
more than one chip?
Make 1000 of these chips and you only have
to be 2.7 time faster than a P2 333.