Mr. Gorman,
There are a few major open source hardware projects (most are listed on wikipedia's open source hardware page.
What is your opinion of the open source hardware projects that you are aware of, and do any stand chance of becoming success stories like so many of the software projects have?
Also, have you come across many business opportunities that could be filled by open source hardware (care the describe them?), or do companies shy away from open source hardware for some reason (if so, why?)?
And, FPGAs are really useful as an implementation, not just for prototyping. It can cost up to $30M to set up an ASIC line, not counting engineering or the $100K+ for the tools...
Yep, some FPGA's are cheap enough that they can be used for both prototyping and production. But comparing one of the least expensive FPGA's in the world with a near maximum "up to" price for ASIC development is not useful. And yes, I'm an FPGA engineer as well (waiting impatiently for 7.1.2i to be released), so I'm not saying this for my love of ASIC's.
Competitive ASIC's can be developed for WELL under $1M (probably even well under $500k for a good number of devices) including tools and NRE/prototype costs. Still doesn't remove the fact that it takes considerable volume to come out ahead against a $7 FPGA. Against a 2VP50, at closing in on $1k a piece, it's a completely different story.
Those with four-wheel-drive vehicles (that actually leave the pavement) know that sometimes after getting stuck, you can move the steering wheel from side to side as a way of trying to gain traction from the sides of the rut you're in.
And sometimes moving the wheels from side to side digs a deeper hole. Seems like it would all depend on how much traction there is to be had - and if it is surrounded by fine grain sand, I don't hold high hopes.
While neat in theory, and useful in certain applications, in general there are a few practical problems with making "many devices share the low-voltage supply":
1. Current flow goes up as voltage goes down (to get the same number of Watts). You don't want to be transmitting a high DC current because series resistance will eat your lunch: Current * Resistance = Voltage drop (aka V=IR, aka Ohm's law).
2. Following on #1, all the devices sharing one supply need to be relatively close to it.
3. Even for low current applications, different devices need different and sometimes multiple voltage rails. Do you supply them all, or just some of them make the target derive the others?
4. Following on both #1 and #3, DC voltage and more importantly, power requirements change over time, so in the end, you'd likely end up with what you have now... multiple DC supplies, some for older devices, some for newer devices.
Now, a number of these problems could be avoided if you used a high enough DC voltage (let's say 48V), but now you have a safety issue if high currents can be delivered, not to mention that each device would need to step down the 48V - so you end up with the same thing you have now.
> Its going to get worse though. As more people use it and when it goes out of beta and some > spammers can start getting accounts and testing...
While I wish spammers were that stupid (maybe they are on your planet), here on Earth basicly anyone that wants a gmail invite can get one by asking in almost any news group, forum, or mailing list. Or they can visit one of the many gmail invite web sites.
Rest assured they were some of the first ones to get gmail accounts and have been banging on the spam filter ever since. There will be no sudden spam spike when gmail goes fully public.
In #11109104, Cyryathron wrote >> >> What crazy things will they think of next? >> Power over Ethernet? >> Internet over Powerlines? > > Power over Wireless Ethernet > POW!
How about power over laser?
In the 1991 through 1993, I worked at Electrospace Systems, Inc (aka ESI) and one of their main businesses were secure phone systems for the government, and the government wanted a phone that would survive a HEMP event. Since you can't have any electrical cables running in or out of the thing, how do you power it?
The engineers immediately thought of using batteries, but after realizing how much power it would take to power the phone, they looked for alternative ideas. Remember, this is back before 3.3V LVCMOS, so most electronics tended to be pretty power hungry.
Then someone came up with the idea to recover the power from the received laser light! Sounded like the perfect solution until they realized how much laser power would be required. If someone were to disconnect the fiber and wave it around the room, serious eye damage (and possibly property damage?) would have resulted.
They went with the battery packs. Small lead-acid, as I recall, that would last 2 to 4 days, depending on usage.
It isn't hard to know what the CEO's of most public companies are paid.
The EA 2004 Proxy statement shows that CEO Probst is being paid $672k this year, plus $781k in bonus(es). Must be tough times, having to take a cut from last year when he made a combined $1.8M rather than $1.45M.
If the auther is really interested in trying to get some attention to the matter, they could attempt to contact the board of directors, for all good that would do.
As much as it pains me to, I must agree with your general theme that something is missing in Motorola's processor development. As a embedded hardware engineer, I've watched them stumble over themselves time and again on the PowerQUICC II:
- Just last year they reached core speeds they promised back in 2000 (or was it 1999?).
- PCI support was two years late (or was it three)?
- Power dissipation has been higher than expected.
- Some clock speeds require you to run a different voltage, while other other clock speeds don't work at all (if you use certain clock multipliers).
We still actively design in their parts because they are a perfect fit, but we don't trust them to deliver their next feature on time (last Oct they promised the 8270 and related devices would be in production by December... here we are in March and now they are promising May).
I hope they can get their act together, cause when they finally release a product, it works like a hose.
And I'm not kidding. Using dBm = 10 log10[ P / 1 mW], you get 27 dBm.
Most lasers in the telecommunications world run between -10 dBm and 5 dBm. Over a good fiber link, you can reach over 100km with a couple dBm.
EDFAs and Raman amplifiers may be up in the 20 or 30 dBm range, but they are not widely used, nor will they ever be. You only need that much power for very long runs - like between remote cities in the mid-West US.
Either I'm a spam processing machine, or some of these estimates are WAY overstated. After running through two filters, I end up only seeing 20 TO 40 spam's a day, and it takes me all of 20 or 30 seconds to deal with them - for the WHOLE DAY. Do these people keep their delete key in their drawer or what?
And the person quoted about the cost of setting up spam filters and following up on incorrect filtering seems to ignore the fact that the effort for this person to do this is spread across all the users... thousands of them (or tens or hundreds of thousands, in this case).
> That must be a big lab! Or maybe they had 100km of fibre > and they just looped it round and round and round.;)
Fiber without the colored "protective insulation" takes up surprisingly little space, and weighs next to nothing. 100km of fiber could be picked up by with one hand if mounted on single spool.
In our lab, we have four fiber spools (two 20km and two 40km) that can be connected together to create various distances. Each is mounted in a plastic case that is about a foot in diameter and 4 inches wide.
> And it's gibibit, not gigabit (unless it's actually 1 000 000 000 bits per sec)
Considering they are likely referring to a gigabit Ethernet connection, gigabit is correct.
GbE uses a 125.00 MHz (or divided version of that) reference clock, multiplied by ten, to transmit data with. When you remove the 25% 8B/10B overhead (how efficient is Ethernet again?), it effectively transmits a bit on every clock cycle of the 1000 MHz clock.
gibi (and kibi) are typically used with regard to things that have address lines (memory).
Slashdot Mirror
What is your opinion of the open source hardware projects that you are aware of, and do any stand chance of becoming success stories like so many of the software projects have?
Also, have you come across many business opportunities that could be filled by open source hardware (care the describe them?), or do companies shy away from open source hardware for some reason (if so, why?)?
Thank you!
And, FPGAs are really useful as an implementation, not just for prototyping. It can cost up to $30M to set up an ASIC line, not counting engineering or the $100K+ for the tools ...
Yep, some FPGA's are cheap enough that they can be used for both prototyping and production. But comparing one of the least expensive FPGA's in the world with a near maximum "up to" price for ASIC development is not useful. And yes, I'm an FPGA engineer as well (waiting impatiently for 7.1.2i to be released), so I'm not saying this for my love of ASIC's.
Competitive ASIC's can be developed for WELL under $1M (probably even well under $500k for a good number of devices) including tools and NRE/prototype costs. Still doesn't remove the fact that it takes considerable volume to come out ahead against a $7 FPGA. Against a 2VP50, at closing in on $1k a piece, it's a completely different story.
Those with four-wheel-drive vehicles (that actually leave the pavement) know that sometimes after getting stuck, you can move the steering wheel from side to side as a way of trying to gain traction from the sides of the rut you're in.
And sometimes moving the wheels from side to side digs a deeper hole. Seems like it would all depend on how much traction there is to be had - and if it is surrounded by fine grain sand, I don't hold high hopes.
It's been a great missions, regardless.
While neat in theory, and useful in certain applications, in general there are a few practical problems with making "many devices share the low-voltage supply":
1. Current flow goes up as voltage goes down (to get the same number of Watts). You don't want to be transmitting a high DC current because series resistance will eat your lunch: Current * Resistance = Voltage drop (aka V=IR, aka Ohm's law).
2. Following on #1, all the devices sharing one supply need to be relatively close to it.
3. Even for low current applications, different devices need different and sometimes multiple voltage rails. Do you supply them all, or just some of them make the target derive the others?
4. Following on both #1 and #3, DC voltage and more importantly, power requirements change over time, so in the end, you'd likely end up with what you have now... multiple DC supplies, some for older devices, some for newer devices.
Now, a number of these problems could be avoided if you used a high enough DC voltage (let's say 48V), but now you have a safety issue if high currents can be delivered, not to mention that each device would need to step down the 48V - so you end up with the same thing you have now.
And a "lesser size" means a HIGHER compression ratio... so it'll be even more than the "insane" 55:1, as you called it.
I meant quality at the current compression ratio. In Theory, MPEG 4 won't need a 55:1 compression ratio just because it will be smaller to begin with.
Huh? You just did it again. What is smaller to begin with? Certainly not the original stream - it is a known, fixed size.
[...]which means the MPEG2 signal is compressed somewhere between 49-55:1.. That's insane, and MPEG 4 will hopefully lessen the compression ratio.
Huh? Hopefully MPEG 4 will increase the compression ratio.
> Its going to get worse though. As more people use it and when it goes out of beta and some
> spammers can start getting accounts and testing...
While I wish spammers were that stupid (maybe they are on your planet), here on Earth basicly anyone that wants a gmail invite can get one by asking in almost any news group, forum, or mailing list. Or they can visit one of the many gmail invite web sites.
Rest assured they were some of the first ones to get gmail accounts and have been banging on the spam filter ever since. There will be no sudden spam spike when gmail goes fully public.
In #11109104, Cyryathron wrote
>>
>> What crazy things will they think of next?
>> Power over Ethernet?
>> Internet over Powerlines?
>
> Power over Wireless Ethernet
> POW!
How about power over laser?
In the 1991 through 1993, I worked at Electrospace Systems, Inc (aka ESI) and one of their main businesses were secure phone systems for the government, and the government wanted a phone that would survive a HEMP event. Since you can't have any electrical cables running in or out of the thing, how do you power it?
The engineers immediately thought of using batteries, but after realizing how much power it would take to power the phone, they looked for alternative ideas. Remember, this is back before 3.3V LVCMOS, so most electronics tended to be pretty power hungry.
Then someone came up with the idea to recover the power from the received laser light! Sounded like the perfect solution until they realized how much laser power would be required. If someone were to disconnect the fiber and wave it around the room, serious eye damage (and possibly property damage?) would have resulted.
They went with the battery packs. Small lead-acid, as I recall, that would last 2 to 4 days, depending on usage.
The EA 2004 Proxy statement shows that CEO Probst is being paid $672k this year, plus $781k in bonus(es). Must be tough times, having to take a cut from last year when he made a combined $1.8M rather than $1.45M.
If the auther is really interested in trying to get some attention to the matter, they could attempt to contact the board of directors, for all good that would do.
- Just last year they reached core speeds they promised back in 2000 (or was it 1999?).
- PCI support was two years late (or was it three)?
- Power dissipation has been higher than expected.
- Some clock speeds require you to run a different voltage, while other other clock speeds don't work at all (if you use certain clock multipliers).
We still actively design in their parts because they are a perfect fit, but we don't trust them to deliver their next feature on time (last Oct they promised the 8270 and related devices would be in production by December... here we are in March and now they are promising May). I hope they can get their act together, cause when they finally release a product, it works like a hose.
And I'm not kidding. Using dBm = 10 log10[ P / 1 mW], you get 27 dBm.
Most lasers in the telecommunications world run between -10 dBm and 5 dBm. Over a good fiber link, you can reach over 100km with a couple dBm.
EDFAs and Raman amplifiers may be up in the 20 or 30 dBm range, but they are not widely used, nor will they ever be. You only need that much power for very long runs - like between remote cities in the mid-West US.
Either I'm a spam processing machine, or some of these estimates are WAY overstated. After running through two filters, I end up only seeing 20 TO 40 spam's a day, and it takes me all of 20 or 30 seconds to deal with them - for the WHOLE DAY. Do these people keep their delete key in their drawer or what?
And the person quoted about the cost of setting up spam filters and following up on incorrect filtering seems to ignore the fact that the effort for this person to do this is spread across all the users... thousands of them (or tens or hundreds of thousands, in this case).
Marc
> That must be a big lab! Or maybe they had 100km of fibre ;)
> and they just looped it round and round and round.
Fiber without the colored "protective insulation" takes up surprisingly little space, and weighs next to nothing. 100km of fiber could be picked up by with one hand if mounted on single spool.
In our lab, we have four fiber spools (two 20km and two 40km) that can be connected together to create various distances. Each is mounted in a plastic case that is about a foot in diameter and 4 inches wide.
> And it's gibibit, not gigabit (unless it's actually 1 000 000 000 bits per sec)
Considering they are likely referring to a gigabit Ethernet connection, gigabit is correct.
GbE uses a 125.00 MHz (or divided version of that) reference clock, multiplied by ten, to transmit data with. When you remove the 25% 8B/10B overhead (how efficient is Ethernet again?), it effectively transmits a bit on every clock cycle of the 1000 MHz clock.
gibi (and kibi) are typically used with regard to things that have address lines (memory).