My 4-month old daughter decided to wake up at 12:30am, and I went out and watched Leo for about 20 minutes. I saw 6 Leonid meteors in that time (and two non-Leonid).
I realize that as we enter the debris path this number will rise, but I thought people might be curious what last night was like.
...which is real-time 3D raytracing (or something similar) at 60 frames per second, 32bits per pixel, at 1600x1200 resolution. Although personally, I would go for 1024x768 resolution.
Your analogy with sound cards is a good one except that sound cards have pretty much achieved their goal: essential 20 or more digital channels, 44.1kHz stereo or four speaker out. With 3D video cards, I personally think we are still five or more years off for the goal.
Currently real-time rendering in 3D games and applications is done with relatively low polygon counts, shadows are done using a hack (if they are done at all), reflections are also done using a hack. Shapes are confined to rectangular solids, or now, bezel curves. Ideally, the the gaming developers and players would like real-time rendering at high resolutions using extremely high polygon counts (infinite would be ideal), and realistic shadows, reflections and shading. We are a long way from this goal. Look at the very best rendered still images on PC, and then compare them to a game like realtime Quake3 (pretty much state-of-the-art for PC's). There's a huge difference.
Sound card technology is essentially commoditized because sound cards have no more technical hurdles to complete. 3D video cards are 3-5 years away from this state, and big leaps forward in technology (such as the Voodoo5) are exciting because they take us one step closer to that goal.
This is a great posting - extremely well written, interesting and exactly on topic. It's been up (as of writing this) for over 4 hours now. Does anyone moderate it up? No... Instead at the topic we have an "insightful" poorly written comment on how pointless the search for fusion is. You have to love moderation... mbkennel, if I had any moderation points, I'd slap your posting up to the top. Thanks for the information.
It will certainly need more than one molecule to store one bit since the flipping the bit due to radiation (cosmic or otherwise) would probably be pretty easy with single molecule memory. It's a real problem with memory today. So, they would probably need triple redundancy - or three bits.
Then there's the problem of wiring up this memory, addressing it, and there's no discussion performance versus current memories.
Also, any speculation of having working implementations in the next few years needs to do a reality check versus the real time it takes to investigate new technology.
Re:'c' isn't the problem. Transmission line delay
on
Pentium III hits 1Ghz
·
· Score: 1
Not that I think anybody reads posts to old stories here at Slashdot, but one point is that copper interconnect on both of the current techniques, doesn't really lower resistance. The resistance stays the same - the technique has difficulty making tall thin wires (which are the way current lower-level wires look like), so instead the copper process reduces the thickness/height of the wire by a little less than half. Thus the capacitance of the wires drops while the resistance essentially stays the same.
No, I don't work for them... I have a brokerage account with WR Hambrect and they sent me an email last week saying that they were participating in the IPO. BTW, these are the same guys that are the lead underwriter in the "OpenIPO" of Andover.Net.
WRONG. I have been in quite a few IPO's and when you are really IN on the IPO, you can't sell until the SEC lockout period ends. That period is usually about 6 months, at which point, if you work for the company you still can't sell all of it.
David, you're coming on pretty strong. Relax.
This is true if you work for the company involved in the IPO or its subsidaries, but it's different if you are going through a brokerage. While what you are saying is technically true, the reality is less black and white. If you participate in an IPO through a brokerage, they request that you hold your shares until specified period. They can't stop you, however, if you decide to sell before that.
For example, the following is taken from Waterhouse Securities IPO FAQ:
"How long should I hold shares I purchased in an IPO?
Registered customers who have track records for buying and holding IPO shares in their account beyond 60 days will have priority in future IPO offerings over other customers. TD Waterhouse encourages this buy-and-hold approach for IPO investing. Selling shares within the 60-day period may prevent your from participating in future IPO offerings. "
My accounts at E*Trade and WR Hambrect have similar sounding wording.
My point is that you can be "in" on an IPO and still sell on the same day as the IPO - it's just not encouraged.
My guess is somewhere around 50nm quantum issues will start to serious crop up, so this is about 4-5 process generations from todays 180nm (0.18um). OTOH, life will start to get very difficult somewhere around 0.08nm because channel leakage (current though the transistor when the transistor is supposed to be turned off) will make it very hard to use many of the common circuit designs used in today's CPU's and will increase power to the point where cooling will be a serious issue.
GaAs, as other posters pointed out, is alive and well in some specialized applications. I have no idea where the semiconductor industry will be when we hit the wall around 50nm, but I don't think GaAs will be the answer - any more than Cu interconnect does much more than get us a little further along the performance curve.
Personally, my guess will be some form of stacked 3D semicondcutor process based on a 0.12um technology. Just being able to stack two transistors on top of each other (an n-FET and a p-FET) would cut chip area massively (due to the elimination of extra space between wells). But that's just my guess. Optical interconnect would be a big plus - an optical clock with really low skew would improve performance today by 10-20%.
Pentiums have an internal PLL which generates the clock, and one or two pins (depends on the model) which determines the multiplier to use on the PLL. In theory, if you were good enough to be able to solder off the old CPU and solder in a new CPU, you could tie off the bus fraction pins (BF0 and BF1) of the new Pentium CPU and set it to whatever clock multipler you wanted. The soldering would be the hard bit. Hardwiring the BF pins would be easy after that.
In the unlikely event that the processor is not soldered to the board, setting the BF pins to override the laptop's board setting would easy with a pair of wire cutters and a wirewrapping tool.
A 750MHz FSB? What are they planning on using to do this miracle? On-chip fiber optic? Current technology tops out at a 200MHz FSB and Motorola is going to try and achieve a 3.5X improvement in the state of the art in two years? I think not. Most likely this will be running at 2GHz on a 266-300MHz FSB - although possibly this 266MHz FSB could be double-pumped.
If you can cite something - anything - to back up this claim I would be extremely interested in reading about it.
Andover is using the OpenIPO system - which is essentially a Dutch auction IPO system designed to stop the volatility inherent the "classic" IPO where money is left on the table to generate interest in the stock and drive up the price. Judging from past OpenIPO IPO performances (Salon.Com, and Ravenswood winery) there will be very little movement in the stock once it's actually on the market due to the stock being auctioned to the public in the pre-IPO.
I totally support the OpenIPO model (and have had an account with WR Hambrect since the start) and I wish Andover the best of luck, but if anyone is trying to get in on the pre-IPO through OpenIPO in the hopes of making money through point "B." in the above post, then they will almost certainly be disappointed.
Actually, they do quote ByteMarks on the page. And now there's new tactic of taking Intel's benchmarks, and throwing out the scores that don't show what they want them to show.
Taken from the bottom of the page:
"The 500MHz G4 Processor, with Velocity Engine, is an average of 2.94 times as fast as the fastest Pentium III (600MHz)
Based on BYTEmark integer index processor scores."
I sincerely doubt that the G4 SpecInt and SpecFP (which they don't quote strangely enough) is twice as fast as the Pentium III's.
OTOH, that LCD monitor sure looks great. I wonder where and when I can get one of those - and more importantly how much will it cost.
I'll echo the rest of the crowd. Tom's site is slow, hard to read, full of arrogant attitude, and occassionally he posts inaccurate misinformation and unproven rumors as if they were factual. I liked Tom's site a few years ago, but in the last year he seems to spend more time flaming people and talking about prime vacation spots than posting useful hardware info. This is an interesting story, but I'm unimpressed with the fact that most people seem to taken this unproven rumor as fact despite the unreliable source. I like Anand's. http://www.anandtech.com/ Thresh's Firing Squad does very good hardware reviews, but they don't do as many as I would like. http://www.firingsquad.com SharkyExtreme is good - although they, too, have a tendancy to print rumors ahead of facts. http://www.sharkyextreme.com Agn3D's hardware site has pretty good hardware reviews - although not, IMO, the quality of those sites listed above. http://www.agnhardware.com Then there's plenty of smaller sites that don't update as regularly that still contain useful hardware information... lostcircuits, jeff's hardware site. There's plenty of links to these sites and more at Anand's site.
Have you ever seen a 24-year old try and rent a car? I believe most businesses have the sign "we reserve the right to refuse business to customers at our discretion" or something like that.
Discrimination on the basis of age is illegal when you are applying for housing, loans or a job - not when you are buying tickets to a movie.
Slashdot Mirror
My 4-month old daughter decided to wake up at 12:30am, and I went out and watched Leo for about 20 minutes. I saw 6 Leonid meteors in that time (and two non-Leonid).
I realize that as we enter the debris path this number will rise, but I thought people might be curious what last night was like.
...which is real-time 3D raytracing (or something similar) at 60 frames per second, 32bits per pixel, at 1600x1200 resolution. Although personally, I would go for 1024x768 resolution.
Your analogy with sound cards is a good one except that sound cards have pretty much achieved their goal: essential 20 or more digital channels, 44.1kHz stereo or four speaker out. With 3D video cards, I personally think we are still five or more years off for the goal.
Currently real-time rendering in 3D games and applications is done with relatively low polygon counts, shadows are done using a hack (if they are done at all), reflections are also done using a hack. Shapes are confined to rectangular solids, or now, bezel curves. Ideally, the the gaming developers and players would like real-time rendering at high resolutions using extremely high polygon counts (infinite would be ideal), and realistic shadows, reflections and shading. We are a long way from this goal. Look at the very best rendered still images on PC, and then compare them to a game like realtime Quake3 (pretty much state-of-the-art for PC's). There's a huge difference.
Sound card technology is essentially commoditized because sound cards have no more technical hurdles to complete. 3D video cards are 3-5 years away from this state, and big leaps forward in technology (such as the Voodoo5) are exciting because they take us one step closer to that goal.
This is a great posting - extremely well written, interesting and exactly on topic. It's been up (as of writing this) for over 4 hours now. Does anyone moderate it up? No... Instead at the topic we have an "insightful" poorly written comment on how pointless the search for fusion is. You have to love moderation... mbkennel, if I had any moderation points, I'd slap your posting up to the top. Thanks for the information.
It will certainly need more than one molecule to store one bit since the flipping the bit due to radiation (cosmic or otherwise) would probably be pretty easy with single molecule memory. It's a real problem with memory today. So, they would probably need triple redundancy - or three bits.
Then there's the problem of wiring up this memory, addressing it, and there's no discussion performance versus current memories.
Also, any speculation of having working implementations in the next few years needs to do a reality check versus the real time it takes to investigate new technology.
Not that I think anybody reads posts to old stories here at Slashdot, but one point is that
copper interconnect on both of the current techniques, doesn't really lower resistance. The resistance stays the same - the technique has difficulty making tall thin wires (which are the way current lower-level wires look like), so instead the copper process reduces the thickness/height of the wire by a little less than half. Thus the capacitance of the wires drops while the resistance essentially stays the same.
No, I don't work for them... I have a brokerage account with WR Hambrect and they sent me an email last week saying that they were participating in the IPO. BTW, these are the same guys that are the lead underwriter in the "OpenIPO" of Andover.Net.
http://www.wrhambrecht.com/
WRONG. I have been in quite a few IPO's and when you are really IN on the IPO, you can't sell until the SEC lockout period ends. That period is usually about 6 months, at which point, if you work for the company you still can't sell all of it.
David, you're coming on pretty strong. Relax.
This is true if you work for the company involved in the IPO or its subsidaries, but it's different if you are going through a brokerage. While what you are saying is technically true, the reality is less black and white. If you participate in an IPO through a brokerage, they request that you hold your shares until specified period. They can't stop you, however, if you decide to sell before that.
For example, the following is taken from Waterhouse Securities IPO FAQ:
"How long should I hold shares I purchased in an IPO?
Registered customers who have track records for buying and holding IPO shares in their account beyond 60 days will have priority in future IPO offerings over other customers. TD Waterhouse encourages this buy-and-hold approach for IPO investing. Selling shares within the 60-day period may prevent your from participating in future IPO offerings. "
My accounts at E*Trade and WR Hambrect have similar sounding wording.
My point is that you can be "in" on an IPO and still sell on the same day as the IPO - it's just not encouraged.
My guess is somewhere around 50nm quantum issues will start to serious crop up, so this is about 4-5 process generations from todays 180nm (0.18um). OTOH, life will start to get very difficult somewhere around 0.08nm because channel leakage (current though the transistor when the transistor is supposed to be turned off) will make it very hard to use many of the common circuit designs used in today's CPU's and will increase power to the point where cooling will be a serious issue.
GaAs, as other posters pointed out, is alive and well in some specialized applications. I have no idea where the semiconductor industry will be when we hit the wall around 50nm, but I don't think GaAs will be the answer - any more than Cu interconnect does much more than get us a little further along the performance curve.
Personally, my guess will be some form of stacked 3D semicondcutor process based on a 0.12um technology. Just being able to stack two transistors on top of each other (an n-FET and a p-FET) would cut chip area massively (due to the elimination of extra space between wells). But that's just my guess. Optical interconnect would be a big plus - an optical clock with really low skew would improve performance today by 10-20%.
You wouldn't need to really jumper up the clock.
Pentiums have an internal PLL which generates the clock, and one or two pins (depends on the model) which determines the multiplier to use on the PLL. In theory, if you were good enough to be able to solder off the old CPU and solder in a new CPU, you could tie off the bus fraction pins (BF0 and BF1) of the new Pentium CPU and set it to whatever clock multipler you wanted. The soldering would be the hard bit. Hardwiring the BF pins would be easy after that.
In the unlikely event that the processor is not soldered to the board, setting the BF pins to override the laptop's board setting would easy with a pair of wire cutters and a wirewrapping tool.
A 750MHz FSB? What are they planning on using to do this miracle? On-chip fiber optic? Current technology tops out at a 200MHz FSB and Motorola is going to try and achieve a 3.5X improvement in the state of the art in two years? I think not. Most likely this will be running at 2GHz on a 266-300MHz FSB - although possibly this 266MHz FSB could be double-pumped.
If you can cite something - anything - to back up this claim I would be extremely interested in reading about it.
They are using OpenIPO - the pre-IPO shares are available to anyone who wants to bid on them using a Dutch auction system.
This is a different situation from the Redhat IPO where the pre-IPO shares are pre-allocated by the underwritting companies.
Go open an account with WR Hambrect and you can get the opportunity to bid on the shares before they are traded on the market.
I agree with you about RH versus Andover.
One minor point regarding "B.".
Andover is using the OpenIPO system - which is essentially a Dutch auction IPO system designed to stop the volatility inherent the "classic" IPO where money is left on the table to generate interest in the stock and drive up the price. Judging from past OpenIPO IPO performances (Salon.Com, and Ravenswood winery) there will be very little movement in the stock once it's actually on the market due to the stock being auctioned to the public in the pre-IPO.
I totally support the OpenIPO model (and have had an account with WR Hambrect since the start) and I wish Andover the best of luck, but if anyone is trying to get in on the pre-IPO through OpenIPO in the hopes of making money through point "B." in the above post, then they will almost certainly be disappointed.
Actually, they do quote ByteMarks on the page. And now there's new tactic of taking Intel's benchmarks, and throwing out the scores that don't show what they want them to show.
Taken from the bottom of the page:
"The 500MHz G4 Processor, with Velocity Engine, is an average of 2.94 times as fast as the fastest Pentium III (600MHz)
Based on BYTEmark integer index processor scores."
I sincerely doubt that the G4 SpecInt and SpecFP (which they don't quote strangely enough) is twice as fast as the Pentium III's.
OTOH, that LCD monitor sure looks great. I wonder where and when I can get one of those - and more importantly how much will it cost.
I'll echo the rest of the crowd. Tom's site is slow, hard to read, full of arrogant attitude, and occassionally he posts inaccurate misinformation and unproven rumors as if they were factual. I liked Tom's site a few years ago, but in the last year he seems to spend more time flaming people and talking about prime vacation spots than posting useful hardware info. This is an interesting story, but I'm unimpressed with the fact that most people seem to taken this unproven rumor as fact despite the unreliable source. I like Anand's. http://www.anandtech.com/ Thresh's Firing Squad does very good hardware reviews, but they don't do as many as I would like. http://www.firingsquad.com SharkyExtreme is good - although they, too, have a tendancy to print rumors ahead of facts. http://www.sharkyextreme.com Agn3D's hardware site has pretty good hardware reviews - although not, IMO, the quality of those sites listed above. http://www.agnhardware.com Then there's plenty of smaller sites that don't update as regularly that still contain useful hardware information... lostcircuits, jeff's hardware site. There's plenty of links to these sites and more at Anand's site.
Have you ever seen a 24-year old try and rent a car? I believe most businesses have the sign "we reserve the right to refuse business to customers at our discretion" or something like that.
Discrimination on the basis of age is illegal when you are applying for housing, loans or a job - not when you are buying tickets to a movie.