Yeah. I keep telling the scientists where I work, "Do your visualization on Macs; there's no better platform. But do your compute on Linux or one of the mono-kernel BSDs." You're absolutely right that heavy real-time threading tends to make a desktop "feel" faster, but that's simply UI responsiveness. If you want to do heavy compute - where context switching incurs a heavy toll on output - then, a monolithic kernel is still the best approach.
A good analogy is the difference between bandwidth and latency. Microkernels, like as is used with MacOS X, offer the best latency because the message passing allows for very fast interrupts. Just as are monolithic kernels better optimized for long tasks, with as few interrupts as necessary.
You'll note that on the low end I listed a Sun 3/80 (or sparc 1) with a cg24 24 bit color card. That would have been manufactured from ~1988 or ~1989, and would have had on the high end no more than 16mb to 32mb of RAM. Most shipped with 8mb.
And BTW: color support has nothing to do with window management. At the time the two competing toolkits were Athena (from the original MIT X11 and X10 distribution), Sun Openlook toolkit, and Motif. These widget toolkits are better compared to modern day GTK+ and QT toolkits in use on GNOME and KDE respectively.
And BTW: I own (and am typing on right now) a 2Ghz Macbook w/ 2GB of RAM. It's a fine machine. But a graphical unix can run all the way down to 4mb of RAM, which twenty years ago was a *huge* amount of memory.
That's a fair point. However, you must admit that within the context of Aqua+DPDF vs. X11 for performance on old hardware, to bring up the historical significance of DPS on NeXTStep is fair. DPS worked great on that old hardware - just as did X11. Which is not to diminish the advances of Aqua on modern 3D accelerated hardware.
An old g3 with a good Rage 128 might have enough horsepower to display an mpeg. But step down to older hardware, and forget it. 2D acceleration for lines and fills was the standard for NuBus video cards from that era (early to late nineties) And before that it was just a flat frame buffer.
This is a great point. Kids today... they just don't realize how good they've got it. I remember Sun 2s having VME 1mb RAM boards populated with 21256 (256K) chips. Work it out, that's 9 chips (parity) per 256K, or 36 chips just for 1MB of RAM. Add on a bunch of TTL and components to fill the board out and you've got one hell of a heat generator. Also, the damn machine was useless without at least 2MB, and really only useful with 4MB (especially if you added a megapixel display).
And that computer did useful work. Back in the day when you were lucky to get 256K - 512K in a desktop computer (and, yes - PCs really maxed out at 640K).
Why do you need 3D acceleration on an old Mac? Most of them never had 3D hardware support anyway. 2D acceleration is perfectly fine for good X11 performance.
What are you smoking? DPS (and the more recent OS X variant DPDF) goes back to 1988, with the introduction of NeXTSTep on the NeXT cube. Mac OS X is really just NeXTStep. Aqua, however, is an Apple addition and - IMO - is a real improvement. But it's also a resource hog.
Mach is a resource hog. But that's not the fault of Aqua. However, Aqua is also a huge resource hog, without much benefit if you only have 2d video acceleration. Might as well just run X11 in that case. And, if you're going to do that - might as well just run Linux or NetBSD as monolithic kernels tend to run much faster.
That's because Mach spends a shit of of resources passing messages and servicing real-time interrupts. That has nothing to do with the difference between X11 vs. Aqua in terms of resource consumption. Stick Yellow Dog or NetBSD on that old G3 if you want performance.
I remember running a full 'nix with X11 and TWM with 4mb of RAM on a Sun 3/80. If you wanted color, a Sun 3/60 could handle it with 8mb; 16mb would give you a "screaming" 4 mips pizza box. When the Sun 3/80s and Sparc 1's came out, a 32mb system with a cg24 sbus card could get you full 24bit color with a megapixel display. And it had plenty of RAM to do real work.
Compare that with a 128mb or 256mb G3 CRT iMac and you've got way more than enough ram and CPU horsepower to run X11 with plenty of useful apps. Christ, I ran X11 on a 486 with 8mb of RAM and a 512kb XVGA card back in 1994 and it worked just fine. (And BTW: NeXTStep on an old cube ran DPS just great in 16mb of RAM too. It's not DPDF that's the hog - it's Aqua).
Re:Paris Hilton or Madame Curie... hmmm
on
Top Ten Geek Girls
·
· Score: 1
Well... the sad truth is that Ms. Franklin died of ovarian cancer in 1958, whereas Crick, Watson, and Wilkins were awarded their Nobel Prizes in 1962. The Nobel Foundation does not award the Nobel Prize posthumously. Thus, she could never have won. But, there is no doubt that Watson's The Double Helix took great liberties in smearing Ms. Franklin unnecessarily.
Paris Hilton or Madame Curie... hmmm
on
Top Ten Geek Girls
·
· Score: 5, Insightful
Did I just see Madame Curie and Rosalind Franklin compared with Paris Hilton and Lisa Simpson? One two time Nobel Prize winner and another near Nobel Prize winner compared to a coke snorting self promoting gamer and a cartoon character.
According to joystik, Sony didn't meet its target of 400,000 units on US launch but instead only shipped 200,000. I've been watching eBay prices decline since launch. At first they spiked up, though it's unclear how many of those very high priced sales were legitimate. Sham bidding of eBay accounts with zero feedback is rampant. However, within a day prices declined to about double retail, and have now declined even further. The average price for a 60GB PS3 is ~$1100 while a 20GB model can be had for about $900. As to volume of units on eBay as a fraction of the total release, I note that at the beginning I saw at least 25,000 units on sale. Now there are about 10,000 units. This represents a range of 10% on Friday and Saturday to 5% of the total release today. That's a *very* large fraction of units up for resale; note that it tells you not what the total fraction of units resold, but only gives snapshots as an indicator.
Double retail is a poor return. If you consider that most people waited in line at least 24 hours in the cold, a double markup represents ~$25/hr. For those who waited 48 hours, it (obviously) drops to $12.50/hr. And, as prices continue to slide, that hourly wage on the markup declines as well. Then one must add eBay fees, shipping hassles, etc. This ignores those who somehow obtained large numbers of units for resale - but they have other problems as prices decline, particularly since they have so much capital tied up in a rapidly depreciating asset.
I note that the xbox 360 stayed above double markup on eBay for well over six weeks after release - well into January. And supplied were very tight well into February. That, for a game machine that shipped with more units than the PS3 on launch. This suggests to me a very lax demand for the PS3. Those who bought in expecting to flip units for triple to five times retail markup are in for quite the disappointment. I predict we will see 1.5 x retail on eBay before Christmas. Further, I predict one will be able to walk into a store and buy a PS3 retail without trouble early after Christmas - and, if demand is really lax - perhaps even before.
I think this indicates a *terrible* launch for Sony and the PS3.
a) there remains enough demand to maintain price above retail. However, as noted, the price continues to fall. You can watch this happen in real time. Thus, a single price snapshot at $1200 is about as worthless a future price indicator as one at $3500.
b) Many sellers are still abusing sham accounts to prop up bogus sale prices so they can relist. Presumably on the hope that it will later sell for more. It probably won't.
But there must be a floor to the price drop, and that floor almost certainly will have to be higher than retail. Sellers spent a great deal of time and energy obtaining those PS3s on launch day. That represents human investment that sellers expect to generate a return from. The divergence between retail and sale price represents the "value" of their time. Assuming a 24 hour wait in line, that means double-retail is worth about $25/hr. 1.5 times retail, is thus worth about $12.50/hr. Then you have to add in eBay costs, packaging for shipment, etc.
Also, I note that the 360 stayed above double markup or above for well over a month after release. The PS3 is hitting that point within a day after release, and is even more constrained for supply than the 360 was. This is an indicator of much less demand then during the launch of the 360.
What you have seen is the effect of too many players in a speculative market.
Oh yeah. This is the clearest example of a modern day tulip mania I've seen. Black Monday back in '87 might also compare. What's striking about though is not the divergence between price expectations among sellers and buyers, but the extreme speed of this market correction. I mean, refresh eBay and you'll see equilibrium slide in real time. Pretty damn cool!
Check ebay. Prices have already crashed to ~$1200 for a 60GB unit, and below $1000 for the 20GB model. Further there are over 25000 units listed and no where near enough buyers. Folks who listed with a buy it now at $3000 - $3500 are going to be sorely disappointed. I suspect that we'll see equilibrium well below twice retail, and possibly less than 1.5 times within a week.
So, how long before I can walk into a Toys 'R Us or Target and pick one off the shelf? Perhaps even before Christmas. This is the most amazing example of an economic boom to bust I've ever seen.
I'm not suggesting people go take out big loans just to re-invest... I'm talking about a situation where you have 30k in cash already and you invest it and then take a 30K loan out against it to buy a car instead of dropping 30k on the car up front.
One more point: money is fungible. It makes no difference whether you take a loan out to buy something while investing the same amount of cash on hand elsewhere vs. simply borrowing the money to invest.
The difference is that if you have 30k cash that you can afford to part with right now then you're not going to go broke if you lose some of it in market fluctuations. It just means having to pay your car loan from your regular income like everybody else instead of your investment income.
Right. Instead, it means you run the risk of losing that $30K investment if the bank calls your loan. This could be due to having missed a loan payment or market losses of your investment that break whatever collateral terms you have for the loan. You risk this while the primary asset - the car - depreciates over time. That is, the collateral soon becomes worth more than the asset (car) itself. And if you miss a payment - *boom!* - the bank wants its collateral. Not the car, but your $30K investment. Have you ever been laid off?
Frankly, I'd rather take out a standard car loan, with the car as collateral, than twist my finances up like you suggest. But - hey - call me a tinfoil hatter if you like. We clearly have different standards for acceptable risk.
Yeah, I understand that. I just wanted him to say it directly. Leverage is great when used to buy an income generating asset with greater cash flow than operating costs plus debt servicing payments. The type of leverage these guys recommend is - IMO - quite risky, however.
The virtue of rental housing is that rental rates are inversly proportional to interest rates. I can refinance down if the interest rates drop, and I can charge higher rates when the interest rates rise. Life is good on both ends of the spectrum.
I own a two-fam near Harvard University. So, when it comes to borrowing for an income generating asset, especially one that tends to appreciate at or above inflation - I'm with you. But that's a different game from borrowing and investing those funds in a financial asset to offset interest payments elsewhere on a depreciating asset. Especially when cash is available for the primary purchase.
And yeah, just hold those 4plexes past this housing downturn and you'll be golden. Good strategy IMO...
Slashdot Mirror
Yeah. I keep telling the scientists where I work, "Do your visualization on Macs; there's no better platform. But do your compute on Linux or one of the mono-kernel BSDs." You're absolutely right that heavy real-time threading tends to make a desktop "feel" faster, but that's simply UI responsiveness. If you want to do heavy compute - where context switching incurs a heavy toll on output - then, a monolithic kernel is still the best approach.
A good analogy is the difference between bandwidth and latency. Microkernels, like as is used with MacOS X, offer the best latency because the message passing allows for very fast interrupts. Just as are monolithic kernels better optimized for long tasks, with as few interrupts as necessary.
Tailor your software to meet your needs.
You'll note that on the low end I listed a Sun 3/80 (or sparc 1) with a cg24 24 bit color card. That would have been manufactured from ~1988 or ~1989, and would have had on the high end no more than 16mb to 32mb of RAM. Most shipped with 8mb.
And BTW: color support has nothing to do with window management. At the time the two competing toolkits were Athena (from the original MIT X11 and X10 distribution), Sun Openlook toolkit, and Motif. These widget toolkits are better compared to modern day GTK+ and QT toolkits in use on GNOME and KDE respectively.
And BTW: I own (and am typing on right now) a 2Ghz Macbook w/ 2GB of RAM. It's a fine machine. But a graphical unix can run all the way down to 4mb of RAM, which twenty years ago was a *huge* amount of memory.
That's a fair point. However, you must admit that within the context of Aqua+DPDF vs. X11 for performance on old hardware, to bring up the historical significance of DPS on NeXTStep is fair. DPS worked great on that old hardware - just as did X11. Which is not to diminish the advances of Aqua on modern 3D accelerated hardware.
An old g3 with a good Rage 128 might have enough horsepower to display an mpeg. But step down to older hardware, and forget it. 2D acceleration for lines and fills was the standard for NuBus video cards from that era (early to late nineties) And before that it was just a flat frame buffer.
sigh, I really should preview more often.
This is a great point. Kids today... they just don't realize how good they've got it. I remember Sun 2s having VME 1mb RAM boards populated with 21256 (256K) chips. Work it out, that's 9 chips (parity) per 256K, or 36 chips just for 1MB of RAM. Add on a bunch of TTL and components to fill the board out and you've got one hell of a heat generator. Also, the damn machine was useless without at least 2MB, and really only useful with 4MB (especially if you added a megapixel display).
And that computer did useful work. Back in the day when you were lucky to get 256K - 512K in a desktop computer (and, yes - PCs really maxed out at 640K).
Why do you need 3D acceleration on an old Mac? Most of them never had 3D hardware support anyway. 2D acceleration is perfectly fine for good X11 performance.
What are you smoking? DPS (and the more recent OS X variant DPDF) goes back to 1988, with the introduction of NeXTSTep on the NeXT cube. Mac OS X is really just NeXTStep. Aqua, however, is an Apple addition and - IMO - is a real improvement. But it's also a resource hog.
That's true. And they'll even run NetBSD! 'Cause NetBSD will run on your freak'n toaster, with enough loving care and attention. :P
Every Mac going back to the old Beige PPC 601 systems had 2D acceleration support.
Mach is a resource hog. But that's not the fault of Aqua. However, Aqua is also a huge resource hog, without much benefit if you only have 2d video acceleration. Might as well just run X11 in that case. And, if you're going to do that - might as well just run Linux or NetBSD as monolithic kernels tend to run much faster.
That's because Mach spends a shit of of resources passing messages and servicing real-time interrupts. That has nothing to do with the difference between X11 vs. Aqua in terms of resource consumption. Stick Yellow Dog or NetBSD on that old G3 if you want performance.
I remember running a full 'nix with X11 and TWM with 4mb of RAM on a Sun 3/80.
s/80/50/
Bullshit.
I remember running a full 'nix with X11 and TWM with 4mb of RAM on a Sun 3/80. If you wanted color, a Sun 3/60 could handle it with 8mb; 16mb would give you a "screaming" 4 mips pizza box. When the Sun 3/80s and Sparc 1's came out, a 32mb system with a cg24 sbus card could get you full 24bit color with a megapixel display. And it had plenty of RAM to do real work.
Compare that with a 128mb or 256mb G3 CRT iMac and you've got way more than enough ram and CPU horsepower to run X11 with plenty of useful apps. Christ, I ran X11 on a 486 with 8mb of RAM and a 512kb XVGA card back in 1994 and it worked just fine. (And BTW: NeXTStep on an old cube ran DPS just great in 16mb of RAM too. It's not DPDF that's the hog - it's Aqua).
Right here on slashdot!
Well... the sad truth is that Ms. Franklin died of ovarian cancer in 1958, whereas Crick, Watson, and Wilkins were awarded their Nobel Prizes in 1962. The Nobel Foundation does not award the Nobel Prize posthumously. Thus, she could never have won. But, there is no doubt that Watson's The Double Helix took great liberties in smearing Ms. Franklin unnecessarily.
Did I just see Madame Curie and Rosalind Franklin compared with Paris Hilton and Lisa Simpson? One two time Nobel Prize winner and another near Nobel Prize winner compared to a coke snorting self promoting gamer and a cartoon character.
I give up.
According to joystik, Sony didn't meet its target of 400,000 units on US launch but instead only shipped 200,000. I've been watching eBay prices decline since launch. At first they spiked up, though it's unclear how many of those very high priced sales were legitimate. Sham bidding of eBay accounts with zero feedback is rampant. However, within a day prices declined to about double retail, and have now declined even further. The average price for a 60GB PS3 is ~$1100 while a 20GB model can be had for about $900. As to volume of units on eBay as a fraction of the total release, I note that at the beginning I saw at least 25,000 units on sale. Now there are about 10,000 units. This represents a range of 10% on Friday and Saturday to 5% of the total release today. That's a *very* large fraction of units up for resale; note that it tells you not what the total fraction of units resold, but only gives snapshots as an indicator.
Double retail is a poor return. If you consider that most people waited in line at least 24 hours in the cold, a double markup represents ~$25/hr. For those who waited 48 hours, it (obviously) drops to $12.50/hr. And, as prices continue to slide, that hourly wage on the markup declines as well. Then one must add eBay fees, shipping hassles, etc. This ignores those who somehow obtained large numbers of units for resale - but they have other problems as prices decline, particularly since they have so much capital tied up in a rapidly depreciating asset.
I note that the xbox 360 stayed above double markup on eBay for well over six weeks after release - well into January. And supplied were very tight well into February. That, for a game machine that shipped with more units than the PS3 on launch. This suggests to me a very lax demand for the PS3. Those who bought in expecting to flip units for triple to five times retail markup are in for quite the disappointment. I predict we will see 1.5 x retail on eBay before Christmas. Further, I predict one will be able to walk into a store and buy a PS3 retail without trouble early after Christmas - and, if demand is really lax - perhaps even before.
I think this indicates a *terrible* launch for Sony and the PS3.
... why are they selling for over 1K?
Two reasons:
a) there remains enough demand to maintain price above retail. However, as noted, the price continues to fall. You can watch this happen in real time. Thus, a single price snapshot at $1200 is about as worthless a future price indicator as one at $3500.
b) Many sellers are still abusing sham accounts to prop up bogus sale prices so they can relist. Presumably on the hope that it will later sell for more. It probably won't.
But there must be a floor to the price drop, and that floor almost certainly will have to be higher than retail. Sellers spent a great deal of time and energy obtaining those PS3s on launch day. That represents human investment that sellers expect to generate a return from. The divergence between retail and sale price represents the "value" of their time. Assuming a 24 hour wait in line, that means double-retail is worth about $25/hr. 1.5 times retail, is thus worth about $12.50/hr. Then you have to add in eBay costs, packaging for shipment, etc.
Also, I note that the 360 stayed above double markup or above for well over a month after release. The PS3 is hitting that point within a day after release, and is even more constrained for supply than the 360 was. This is an indicator of much less demand then during the launch of the 360.
What you have seen is the effect of too many players in a speculative market.
Oh yeah. This is the clearest example of a modern day tulip mania I've seen. Black Monday back in '87 might also compare. What's striking about though is not the divergence between price expectations among sellers and buyers, but the extreme speed of this market correction. I mean, refresh eBay and you'll see equilibrium slide in real time. Pretty damn cool!
Check ebay. Prices have already crashed to ~$1200 for a 60GB unit, and below $1000 for the 20GB model. Further there are over 25000 units listed and no where near enough buyers. Folks who listed with a buy it now at $3000 - $3500 are going to be sorely disappointed. I suspect that we'll see equilibrium well below twice retail, and possibly less than 1.5 times within a week.
So, how long before I can walk into a Toys 'R Us or Target and pick one off the shelf? Perhaps even before Christmas. This is the most amazing example of an economic boom to bust I've ever seen.
I'm not suggesting people go take out big loans just to re-invest ... I'm talking about a situation where you have 30k in cash already and you invest it and then take a 30K loan out against it to buy a car instead of dropping 30k on the car up front.
One more point: money is fungible. It makes no difference whether you take a loan out to buy something while investing the same amount of cash on hand elsewhere vs. simply borrowing the money to invest.
The difference is that if you have 30k cash that you can afford to part with right now then you're not going to go broke if you lose some of it in market fluctuations. It just means having to pay your car loan from your regular income like everybody else instead of your investment income.
Right. Instead, it means you run the risk of losing that $30K investment if the bank calls your loan. This could be due to having missed a loan payment or market losses of your investment that break whatever collateral terms you have for the loan. You risk this while the primary asset - the car - depreciates over time. That is, the collateral soon becomes worth more than the asset (car) itself. And if you miss a payment - *boom!* - the bank wants its collateral. Not the car, but your $30K investment. Have you ever been laid off?
Frankly, I'd rather take out a standard car loan, with the car as collateral, than twist my finances up like you suggest. But - hey - call me a tinfoil hatter if you like. We clearly have different standards for acceptable risk.
Yeah, I understand that. I just wanted him to say it directly. Leverage is great when used to buy an income generating asset with greater cash flow than operating costs plus debt servicing payments. The type of leverage these guys recommend is - IMO - quite risky, however.
The virtue of rental housing is that rental rates are inversly proportional to interest rates. I can refinance down if the interest rates drop, and I can charge higher rates when the interest rates rise. Life is good on both ends of the spectrum.
I own a two-fam near Harvard University. So, when it comes to borrowing for an income generating asset, especially one that tends to appreciate at or above inflation - I'm with you. But that's a different game from borrowing and investing those funds in a financial asset to offset interest payments elsewhere on a depreciating asset. Especially when cash is available for the primary purchase.
And yeah, just hold those 4plexes past this housing downturn and you'll be golden. Good strategy IMO...