Professional wedding photographers shoot 300+ pics per event and rarely get better than 25% that come out with any sort of quality
Well, there are photographers, and there are "photographers".
I know plenty in the "I'll shoot a bunch, and end up with 10% to 25% that are really good" category. On the other hand, I know a very few who take a few shots, and end up with 75%-90% that are really worth it.
Making a blanket statement like that about photographers is just silly. Of course, since you're talking about *wedding* photographers, maybe it's not quite as silly. : )
Many believe buying a better camera with greater megapixels etc will make them a better photographer.
I don't know a single person who thinks that.
On the other hand, since my camera only has about 1/3 of the resolution of my dye-sub printer, I know that I'll get higher-quality prints if I get a camera with more pixels. My photographic skills will, of course, be unchanged.
Sure, the pixels in the CCD are large, giving it better light sensitivity. And sure, it's got a nice lens. Terrific.
On the other hand, consider that (a) it cost hundreds of thousands of dollars just for the fuel to get the launch craft off of the ground, ignoring all of the personnel costs, cost of the rover, etc., and that (b) this thing was only designed to travel about 600 yards.
Given the enormous expense and extremely tiny travelling range, I would think that throwing an extra thousand bucks on a higher-res CCD and a "longer" (telephoto) lens would be a no-brainer. Can't travel to the mountains? At least you'd be able to *see* them.
Meeting the benchmarks is easy. Doing that cheaply is another matter!
The greatest point about a system like the one I described is that not only would it give extremely high performance for a relatively modest price, it could also be made to be very easy to use. That would probably sell a lot of them. Of course, that could potentially put a lot of admins out of business. : )
I thought that one of Transmeta's selling points was the the chips were relatively inexpensive, compared to the AMD/Intel chips, but I could be wrong. The chips probably use much less power mostly because they likely have far fewer transistors than a P4, Athlon, or Opteron. It still may be more than $2,000, but it would still likely be MUCH cheaper (say, an order of magnitude) than trying to do it by building a bunch of Intel/AMD machines. (more below)
I mentioned a billion-page per day number simply because so many people's web servers seem to buckle under a slashdotting load of tens or hundreds of thousands of hits per *hour*.
I'm guessing that it would probably be able to hit the million-per-minute mark with fewer processers, as these sound to be significantly better performers than StrongARMs.
(more info on doing it with Intel/AMD machines: I use 7 dual-CPU front-end machines for handling the Perl CGI and dishing out HTTP/HTTPS for my office. They'll handle ten million hits per day with relative ease. (actually, they'd handle MUCH more if it were static HTML, a lot of the CGI work is pretty intensive) However, it's expensive. 7 good-quality rack-mount chassis' don't come cheap. 7 motherboards, 7 hard drives, having 7 sets of memory. And the memory is mostly for disk cache, so I'm duplicating the cache on each one. By combining a good number of these inexpensive chips on a well-designed motherboard, you'd save the cost of 6 chassis, 6 motherboards, 6 sets of RAM, 6 disks, etc.. That's several thousand in savings right there.)
In 1998, some engineers at Corel took 10 StrongARMs and connected them on a custom backplane, made a couple of modifications to Apache, and were able to dish out close to one million web pages per minute.
I'd love to see someone put 8 of these on a board with a gig of memory, and two ethernet jacks. One would go to the network, the other would go to your file server/SAN/NAS/other_buzzword.
Put 2 gigs of memory on it for disk caching, and for a pretty low amount of money and electrical power, you could dish out VERY large numbers of web pages.
Shoot, take it farther: Have another unit based on them that runs LVS as a load-balancer, and put several of the servers behind it. All of the sudden, for $2000, you'd have the capability to dish out a billion web pages per day (or more), with load-balancing and realtime failover to boot!
a good SMP OS would handle multitasking properly among the CPUs. Your web browser would never interrupt your mp3 player again, and the UI would be unhindered by background processes.
Be careful what you say, or someone who's never used an SMP system will come along and tell you you're wrong. At least that's usually what happens when I tell people about my SMP desktops!
4 gigE ports, each on it's own PCI-X controller. Between the two Xeons and whatever amount of memory you through at it, one of these could *easily* handle a great deal of BGP sessions, load-balancing, failover, as well as VPN and encryption.
With a board like that, a couple of Xeons, and a gig of memory, these could out-perform some very, very expensive commercial routers.
Are you sure that USB drive isn't Linux compatible? All of the USB pen drives I've seen run as generic mass storage devices, and hence, are compatible with Linux. Perhaps there are a couple of oddballs out there, but I'm betting that it really will work with Linux.
It's interesting that Intel has, according to the article, been doing this since 2002, but just now dropped it on the market. Just like the "Extreme Edition" Pentium IV, this certainly seems like just another tactic to combat the fact that the Opteron, for most server uses, is beating the Xeon line very, very handily.
There are several new technologies that either are speeding up chips, or will speed up chips, and the best part is that they'll all work together.
For some time, SOI (silicon-on-insulator) has been helping chip manufacturers squeeze out extra performance. And the straining of the silicon lattice (strained silicon) helps as well. And you can combine them into SSOI, strained-silicon-on-insulator.
Well, there's also one other technology that's been developed, called "fully depleted silicon". And guess what - it should/will be possible to make fully-depleted, strained silicon-on-insulator chips. (FDSSOI?)
Between moving to 90 nm, then 65nm, and then further, as well as integrating high-K dialectrics and fully-depleted, strained silicon-on-insulator manufacturing technologies, we've still got a lot of headroom to keep cranking out faster and faster processors. Moore's law has still got a long time to live. And that's even if we don't make any new breakthroughs, but my guess is that the chip makers will continue to pull aces out of their sleeves, so to speak.
However, non-technical people are still buying Intel/AMD-based computers because they have the largest processor speed posted on the shelf
I don't know a single non-technical person that's made their paltform choice based on that - they do it on the fact that they've had some exposure to Windows and Office, and aren't really up to trying anything different.
Like Lothar of the Hill people said, "Because it is different, we naturally fear it!"
"TCO" is one of those things that gets thrown about with wild numbers, and can be used to prove any or both sides of nearly any argument, depending on how you look at the numbers.
Here's how the TCO has worked out for me over the past 10 years: I build white-box servers for our company, and they have just as good of performance and stability as any x86 from a "big-brand". We've saved a LOT of money right there.
Then, we've used Linux instead of purchasing very expensive OS and software licenses. We've saved even MORE money there.
As for drivers and OS support, I've got machines that are far, far too old to be of use for anything, and the drivers and OS are still there. I've even got some RAID cards that are (IMHO) too old to be of use for anything but a curiosity, and the drivers are still under active maintenance!
Overall, it's been a decade of "linux-on-x86" being just as cheap as it's touted to be.
They say their server ran $5k for two 1GHz chips, 4 GB memory, two 10K drives, four ethernet ports, and hardware SSL encryption.
I priced out the componants to build a dual Opteron 242 with the same trimmings (including redundant power supplies), sans the SSL encryption. It came to about $2500. In other words, half of the cost of the Sun system.
Now, the omission of the SSL encryption may make you cry foul. I don't believe it should, very few DBMS servers use significant amounts of SSL encryption. But, hey, let's throw one in anyway at $1,000. That still means the Opteron cost 30% less than the Sun.
Now, that's significant, but not an order of magnitude. There is one other catch, however: Sun's chips have a 128-bit PC2100 memory controller on each chip, where the Opteron has a 128-bit PC2700 memory controller. And a 25% improvement in memory bandwidth on a DBMS server is a very significant thing.
Now I don't have the two systems handy to benchmark, so I can't definitively say which would come out on top: But it's going to be very hard to turn down a system that has 25% more memory bandwidth, 60% higher clock, and costs half as much. There's always a chance the Sun will still come out a little ahead of the Opteron, but it's not going to come out THAT much ahead.
.... but a 64-bit address space and very fast memory controllers do make for good performance!
Every year, as our business has grown, I've had to upgrade our DBMS server to keep up. We've gone from a 4xP3 Xeon to a 2x AthlonMP to a 2xP4 Xeon, and next year it will be a 2x or 4x Opteron.
In every case, when the machine is finally hit it's max capacity, the CPU's were nearly *never* at full use. Even though the entire operation was running from memory and cache(the disk lights rarely blinked), the memory bandwidth has always been the limitation. Between the Opterons having VERY fast memory controllers (and each chip having it's own controller), and the ability to address vast amounts of memory, it's a recipe for letting those CPU's fulfill just a bit more of their true potential! : )
RDBMS systems are your killer app. Opterons are well-suited to RDBMS work, to the point of nearly seeming intended for it. Between the "big iron" memory architecture and the 64-bit address space, AMD has finally provided commodity hardware that can truly tackle real, heavy database environments.
The only reason I didn't buy an Opteron for our main RDBMS server this year was because they weren't ready for our peak season. This coming year, I'll be getting a minimum of a dual-opteron, more likely a quad - and getting it for a fraction of what similar performance would cost from Sun.
Do you like to be hands on with your computer/network hardware? I personally like to be able to take a machine down and rip into it on a workbench.
I actually prefer working on rack-mounted machines, provided they're on rails. It's terrific to simply pull on a machine, have it come sliding out, pop the top off, and have everything exposed. With better-designed rack-mounts, it's actually easier than working on a regular case!
Of course, as others have said, I would never go rack-mount in my home unless I had a very good number of servers in it for some reason. While it makes it nice, the cost premium does make it less attractive if you're not serious about it.
He wants to fly his plane out of their, instead of having it shipped. He's willing to pay good money for the fuel, right?
Well, it's easy. Let him use the phone to call and arrange for a fuel shipment. He can pay someone to bring him the fuel that he needs, and he's getting free room and board.
Sure, instead of the $400 or so the AVGAS would normally run him, it might cost him $15,000 to have it delivered. And it might take some time. So what?
He'll get to fly his plane out of there, and he'll have a little bit of time to sit around thinking "Gee, I really SHOULD have planned that better. Maybe next time I'll check if I should turn around BEFORE I get to the point-of-no-return."
It might not be the cost of the fuel, but rather getting more delivered. I don't think they can just send out for another hundred gallons when they're running low.
If the guy's go that much money, let him pay to have fuel delivered for his plane. So it takes a little time, so what? A week or two of sitting around might let him think about how to plan things better next time. And letting him pay the full amount of delivery would maybe wake him up to reality. And, best of all, he gets to fly his own plane out of there. Not a bad solution at all, in my mind.
Having an idiot who is so imbecilic as to not even account for possible headwinds drop in out of the blue and expect you to give him the world on a silver platter is entirely another.
Sometimes the best thing you can do for a person is let them work their own way out of the hole they dug for themselves.
This guy's being fed and sheltered. And he's been offered ways to get off of the island. If he's (a) wealthy enough to take on this sort of adventure, and (b) stupid enough to not plan for the nearly inevitable ("Oh, no, there couldn't be any headwinds!"), then you can't exactly blame me if I don't cry him a river for not having someone cater to him like the prince he evidently believes himself to be.
Sell him the gas at a markup, give him a bill for the food and shelter, and call it a day.
There are two problems:
First, to sell the gasoline at cost would still be an exhoribant amount of money, because you have to factor in the price of delivery to Antarctica, which ain't exactly cheap.
Second, even if they recovered the delivery cost, they're still short of fuel, and won't be able to just send out for it the next day. They need their fuel for their own uses, and can't just give it up for any Joe that happens to take off without enough of his own.
Slashdot Mirror
Professional wedding photographers shoot 300+ pics per event and rarely get better than 25% that come out with any sort of quality
Well, there are photographers, and there are "photographers".
I know plenty in the "I'll shoot a bunch, and end up with 10% to 25% that are really good" category. On the other hand, I know a very few who take a few shots, and end up with 75%-90% that are really worth it.
Making a blanket statement like that about photographers is just silly. Of course, since you're talking about *wedding* photographers, maybe it's not quite as silly. : )
steve
Many believe buying a better camera with greater megapixels etc will make them a better photographer.
I don't know a single person who thinks that.
On the other hand, since my camera only has about 1/3 of the resolution of my dye-sub printer, I know that I'll get higher-quality prints if I get a camera with more pixels. My photographic skills will, of course, be unchanged.
steve
Sure, the pixels in the CCD are large, giving it better light sensitivity. And sure, it's got a nice lens. Terrific.
On the other hand, consider that (a) it cost hundreds of thousands of dollars just for the fuel to get the launch craft off of the ground, ignoring all of the personnel costs, cost of the rover, etc., and that (b) this thing was only designed to travel about 600 yards.
Given the enormous expense and extremely tiny travelling range, I would think that throwing an extra thousand bucks on a higher-res CCD and a "longer" (telephoto) lens would be a no-brainer. Can't travel to the mountains? At least you'd be able to *see* them.
steve
Meeting the benchmarks is easy. Doing that cheaply is another matter!
The greatest point about a system like the one I described is that not only would it give extremely high performance for a relatively modest price, it could also be made to be very easy to use. That would probably sell a lot of them. Of course, that could potentially put a lot of admins out of business. : )
steve
I thought that one of Transmeta's selling points was the the chips were relatively inexpensive, compared to the AMD/Intel chips, but I could be wrong. The chips probably use much less power mostly because they likely have far fewer transistors than a P4, Athlon, or Opteron. It still may be more than $2,000, but it would still likely be MUCH cheaper (say, an order of magnitude) than trying to do it by building a bunch of Intel/AMD machines. (more below)
I mentioned a billion-page per day number simply because so many people's web servers seem to buckle under a slashdotting load of tens or hundreds of thousands of hits per *hour*.
I'm guessing that it would probably be able to hit the million-per-minute mark with fewer processers, as these sound to be significantly better performers than StrongARMs.
(more info on doing it with Intel/AMD machines: I use 7 dual-CPU front-end machines for handling the Perl CGI and dishing out HTTP/HTTPS for my office. They'll handle ten million hits per day with relative ease. (actually, they'd handle MUCH more if it were static HTML, a lot of the CGI work is pretty intensive) However, it's expensive. 7 good-quality rack-mount chassis' don't come cheap. 7 motherboards, 7 hard drives, having 7 sets of memory. And the memory is mostly for disk cache, so I'm duplicating the cache on each one. By combining a good number of these inexpensive chips on a well-designed motherboard, you'd save the cost of 6 chassis, 6 motherboards, 6 sets of RAM, 6 disks, etc.. That's several thousand in savings right there.)
steve
In 1998, some engineers at Corel took 10 StrongARMs and connected them on a custom backplane, made a couple of modifications to Apache, and were able to dish out close to one million web pages per minute.
I'd love to see someone put 8 of these on a board with a gig of memory, and two ethernet jacks. One would go to the network, the other would go to your file server/SAN/NAS/other_buzzword.
Put 2 gigs of memory on it for disk caching, and for a pretty low amount of money and electrical power, you could dish out VERY large numbers of web pages.
Shoot, take it farther: Have another unit based on them that runs LVS as a load-balancer, and put several of the servers behind it. All of the sudden, for $2000, you'd have the capability to dish out a billion web pages per day (or more), with load-balancing and realtime failover to boot!
steve
a good SMP OS would handle multitasking properly among the CPUs. Your web browser would never interrupt your mp3 player again, and the UI would be unhindered by background processes.
Be careful what you say, or someone who's never used an SMP system will come along and tell you you're wrong. At least that's usually what happens when I tell people about my SMP desktops!
steve
I'd like to see one based on this bad boy.
4 gigE ports, each on it's own PCI-X controller. Between the two Xeons and whatever amount of memory you through at it, one of these could *easily* handle a great deal of BGP sessions, load-balancing, failover, as well as VPN and encryption.
With a board like that, a couple of Xeons, and a gig of memory, these could out-perform some very, very expensive commercial routers.
steve
Are you sure that USB drive isn't Linux compatible? All of the USB pen drives I've seen run as generic mass storage devices, and hence, are compatible with Linux. Perhaps there are a couple of oddballs out there, but I'm betting that it really will work with Linux.
It's interesting that Intel has, according to the article, been doing this since 2002, but just now dropped it on the market. Just like the "Extreme Edition" Pentium IV, this certainly seems like just another tactic to combat the fact that the Opteron, for most server uses, is beating the Xeon line very, very handily.
steve
There are several new technologies that either are speeding up chips, or will speed up chips, and the best part is that they'll all work together.
For some time, SOI (silicon-on-insulator) has been helping chip manufacturers squeeze out extra performance. And the straining of the silicon lattice (strained silicon) helps as well. And you can combine them into SSOI, strained-silicon-on-insulator.
Well, there's also one other technology that's been developed, called "fully depleted silicon". And guess what - it should/will be possible to make fully-depleted, strained silicon-on-insulator chips. (FDSSOI?)
Between moving to 90 nm, then 65nm, and then further, as well as integrating high-K dialectrics and fully-depleted, strained silicon-on-insulator manufacturing technologies, we've still got a lot of headroom to keep cranking out faster and faster processors. Moore's law has still got a long time to live. And that's even if we don't make any new breakthroughs, but my guess is that the chip makers will continue to pull aces out of their sleeves, so to speak.
steve
However, non-technical people are still buying Intel/AMD-based computers because they have the largest processor speed posted on the shelf
I don't know a single non-technical person that's made their paltform choice based on that - they do it on the fact that they've had some exposure to Windows and Office, and aren't really up to trying anything different.
Like Lothar of the Hill people said, "Because it is different, we naturally fear it!"
steve
Speaking of video game sounds being reused...
I was amazed, while playing Revenant, to occasionally hear not-so-obvious background sounds that were directly from Dungeon Keeper. Pretty hilarious!
steve
I thought people bought Apple for the interface, not so they would have to fiddle with hardware for themselves!
; )
steve
"TCO" is one of those things that gets thrown about with wild numbers, and can be used to prove any or both sides of nearly any argument, depending on how you look at the numbers.
Here's how the TCO has worked out for me over the past 10 years: I build white-box servers for our company, and they have just as good of performance and stability as any x86 from a "big-brand". We've saved a LOT of money right there.
Then, we've used Linux instead of purchasing very expensive OS and software licenses. We've saved even MORE money there.
As for drivers and OS support, I've got machines that are far, far too old to be of use for anything, and the drivers and OS are still there. I've even got some RAID cards that are (IMHO) too old to be of use for anything but a curiosity, and the drivers are still under active maintenance!
Overall, it's been a decade of "linux-on-x86" being just as cheap as it's touted to be.
steve
If they really do get the price that low, I might even buy some of their overpriced CPU's for some of my servers, just to thank them!
; )
steve
They say their server ran $5k for two 1GHz chips, 4 GB memory, two 10K drives, four ethernet ports, and hardware SSL encryption.
I priced out the componants to build a dual Opteron 242 with the same trimmings (including redundant power supplies), sans the SSL encryption. It came to about $2500. In other words, half of the cost of the Sun system.
Now, the omission of the SSL encryption may make you cry foul. I don't believe it should, very few DBMS servers use significant amounts of SSL encryption. But, hey, let's throw one in anyway at $1,000. That still means the Opteron cost 30% less than the Sun.
Now, that's significant, but not an order of magnitude. There is one other catch, however: Sun's chips have a 128-bit PC2100 memory controller on each chip, where the Opteron has a 128-bit PC2700 memory controller. And a 25% improvement in memory bandwidth on a DBMS server is a very significant thing.
Now I don't have the two systems handy to benchmark, so I can't definitively say which would come out on top: But it's going to be very hard to turn down a system that has 25% more memory bandwidth, 60% higher clock, and costs half as much. There's always a chance the Sun will still come out a little ahead of the Opteron, but it's not going to come out THAT much ahead.
steve
.... but a 64-bit address space and very fast memory controllers do make for good performance!
Every year, as our business has grown, I've had to upgrade our DBMS server to keep up. We've gone from a 4xP3 Xeon to a 2x AthlonMP to a 2xP4 Xeon, and next year it will be a 2x or 4x Opteron.
In every case, when the machine is finally hit it's max capacity, the CPU's were nearly *never* at full use. Even though the entire operation was running from memory and cache(the disk lights rarely blinked), the memory bandwidth has always been the limitation. Between the Opterons having VERY fast memory controllers (and each chip having it's own controller), and the ability to address vast amounts of memory, it's a recipe for letting those CPU's fulfill just a bit more of their true potential! : )
steve
RDBMS systems are your killer app. Opterons are well-suited to RDBMS work, to the point of nearly seeming intended for it. Between the "big iron" memory architecture and the 64-bit address space, AMD has finally provided commodity hardware that can truly tackle real, heavy database environments.
The only reason I didn't buy an Opteron for our main RDBMS server this year was because they weren't ready for our peak season. This coming year, I'll be getting a minimum of a dual-opteron, more likely a quad - and getting it for a fraction of what similar performance would cost from Sun.
steve
Do you like to be hands on with your computer/network hardware? I personally like to be able to take a machine down and rip into it on a workbench.
I actually prefer working on rack-mounted machines, provided they're on rails. It's terrific to simply pull on a machine, have it come sliding out, pop the top off, and have everything exposed. With better-designed rack-mounts, it's actually easier than working on a regular case!
Of course, as others have said, I would never go rack-mount in my home unless I had a very good number of servers in it for some reason. While it makes it nice, the cost premium does make it less attractive if you're not serious about it.
steve
He wants to fly his plane out of their, instead of having it shipped. He's willing to pay good money for the fuel, right?
Well, it's easy. Let him use the phone to call and arrange for a fuel shipment. He can pay someone to bring him the fuel that he needs, and he's getting free room and board.
Sure, instead of the $400 or so the AVGAS would normally run him, it might cost him $15,000 to have it delivered. And it might take some time. So what?
He'll get to fly his plane out of there, and he'll have a little bit of time to sit around thinking "Gee, I really SHOULD have planned that better. Maybe next time I'll check if I should turn around BEFORE I get to the point-of-no-return."
steve
It might not be the cost of the fuel, but rather getting more delivered. I don't think they can just send out for another hundred gallons when they're running low.
If the guy's go that much money, let him pay to have fuel delivered for his plane. So it takes a little time, so what? A week or two of sitting around might let him think about how to plan things better next time. And letting him pay the full amount of delivery would maybe wake him up to reality. And, best of all, he gets to fly his own plane out of there. Not a bad solution at all, in my mind.
steve
Helping someone is one thing.
Having an idiot who is so imbecilic as to not even account for possible headwinds drop in out of the blue and expect you to give him the world on a silver platter is entirely another.
Sometimes the best thing you can do for a person is let them work their own way out of the hole they dug for themselves.
This guy's being fed and sheltered. And he's been offered ways to get off of the island. If he's (a) wealthy enough to take on this sort of adventure, and (b) stupid enough to not plan for the nearly inevitable ("Oh, no, there couldn't be any headwinds!"), then you can't exactly blame me if I don't cry him a river for not having someone cater to him like the prince he evidently believes himself to be.
steve
Sell him the gas at a markup, give him a bill for the food and shelter, and call it a day.
There are two problems:
First, to sell the gasoline at cost would still be an exhoribant amount of money, because you have to factor in the price of delivery to Antarctica, which ain't exactly cheap.
Second, even if they recovered the delivery cost, they're still short of fuel, and won't be able to just send out for it the next day. They need their fuel for their own uses, and can't just give it up for any Joe that happens to take off without enough of his own.
steve
4 months?
The dates between 2.4.1 and 2.4.17 are 1/29/2001 and 12/21/2201, respectively.
Mebbe it's just me, but that seems like a bit more than four months.
steve