There are a fair amount of people in the world who don't get motion sick. I can read or use a computer in the front or back seats while on the road just fine. Busses too. I've been out to sea in boats smaller than the cruise liner they're talking about and I wasn't ever bothered (ferry in the english channel, and small cruise ship in the north sea).
Worth a note: we don't have a 640-node cluster, it's really a 512 and 128 node clusters. Theoretical peak was I think 8 Tflop, and we got 4.6 or about 57%. VT's first entry only hit 54% if I remember what our parallel software expert told me right, so with more time to tune we'll probably get higher again for SC05's list. (VT's up to what, 61% or so now?)
Computer Science already has HAL as a cluster name. Having the new one be HAL would only confuse matters. Especially since HAL contains the old turing machines!
Really, that's all we're talking about. More and more and more lag. Truly awful lag. Server killing connection swamping videocard toasting lag.
At least, that's what happend in Asheron's Call when there were dev-run events. It's what happens in CoH. I can only assume it happens the same in other games.
The little java free-MMO called Runescape does a version of it. Your character exists only once on the central DB servers. I suspect those DB servers are running some sort of replication between the US and Euro server clusters, and then the front end worlds all pull off that same, single, DB.
You only see people on your server, (so if you're on NY1 you don't see anyone but those on NY1) but if you logout and log back in to a different server (NY2) your character is just as it was from NY1.
The problem with what you're suggesting is that some games (like horizons, guild wars, etc) have a fully-seamless world -- no zoneing, ever (well unless you teleport but that's something else now isn't it?). If you have no zoneing, then you've got a harder time breaking players up into distinct chunks -- players who are on the border of two server's zones need updates from both zones, etc.
On the other hand if you have strict zones (like CoH), you only have so many zones. That makes it hard to load balance things, so what CoH does is starts pushing people to an overflow zone (aka "Atlas Park 2") when they try to zone in or log in to a too full zone.
But too many of "Atlas Park n" and it gets hard to meet up with your friends to do missions and such.
I guess my point is: it is solved, but the solutions bring up problems of their own.
He could have done a CSS version of the exact mac effect if he'd spent long enough dwaddling with the images/magnification. http://alistapart.com/articles/sprites/ tells ya how.
Most of the MMORPGs out there have players doing the same thing. Heck, players have written tools to automate the process for Asheron's Call -- I assume the same is true of other games.
Don't forget to include other applications that (could) drive this beyond games. (which makes me wonder why they aren't...)
What are the DoE's ASCI centers doing? This includes things such as Rigid Body Dynamics, Collision Detection, Fluid Simulation, Soft Bodies and Fracturing of objects. is a pretty damn close description of what they do. They do it massivly parallel of course, in order to get decent performance out of it. But there's no reason they shouldn't be able to have these little add-on cards doing the number crunching with the CPU mediating networking and the OS overhead.
If these little guys work as promised and catch on, I wouldn't be surprized if we start seeing a lot of CPU-soldered-on-to-motherboard solutions coming out, with the PPU and GPU the "high cost" items you actually slot into your computer.
It will raise the complexity however, because you'd really like your PPU to do some direct geometry manipulation, not have the results fed to the GPU for display via the CPU.
Windows is pretty, with windows it just works, you don't have to support multiple windows configurations.
What, am I the only one who remembers the days of DOS/Win3.1/Win95/WinNT overlap? emm386.dll? I'll admit, I was only getting started in sysadmin as a summer job back then, but I've seen a shop try to deal with supporting all of those at the same time except NT. Speaking as a user (gamer), I can tell you things sure don't work that smooth between versions of windows either.
And yes you could argue that the above is equiv to RH 7, 8, 9... but that's not accurate. RH is targetted to the same people whereas Win95 and WinNT certainly weren't.
He's saying that linux has lost (or will lose) because it hasn't done anything about the MS monopoly (due to it's fragmented nature). Duh! It's a monopoly. High costs to entry, hard to dislodge. MS had their monopoly with Windows 3 at the latest (realistically, probably earlier with DOS). We didn't start really seeing linux distros until what, 94-95? That's post 3.1, that's starting to push Win95.
That linux has not only survived, but flourished against a monopoly should actually tell you a lot more than his scare-blog-post. Microsoft has successfully crushed 2 other competetor OSes (OS/2 and Netware) and has an assist on a 3rd (BeOS - the Microsoft OS lockin on hardware vendors).
Now, I'm not saying that the LSB isn't a great thing. I'm just saying that I don't think multiple distros (or desktop enviroments) is bad.
Beyond the GPU, any intensive computation application gets benefits from the second CPU.
Our local (to UIUC) parallel software master, working on the turing xserve cluster is pulling about 95% (I think, don't quote me) of theoretical peak performance in linpack running on 1 cpu on 1 xserve. Bring that up to both cpus in one and he said it dropped to around 50%.
Why? The OS has to run somewhere. When it's running, that processor is stuck with it. The other processor is stuck waiting for the OS, and then things can pick up again.
Now, we haven't yet finished tuning the systems to make the OS do as little as possible. (they're still running GUIs, so we can remote desktop into them amoung other things.) But still that's quite a performance hit!
He said two machines running 1 CPU each over myrinet were still in the 90%ish of theoretical peak.
So can we quite rehashing this stupid topic every time dual core CPUs comes up? Yes it'll help. No it won't double your game performance (unless it's written for a dual-core cpu), and probably it won't even double it then, because there's still teamspeak/windows/aim/virus scan/etc running that need cpu time.
Don't offer much benefit... for games that aren't written to use them.
However lots of games could be very easily written to use them. Not to mention they offer immediate (small) performance benefits to gamers of being able to offload everything but the game onto the second processor. Windows, the mp3 player, the virus scanner, web browser showing a stuipd flash ad, the (speedhack... er I mean) "stats" addon you have running for your favorite MMO/FPS...
"But writing parallel software is hard!" is the usual refrain I hear. Oh really now? Every bot in a match can be it's own process, talking to the game server just like a player. That's trivially parallelizable.
If you've got a problem that's trivially parallelizable, then sure grid computing is great! RC5, seti@home, and similar projects can benefit from grid computing (really, that's what grid computing is -- someone else's code able to run on your machine when it's idle and do work).
However, don't even begin to think you'll be solving anything that requires any sort of processor to processor communication. Rocket simulation (our local favorite example here at UIUC) for instance is heavily communication based.
The linpack benchmark that top500 uses also needs a low-latency interconnect to perform really well, so don't expect to see "the grid" sitting up at the #1 supercomputer slot on top500.org anytime soon (or really, ever, unless someone develops FTL networking). Latency on the internet in general (and specifically around the world thru all those switches and latest_slashdot_hot_chick_movie.torrent packets) is nothing near what a supercomputer needs.
Now, there are research groups looking at ways of making communiation delays less of a problem, including the one I was in while I was in grad school. There's a number of ways to do it, but none of them I've seen are going to take on worldwide-network-latency and survive with their performance intact.
Even something as "simple" as chess wants to have a fast interconnect - every node that's gotten stranded working on low-priority (bad move) work is a wasted node you may as well not have.
Do the math. (anyone else remember those old Jaguar ads?) A basic Xserve runs right at 3 grand. Upgrade that some (memory to the dual proc 2.3 ghz model), throw in overhead cost for a fast network, some storage, head nodes, etc... and how much are you actually talking in terms of (amortized cost) for an xserve in a cluster? Let's say 5k (an overestimate I'd say, but I didn't go try to price out a fast interconnect unlike hitting apple's page for an Xserve price real quick).
That's for a dual processor box too, so cost per processor is more like 2.5k. So, ask now how long you have to use that machine before you've made the cost of using sun's service?
Let's simplify: 4.8k for the xserve, 2.4k per processor. That's only 100 days running flat out. A third of a year (less!). Yeah, you aren't likely to have the machine loaded 100%, 24/7 yourself, but I'm sure when the Xserve cluster here at UIUC is officially open it will be pretty much running 100% 24/7, because the old PC cluster it's replacing was always running nearly flat-out. The apple replacments are currently running about 40% load and they aren't even open to "the public" as it were. (with 10% additional being used to tune the cluster by our resident parallel expert)
Yes, there's system administration overhead, and cooling and power and all sorts of other things you need to take care of too. And, for places that aren't a university that don't have the resources to deal with that, Sun's grid may be a good deal. Or places that won't load the machine up 100% 24/7.
But for a large compute cluster that you're going to keep pretty busy, you can do better.
He's offering them virtual gold for real gold and they're taking him up on the offer. There's no possible way to call this scamming as long as he's delivering the product.
You may not like it. You may consider it exploiting or breaking the ToS. Go for it, I'm a bit peeved at people in AC that'd been doing that (a lot of duping going on there too the devs and MS didn't have the balls to ban), but mainly because when I eventually quit, I couldn't ebay my house/account.
So what you're saying is that your internet pipe is faster than your hard drive? Tell you what I'll go buy you a nice 15k scsi disk, you buy me a new internet connection, k?
On the other hand, if you assume Cringely is right and that it's for downloads AND that those downloads will be faster than going to the rental store, you're left with only a couple obvious things:
1) Streaming. You'd be buffering for the 15 minutes of going-to-store then play and rely on the buffer.
2) You'll need a big pipe. The rate of 3-10Mbps for a dvd video (dunno about HD video) isn't likely to go down too much more with other compression. On the other hand, with cable companies talking about upping their service to 4Mbps or 6Mbps, and baby bells trialing FTTP that's approaching feasibility.
I'd mark it as more a 2k6 thing than 2k5 but who knows maybe it'll take off as a driving force behind FTTP. I'm sure I could figure something to use my bandwidth there for since I'm not a huge movie watcher.
I'd like to note that this case would already be over in CoH's favor if they didn't have the "we own everything you create on CoH" clause in their EULA. If they could tell Marvell, "Sorry buds we don't own any of that, we have the rights necessary to maintain the servers but those characters are the player's creation and owned by them." they'd probably never have gotten sued. (aka, we're just a carrier, just like all the baby bells & cable companies)
So despite being a player of CoH, I have remakably limited simpathy for them and think that that excerpt from their defense is pretty shaky at best. If, like a pencil, they left me in ownership of my character, then I'd agree it's a good defense.
For those of us who work with supercomputers, it still means that. And "how much more power do we need" and "how much cooling do we need" and "there are how-many-thousnad-ethernet-cables?"
It's okay, capes in CoH don't actually exist as "physical" objects (in the virtual world). Your hair, armor, and cape happily intersect and move through each other.
Given that, we can infer that capes don't have any real physics in the game. On your machine, they may, but they don't on the servers, which means they can't check for cape jet engine collisions.
Besides, in CoH aircraft would be point-sources just like cars. You'd just get shoved to the side.
The real problem is how do you power cars/trucks/and friends? A huge portion of our energy use (and in particular oil use -- because a lot of our electricity is nuclear and coal based) is there -- something like 20-30% if I recall correctly from one of my classes.
You aren't going to be sticking a reactor on a car or truck. Works okay for the navy's ships but they're pretty large. You might be able to get a reactor stuck into a train because they also are pretty big. You can of course also run electric power along the rails, so if you have fixed-site generation you could just power the trains that way, though it means a lot of infrastructure that we don't have would have to be added to the tracks.
The touted hydrogen economy doesn't seem like it's there yet, even if we could snap our fingers and have nuclear power generation across the country for free tomorrow. Possibly that'd be okay for home heating and such, but I don't think it'll be viable for transport. Electric cars could solve it for a lot of people theoretically, but it's doubtful they'd really go for it looking out at the sea of Urban Enviromental Assault Vehicles in parking lots.
Slashdot Mirror
There are a fair amount of people in the world who don't get motion sick. I can read or use a computer in the front or back seats while on the road just fine. Busses too. I've been out to sea in boats smaller than the cruise liner they're talking about and I wasn't ever bothered (ferry in the english channel, and small cruise ship in the north sea).
So it's a critera for the job. So what?
Worth a note: we don't have a 640-node cluster, it's really a 512 and 128 node clusters. Theoretical peak was I think 8 Tflop, and we got 4.6 or about 57%. VT's first entry only hit 54% if I remember what our parallel software expert told me right, so with more time to tune we'll probably get higher again for SC05's list. (VT's up to what, 61% or so now?)
Netfiles will un-quota then every... hour I think Ben said, so try back in a bit. The PDF's kinda big.
Computer Science already has HAL as a cluster name. Having the new one be HAL would only confuse matters. Especially since HAL contains the old turing machines!
Probably in-game advertising.
There's already free graphical MMOs out there, supported by ads and "premium" paid subscribers (ala runescape).
Really, that's all we're talking about. More and more and more lag. Truly awful lag. Server killing connection swamping videocard toasting lag.
At least, that's what happend in Asheron's Call when there were dev-run events. It's what happens in CoH. I can only assume it happens the same in other games.
The little java free-MMO called Runescape does a version of it. Your character exists only once on the central DB servers. I suspect those DB servers are running some sort of replication between the US and Euro server clusters, and then the front end worlds all pull off that same, single, DB.
You only see people on your server, (so if you're on NY1 you don't see anyone but those on NY1) but if you logout and log back in to a different server (NY2) your character is just as it was from NY1.
The problem with what you're suggesting is that some games (like horizons, guild wars, etc) have a fully-seamless world -- no zoneing, ever (well unless you teleport but that's something else now isn't it?). If you have no zoneing, then you've got a harder time breaking players up into distinct chunks -- players who are on the border of two server's zones need updates from both zones, etc.
On the other hand if you have strict zones (like CoH), you only have so many zones. That makes it hard to load balance things, so what CoH does is starts pushing people to an overflow zone (aka "Atlas Park 2") when they try to zone in or log in to a too full zone.
But too many of "Atlas Park n" and it gets hard to meet up with your friends to do missions and such.
I guess my point is: it is solved, but the solutions bring up problems of their own.
He could have done a CSS version of the exact mac effect if he'd spent long enough dwaddling with the images/magnification. http://alistapart.com/articles/sprites/ tells ya how.
Most of the MMORPGs out there have players doing the same thing. Heck, players have written tools to automate the process for Asheron's Call -- I assume the same is true of other games.
Don't forget to include other applications that (could) drive this beyond games. (which makes me wonder why they aren't...)
What are the DoE's ASCI centers doing? This includes things such as Rigid Body Dynamics, Collision Detection, Fluid Simulation, Soft Bodies and Fracturing of objects. is a pretty damn close description of what they do. They do it massivly parallel of course, in order to get decent performance out of it. But there's no reason they shouldn't be able to have these little add-on cards doing the number crunching with the CPU mediating networking and the OS overhead.
If these little guys work as promised and catch on, I wouldn't be surprized if we start seeing a lot of CPU-soldered-on-to-motherboard solutions coming out, with the PPU and GPU the "high cost" items you actually slot into your computer.
It will raise the complexity however, because you'd really like your PPU to do some direct geometry manipulation, not have the results fed to the GPU for display via the CPU.
Windows is pretty, with windows it just works, you don't have to support multiple windows configurations.
What, am I the only one who remembers the days of DOS/Win3.1/Win95/WinNT overlap? emm386.dll? I'll admit, I was only getting started in sysadmin as a summer job back then, but I've seen a shop try to deal with supporting all of those at the same time except NT. Speaking as a user (gamer), I can tell you things sure don't work that smooth between versions of windows either.
And yes you could argue that the above is equiv to RH 7, 8, 9... but that's not accurate. RH is targetted to the same people whereas Win95 and WinNT certainly weren't.
He's saying that linux has lost (or will lose) because it hasn't done anything about the MS monopoly (due to it's fragmented nature). Duh! It's a monopoly. High costs to entry, hard to dislodge. MS had their monopoly with Windows 3 at the latest (realistically, probably earlier with DOS). We didn't start really seeing linux distros until what, 94-95? That's post 3.1, that's starting to push Win95.
That linux has not only survived, but flourished against a monopoly should actually tell you a lot more than his scare-blog-post. Microsoft has successfully crushed 2 other competetor OSes (OS/2 and Netware) and has an assist on a 3rd (BeOS - the Microsoft OS lockin on hardware vendors).
Now, I'm not saying that the LSB isn't a great thing. I'm just saying that I don't think multiple distros (or desktop enviroments) is bad.
Beyond the GPU, any intensive computation application gets benefits from the second CPU.
Our local (to UIUC) parallel software master, working on the turing xserve cluster is pulling about 95% (I think, don't quote me) of theoretical peak performance in linpack running on 1 cpu on 1 xserve. Bring that up to both cpus in one and he said it dropped to around 50%.
Why? The OS has to run somewhere. When it's running, that processor is stuck with it. The other processor is stuck waiting for the OS, and then things can pick up again.
Now, we haven't yet finished tuning the systems to make the OS do as little as possible. (they're still running GUIs, so we can remote desktop into them amoung other things.) But still that's quite a performance hit!
He said two machines running 1 CPU each over myrinet were still in the 90%ish of theoretical peak.
So can we quite rehashing this stupid topic every time dual core CPUs comes up? Yes it'll help. No it won't double your game performance (unless it's written for a dual-core cpu), and probably it won't even double it then, because there's still teamspeak/windows/aim/virus scan/etc running that need cpu time.
Don't offer much benefit... for games that aren't written to use them.
However lots of games could be very easily written to use them. Not to mention they offer immediate (small) performance benefits to gamers of being able to offload everything but the game onto the second processor. Windows, the mp3 player, the virus scanner, web browser showing a stuipd flash ad, the (speedhack... er I mean) "stats" addon you have running for your favorite MMO/FPS...
"But writing parallel software is hard!" is the usual refrain I hear. Oh really now? Every bot in a match can be it's own process, talking to the game server just like a player. That's trivially parallelizable.
If you've got a problem that's trivially parallelizable, then sure grid computing is great! RC5, seti@home, and similar projects can benefit from grid computing (really, that's what grid computing is -- someone else's code able to run on your machine when it's idle and do work).
However, don't even begin to think you'll be solving anything that requires any sort of processor to processor communication. Rocket simulation (our local favorite example here at UIUC) for instance is heavily communication based.
The linpack benchmark that top500 uses also needs a low-latency interconnect to perform really well, so don't expect to see "the grid" sitting up at the #1 supercomputer slot on top500.org anytime soon (or really, ever, unless someone develops FTL networking). Latency on the internet in general (and specifically around the world thru all those switches and latest_slashdot_hot_chick_movie.torrent packets) is nothing near what a supercomputer needs.
Now, there are research groups looking at ways of making communiation delays less of a problem, including the one I was in while I was in grad school. There's a number of ways to do it, but none of them I've seen are going to take on worldwide-network-latency and survive with their performance intact.
Even something as "simple" as chess wants to have a fast interconnect - every node that's gotten stranded working on low-priority (bad move) work is a wasted node you may as well not have.
Thank god your average linux account can't go modify the kernel, unlike your average windows account! Maybe now they'll have to finish catching up.
Do the math. (anyone else remember those old Jaguar ads?) A basic Xserve runs right at 3 grand. Upgrade that some (memory to the dual proc 2.3 ghz model), throw in overhead cost for a fast network, some storage, head nodes, etc... and how much are you actually talking in terms of (amortized cost) for an xserve in a cluster? Let's say 5k (an overestimate I'd say, but I didn't go try to price out a fast interconnect unlike hitting apple's page for an Xserve price real quick).
That's for a dual processor box too, so cost per processor is more like 2.5k. So, ask now how long you have to use that machine before you've made the cost of using sun's service?
Let's simplify: 4.8k for the xserve, 2.4k per processor. That's only 100 days running flat out. A third of a year (less!). Yeah, you aren't likely to have the machine loaded 100%, 24/7 yourself, but I'm sure when the Xserve cluster here at UIUC is officially open it will be pretty much running 100% 24/7, because the old PC cluster it's replacing was always running nearly flat-out. The apple replacments are currently running about 40% load and they aren't even open to "the public" as it were. (with 10% additional being used to tune the cluster by our resident parallel expert)
Yes, there's system administration overhead, and cooling and power and all sorts of other things you need to take care of too. And, for places that aren't a university that don't have the resources to deal with that, Sun's grid may be a good deal. Or places that won't load the machine up 100% 24/7.
But for a large compute cluster that you're going to keep pretty busy, you can do better.
He's offering them virtual gold for real gold and they're taking him up on the offer. There's no possible way to call this scamming as long as he's delivering the product.
You may not like it. You may consider it exploiting or breaking the ToS. Go for it, I'm a bit peeved at people in AC that'd been doing that (a lot of duping going on there too the devs and MS didn't have the balls to ban), but mainly because when I eventually quit, I couldn't ebay my house/account.
So what you're saying is that your internet pipe is faster than your hard drive? Tell you what I'll go buy you a nice 15k scsi disk, you buy me a new internet connection, k?
On the other hand, if you assume Cringely is right and that it's for downloads AND that those downloads will be faster than going to the rental store, you're left with only a couple obvious things:
1) Streaming. You'd be buffering for the 15 minutes of going-to-store then play and rely on the buffer.
2) You'll need a big pipe. The rate of 3-10Mbps for a dvd video (dunno about HD video) isn't likely to go down too much more with other compression. On the other hand, with cable companies talking about upping their service to 4Mbps or 6Mbps, and baby bells trialing FTTP that's approaching feasibility.
I'd mark it as more a 2k6 thing than 2k5 but who knows maybe it'll take off as a driving force behind FTTP. I'm sure I could figure something to use my bandwidth there for since I'm not a huge movie watcher.
Or at least change it every so often so when you piss someone off they can't nuke it.
I'd like to note that this case would already be over in CoH's favor if they didn't have the "we own everything you create on CoH" clause in their EULA. If they could tell Marvell, "Sorry buds we don't own any of that, we have the rights necessary to maintain the servers but those characters are the player's creation and owned by them." they'd probably never have gotten sued. (aka, we're just a carrier, just like all the baby bells & cable companies)
So despite being a player of CoH, I have remakably limited simpathy for them and think that that excerpt from their defense is pretty shaky at best. If, like a pencil, they left me in ownership of my character, then I'd agree it's a good defense.
Every third month has 31 days, the rest have 30. That is in fact way easier to remember than the stupid little rhyme I never actually learned.
For those of us who work with supercomputers, it still means that. And "how much more power do we need" and "how much cooling do we need" and "there are how-many-thousnad-ethernet-cables?"
It's okay, capes in CoH don't actually exist as "physical" objects (in the virtual world). Your hair, armor, and cape happily intersect and move through each other.
Given that, we can infer that capes don't have any real physics in the game. On your machine, they may, but they don't on the servers, which means they can't check for cape jet engine collisions.
Besides, in CoH aircraft would be point-sources just like cars. You'd just get shoved to the side.
I'm not sure on how efficient the biodiesel conversion is, but even with the full US corn crop, we couldn't manage all of our oil use as ethanol.
The real problem is how do you power cars/trucks/and friends? A huge portion of our energy use (and in particular oil use -- because a lot of our electricity is nuclear and coal based) is there -- something like 20-30% if I recall correctly from one of my classes.
You aren't going to be sticking a reactor on a car or truck. Works okay for the navy's ships but they're pretty large. You might be able to get a reactor stuck into a train because they also are pretty big. You can of course also run electric power along the rails, so if you have fixed-site generation you could just power the trains that way, though it means a lot of infrastructure that we don't have would have to be added to the tracks.
The touted hydrogen economy doesn't seem like it's there yet, even if we could snap our fingers and have nuclear power generation across the country for free tomorrow. Possibly that'd be okay for home heating and such, but I don't think it'll be viable for transport. Electric cars could solve it for a lot of people theoretically, but it's doubtful they'd really go for it looking out at the sea of Urban Enviromental Assault Vehicles in parking lots.