the price improvement per FLOPS of the GPUs over the CPU is even better now
Yes, but the price improvement per bandwidth and especially latency of the interconnect
between the the two is much worse. Going off-chip for anything has a huge cost; in order for it to make sense, you have to be able to amortize that cost.
And those DSP chips are CPUs in the conventional sense, although they don't have all the niceties that modern CPUs have (which, ironically, also often used to be implemented as co-processors: complicated interrupt/exception handling, virtual address translation). What I'm thinking of are the TMS320* ISA, which is much more like a CPU than is a GPU. Maybe you're thinking of something else.
Well, no, if you want flops/$ then the signal processing chips used in cell phones and MP3 players are the clear winners. There are some real screamers here. But they're a bit complicated to program and don't function well as general purpose processors, which is why they're primarily used in systems where they can be programmed once and then shipped by the million.
As I wrote before, I'm sure there's some workload where it makes sense to mate a T1 and a GPU (besides the obvious one, i.e., rendering graphics). But the relative latency and bandwidth gulf between the CPU and GPU make it impractical in many cases. Want to multiply two numbers? Do it on the chip; it might take more CPU time but it doesn't take any setup, transfer, and teardown time. Want to multiply a million numbers? In that case the setup and init time is amortized and the memory transfers can be pipelined, so it could make sense. Where's the break-even point? I don't know, but I'll wager it's closer to a million than to one.
I remember when it was common practice to buy extra hardware to add to your system to implement fast floating point ops. First it was a box (FPS), then a few cards (Sky), then a card (Mercury), then a daughterboard (everyone), then a chip (Weitek)... and then it was on the CPU and everyone expected it to be there.
But Sun realized that the more things change, the more they stay the same; the reason why vendors got away with making floating point an expensive option was that there are lots of workloads where floating point performance is unimportant. So they applied the RISC principle and chose to not waste a lot of silicon on the T1 implementing instructions that are not needed in their target workload, but instead figure out how to get lots of concurrent threads.
Trying to improve floating point perf on a T1 by adding another card is like trying to figure out how to put wheels on a fish. It might be a cool hack and it might solve some particular problem but it doesn't generalize.
If you want floating point perf and tons of threads, wait for the rock chip from Sun (and hope that Sun stays afloat long enough to ship it). It's like a T1 only moreso, with floating point for each thread.
Wait: Oracle is porting their database to Linux and the headline is "Linux Helping Oracle?" Sounds more like Oracle is helping Linux get into the datacenter, or at the very least, there's a symbiotic relationship. Linux is useful to Oracle (this isn't news, they've been pushing to an all-Linux solution for years) but Oracle is also very useful to Linux. Hate Oracle if you must, but admit that they've put a lot of money into Linux.
Good ideas. Go for it! (or at least, send them the ideas, instead of posting them to slashdot, where they'll probably never be seen by the right folks at NASA...)
I vaguely remember seeing some orbital sim applets on some university web site. Maybe there is some reason why they can't be snarfed by NASA. Or maybe the NASA folks don't think kids can hack this stuff. Maybe I'll start a new web site to gather up this stuff. Who's with me?
Sun has clockless chips up and running (real silicon, not sims) and they have done some interesting things, but they don't have a complete system that's ready to ship. And there are other components out there that use the clockless philosophy to do certain things, but they're not CPUs in any sense. To give credit where credit is due, as the parent post points out, ARM beat Sun out the door with a clockless CPU that is a drop-in replacement (to some degree, anyway -- not clear how much) for an existing, established architecture. But that wasn't/isn't Suns goal (although perhaps it should be...). They're pushing in new directions, not using this to reimplement current architectures.
Aren't these April Fools pranks supposed to be plausible (at least, at first)? Kinda gives it away in the first sentence with the assertion that there's someone at SlashDot who acts based on the results of research...
It would be so nice if it took creativity and hard work and perhaps a little luck to get rich, but apparently a stale idea and a little PHP/MySQL knowledge is enough these days.
How exactly are the shareholders going to be pleased?
Axing senior management isn't going to get Vista out the door any faster -- probably a lot slower because whoever comes it to pick up the pieces is going to have a hell of a job. It might make Windows 2021 (or whatever they're calling Vista 1.1) ship quicker but in the short run, it'll be chaos. Shareholders, for the most part, don't care about the long run -- they care about now.
Either that or to explain why the Grid is having problems handling the bandwidth, etc
Uh, yeah?
Sun claims that their grid is having problems handling the bandwidth of a DDoS. What part of this is hard to believe?
Announce a new service, expect a DDoS. That's the world we live in. The only thing Sun does that's different is admit it instead of dismissing it with some dopey error message (Bad Server! No Cookie for You!)
A badly designed prison is better than a perfect one.
No. A badly designed prison is a waste of money that makes
politicians desperate and willing to spend any amount of money
to build the next version, and a populace eager to pay for it.
I'd rather have some decent computer scientists come in and
design something so powerful and dangerous that people will
realize the implications and squash the project. If we get something
that's 5% better than the last system, then our liberties will get
whittled away so slowly that most people won't ever even notice.
My thought after reading this article was "Either this guy has no kids (and maybe no S.O.)
or else he's in for a rude awakening one day soon!"
I used to do the "code-til-you-drop, then sleep until you can do it again" thing
and I was incredibly productive. Now I have kids... and I'm still productive, but
my life has a lot more structure. Interrupts are not necessarily a bad thing.
If you're working on something important/interesting/compelling, then it's
still going to be important/interesting/whatever after you change
your two-year-old's poopy diaper. And if my code is so disorganized that I
can't remember what I was doing ten minutes later, well, it probably wasn't
going to work anyway!
I thought polo and yachting were the sports for people with too much money and time on their hands.
Well, I guess times change.
Anyone want to partner with me to develop league of swimming hamsters or weight-lifting mice?
Those are sports we can all enjoy (watching, anyway). They could be the WWF of the 21st century.
Wake me up when there's some "stuff that matters".
Or when the Google fanboys go off and start up
a "GoogleDot" web site (which will be the best thing
ever and spend five years in beta...) and
we can have more stories about swimming hamsters.
Just give me until Monday so I can short a few thousand shares of Google...
If Google blocks IE, people are going to take the path of least resistance and
use a different search engine. It's not as if there aren't other perfectly adequate
search engines. Google doesn't have the sort of monopoly necessary to pull this
off.
No... It must be a sign that slashdot readers are mesmerized by
Google when a doodle contest held on Google premises is
a front-page story. Hey, we might get to see what their cubicles
look like!
Slashdots apparent obsession with Google makes People Magazines
"bennifer" story look like an idle fancy. Let's move on.
Slashdot Mirror
Yes, just like almost every other company traded on the NYSE. Companies don't post their results during trading hours.
Nothing to see here, move along.
2. Leave fifteen minutes earlier or ten minutes later.
3. Wonder where everyone else is hiding.
Or just wait until 9:00, when most other people are already at work.
Yes, but the price improvement per bandwidth and especially latency of the interconnect between the the two is much worse. Going off-chip for anything has a huge cost; in order for it to make sense, you have to be able to amortize that cost.
And those DSP chips are CPUs in the conventional sense, although they don't have all the niceties that modern CPUs have (which, ironically, also often used to be implemented as co-processors: complicated interrupt/exception handling, virtual address translation). What I'm thinking of are the TMS320* ISA, which is much more like a CPU than is a GPU. Maybe you're thinking of something else.
As I wrote before, I'm sure there's some workload where it makes sense to mate a T1 and a GPU (besides the obvious one, i.e., rendering graphics). But the relative latency and bandwidth gulf between the CPU and GPU make it impractical in many cases. Want to multiply two numbers? Do it on the chip; it might take more CPU time but it doesn't take any setup, transfer, and teardown time. Want to multiply a million numbers? In that case the setup and init time is amortized and the memory transfers can be pipelined, so it could make sense. Where's the break-even point? I don't know, but I'll wager it's closer to a million than to one.
But Sun realized that the more things change, the more they stay the same; the reason why vendors got away with making floating point an expensive option was that there are lots of workloads where floating point performance is unimportant. So they applied the RISC principle and chose to not waste a lot of silicon on the T1 implementing instructions that are not needed in their target workload, but instead figure out how to get lots of concurrent threads.
Trying to improve floating point perf on a T1 by adding another card is like trying to figure out how to put wheels on a fish. It might be a cool hack and it might solve some particular problem but it doesn't generalize.
If you want floating point perf and tons of threads, wait for the rock chip from Sun (and hope that Sun stays afloat long enough to ship it). It's like a T1 only moreso, with floating point for each thread.
Damn you!
I don't believe this for a second. There's nothing "slowly" about it.
Wait: Oracle is porting their database to Linux and the headline is "Linux Helping Oracle?" Sounds more like Oracle is helping Linux get into the datacenter, or at the very least, there's a symbiotic relationship. Linux is useful to Oracle (this isn't news, they've been pushing to an all-Linux solution for years) but Oracle is also very useful to Linux. Hate Oracle if you must, but admit that they've put a lot of money into Linux.
I vaguely remember seeing some orbital sim applets on some university web site. Maybe there is some reason why they can't be snarfed by NASA. Or maybe the NASA folks don't think kids can hack this stuff. Maybe I'll start a new web site to gather up this stuff. Who's with me?
Sun has clockless chips up and running (real silicon, not sims) and they have done some interesting things, but they don't have a complete system that's ready to ship. And there are other components out there that use the clockless philosophy to do certain things, but they're not CPUs in any sense. To give credit where credit is due, as the parent post points out, ARM beat Sun out the door with a clockless CPU that is a drop-in replacement (to some degree, anyway -- not clear how much) for an existing, established architecture. But that wasn't/isn't Suns goal (although perhaps it should be...). They're pushing in new directions, not using this to reimplement current architectures.
Aren't these April Fools pranks supposed to be plausible (at least, at first)? Kinda gives it away in the first sentence with the assertion that there's someone at SlashDot who acts based on the results of research...
Irony is an anonymous coward telling me to kill myself for publically stating an opinion.
The mixture of these mechanisms by amateurs is not recommended.
Gadzooks.
2) Steal their source code.
3) Quit and start your own site.
4) Profit!
Apparently worked for Orkut and FaceBook!
It would be so nice if it took creativity and hard work and perhaps a little luck to get rich, but apparently a stale idea and a little PHP/MySQL knowledge is enough these days.
Axing senior management isn't going to get Vista out the door any faster -- probably a lot slower because whoever comes it to pick up the pieces is going to have a hell of a job. It might make Windows 2021 (or whatever they're calling Vista 1.1) ship quicker but in the short run, it'll be chaos. Shareholders, for the most part, don't care about the long run -- they care about now.
Uh, yeah?
Sun claims that their grid is having problems handling the bandwidth of a DDoS. What part of this is hard to believe?
Announce a new service, expect a DDoS. That's the world we live in. The only thing Sun does that's different is admit it instead of dismissing it with some dopey error message (Bad Server! No Cookie for You!)
No. A badly designed prison is a waste of money that makes politicians desperate and willing to spend any amount of money to build the next version, and a populace eager to pay for it.
I'd rather have some decent computer scientists come in and design something so powerful and dangerous that people will realize the implications and squash the project. If we get something that's 5% better than the last system, then our liberties will get whittled away so slowly that most people won't ever even notice.
I used to do the "code-til-you-drop, then sleep until you can do it again" thing and I was incredibly productive. Now I have kids... and I'm still productive, but my life has a lot more structure. Interrupts are not necessarily a bad thing. If you're working on something important/interesting/compelling, then it's still going to be important/interesting/whatever after you change your two-year-old's poopy diaper. And if my code is so disorganized that I can't remember what I was doing ten minutes later, well, it probably wasn't going to work anyway!
Anyone want to partner with me to develop league of swimming hamsters or weight-lifting mice? Those are sports we can all enjoy (watching, anyway). They could be the WWF of the 21st century.
But wait, let me patent that, first.
Whether or not this is true, I know I'd pay good money for an mpg of that. (how much Xanax does it require to get a hamster to hover?)
Just give me until Monday so I can short a few thousand shares of Google...
If Google blocks IE, people are going to take the path of least resistance and use a different search engine. It's not as if there aren't other perfectly adequate search engines. Google doesn't have the sort of monopoly necessary to pull this off.
Slashdots apparent obsession with Google makes People Magazines "bennifer" story look like an idle fancy. Let's move on.
This one goes in the "mad scientist" file. Talk about complete loss of perspective.