The "mishmash" of components thing hasn't been true ever since nforce came out. AMD chipsets used to suck hard, but I've never had a problem with an nforce board.
It's pretty surprising given that, clock for clock, a G5 is closer to an Opteron than a Pentium 4. An Opteron at 2.6GHz can certainly keep up with a 3.6GHz P4 in pretty much anything, so a 2.7GHz G5 should certainly be able to do it.
Expect to see it cracked within two weeks. Even faster if its on-chip Intel DRM, because there should be a general crack for that as soon as chips that support it come out.
Sun's SPARCs lag behind even cheap x86 chips in SPEC. The top UltraSPARC is a little more than half the speed of the top Power5 in integer, and less than half the speed in floating-point. The top SPARC64V edges out the top Power5 in integer, but is about 60% as fast in FP code. The top Opteron beats all three in integer, and beats the top UltraSPARC in FP (by a lot), and loses to the top SPARC64V in FP (by a few percent), and to the top Power5 in FP (by a lot).
Spiffy. I'll have to try hard at the patience thing, though, I'm not the most patient of individuals by nature:) Just gotta keep the end result in mind, huh?
Where do you pipe the digital output to? At some point in your system, you've got a DAC. It's highly likely that the DAC in whatever you've plugged in the digital output isn't a good one.
The price of games has hovered around $50 for a decade now. Inflation alone says that the price should be over $60 by now. Have you looked at the price of milk lately???
It's Sony's fault that people are clueless? The 2 teraflops figure is not a meaningless number --- it has meaning within the context of graphics. For example, ATI is perfectly happy to throw around 200+ gigaflops numbers for their R4xx chips.
I'm a programmer. Citing 200 single-precision gigaflops for Cell means something to me. I can look at the 8 SPEs and the 256KB of cache, and get a picture of the real-world performance I'm going to get out of it for different R400 gigaflopstasks. Just because a clueless end-user can't make sense of it is no reason to label a perfectly meaningful number as "hype". It's like saying that BlueGene's 367 teraflops numbers are "hype". It's not like it'll ever hit 300+ teraflops running anything except LinPACK, but it gives programmers an idea of where the ceiling is.
It's not that I'm buying into the hype. It's just that the hype isn't coming from Sony or IBM, but rather from journalists. Since I ignore the technology media completely, and I don't see any reason why anybody would pay attention to them, all I've been seeing are things like IBM's ISSCC presentation.
I just think its silly to pin all this hype on Sony and IBM (eg: PS3 will have 4 Cells), when they didn't actually say any of it! It was all "journalists" extrapolating crap from rumors.
Dude, relax. Where did you read "it'll be like jacking up to the matrix"? Certainly, Sony never said anything like that. Journalists might have, but there is a reason you should never read crap like EGM.
As for the games, yes, the first-gen PS2 games were dissapointing. That was because the PS2 was a complete bitch to program. The second and third-gen games, however, were completely worth it. Actually, I just got a PS2 a couple of days ago (though, my brother has owned one for awhile). I have to say, the new games look as good as my roommate's XBox, though the textures aren't as detailed. They look a heck of lot nicer than my Gamecube, which has been gathering dust in my closet for two years.
Sony never claimed the PS3 is powerful enough to get deep into top 500 lists. "Techies" with not enough knowledge are seeing the 2 teraflops figure and jumping to that conclusion.
Saying the PS3 can do 2 teraflops isn't a lie. The question is, what kind of teraflops are they? The thing is a game machine, so it makes sense that they report the 1.8 teraflops on the graphics card. However, most of those are very limited operations. They're stuff like hardwired perspective correction or interpolation for looking up a texture. Those numbers are entirely relevant if you're doing graphics, not so relevant if you're computing fluid dynamics. In comparison, a supercomputer rated at 2 teraflops can handle 2 trillion general purpose douple-precision operations per second. Great for fluid dynamics calculations, but you don't need that kind of generality for graphics.
Again, I think most of the hype is in the minds of readers who don't really understand the information STI is putting out. IBM and Sony's info on Cell has been rather dry and technical. It's the "tech media" that's been drawing all sorts of conclusions from it.
First, I've never heard of the 170M number until now, the previous poster just made it up. Second, the 70M number doesn't need disproving. It's a perfectly valid number for what it is --- a measure of the console's flat-shading performance. It's like the "11 teraflops" number you see for a supercomputer running LinPACK. It allows people with a clue to gauge what the console will perform like.
Sony never claimed 160 million polygons per second. These numbers have become fabricated over time. They did claim 70M polygons per second, but nobody ever quotes the context. It was 70M unshaded polygons per second. The number for shaded/textured polygons (in the PS2 datasheet) is 20-30M, and the most recent games are indeed in the (lower end) of that range.
I'd also point out the "Toy Story in Real Time" thing was never a Sony claim either. As far as I've been able to track it down, it was some idiot journalist that made the claim, not a Sony spokesperson or any Sony marketing literature.
I think it'd be idiotic to say that the EE sucks. My PS2 plays games a sight better than my 2GHz p4 (with GeForce MX 440). That's quite impressive considering that the PS2's CPU is running at a clockrate nearly 7x slower, and its GPU is running at about half the clockrate. Of course, you're also ignoring the fact that after many years of work, game developers have been able to get the EE to perform reasonably close to its theoretical peak.
The "Cell is overhyped" thing is complete nonsense. These consoles will be more powerful than PCs on introduction (if only for the equivalent of a GeForce 7800GTX running at 550MHz!), and thanks to developers figuring out how to use the Cell's power over time, will still run games as well as contemporary PCs a few years down the line.
Slashdot Mirror
The "mishmash" of components thing hasn't been true ever since nforce came out. AMD chipsets used to suck hard, but I've never had a problem with an nforce board.
In Athlon X2 we trust, amen.
The G3 was never "up to twice as fast as a PII", just like the G5 was never faster than a P4 (which Apple claimed it was).
Except that the developer machines aren't the new Macs. They're fairly pedestrian Intel machines, contemporary to the current line of G5 macs.
Actually. I believe breathing in dihydrogen monoxide causes a fatal condition known as "drowning".
It's pretty surprising given that, clock for clock, a G5 is closer to an Opteron than a Pentium 4. An Opteron at 2.6GHz can certainly keep up with a 3.6GHz P4 in pretty much anything, so a 2.7GHz G5 should certainly be able to do it.
Expect to see it cracked within two weeks. Even faster if its on-chip Intel DRM, because there should be a general crack for that as soon as chips that support it come out.
Sun's SPARCs lag behind even cheap x86 chips in SPEC. The top UltraSPARC is a little more than half the speed of the top Power5 in integer, and less than half the speed in floating-point. The top SPARC64V edges out the top Power5 in integer, but is about 60% as fast in FP code. The top Opteron beats all three in integer, and beats the top UltraSPARC in FP (by a lot), and loses to the top SPARC64V in FP (by a few percent), and to the top Power5 in FP (by a lot).
It's not the volume discount. It's the discount from being an "all Intel" shop like Dell.
If I recall correctly, that's coding theory, and the decimal thing is the principle behind arithmetic coding.
Without the government-funded model, we would never even have gotten into space. If corporations are anything, they are risk-averse.
Spiffy. I'll have to try hard at the patience thing, though, I'm not the most patient of individuals by nature :) Just gotta keep the end result in mind, huh?
Where do you pipe the digital output to? At some point in your system, you've got a DAC. It's highly likely that the DAC in whatever you've plugged in the digital output isn't a good one.
What kit did you use, btw? I'm going to put together a pair of Ellis 1801s this summer and am looking forward to listening to them.
The price of games has hovered around $50 for a decade now. Inflation alone says that the price should be over $60 by now. Have you looked at the price of milk lately???
It's Sony's fault that people are clueless? The 2 teraflops figure is not a meaningless number --- it has meaning within the context of graphics. For example, ATI is perfectly happy to throw around 200+ gigaflops numbers for their R4xx chips.
I'm a programmer. Citing 200 single-precision gigaflops for Cell means something to me. I can look at the 8 SPEs and the 256KB of cache, and get a picture of the real-world performance I'm going to get out of it for different R400 gigaflopstasks. Just because a clueless end-user can't make sense of it is no reason to label a perfectly meaningful number as "hype". It's like saying that BlueGene's 367 teraflops numbers are "hype". It's not like it'll ever hit 300+ teraflops running anything except LinPACK, but it gives programmers an idea of where the ceiling is.
It's not that I'm buying into the hype. It's just that the hype isn't coming from Sony or IBM, but rather from journalists. Since I ignore the technology media completely, and I don't see any reason why anybody would pay attention to them, all I've been seeing are things like IBM's ISSCC presentation.
I just think its silly to pin all this hype on Sony and IBM (eg: PS3 will have 4 Cells), when they didn't actually say any of it! It was all "journalists" extrapolating crap from rumors.
It's Sony's fault that your friend is a complete nitwit? It should teach you not to associate with any idiot who wastes money on game magazines...
Dude, relax. Where did you read "it'll be like jacking up to the matrix"? Certainly, Sony never said anything like that. Journalists might have, but there is a reason you should never read crap like EGM.
As for the games, yes, the first-gen PS2 games were dissapointing. That was because the PS2 was a complete bitch to program. The second and third-gen games, however, were completely worth it. Actually, I just got a PS2 a couple of days ago (though, my brother has owned one for awhile). I have to say, the new games look as good as my roommate's XBox, though the textures aren't as detailed. They look a heck of lot nicer than my Gamecube, which has been gathering dust in my closet for two years.
Anandtech, a site run by and for clueless gamers, or companies, run for and by heartless investors?
Screwed either way.
This is how rumors get started.
Sony never claimed the PS3 is powerful enough to get deep into top 500 lists. "Techies" with not enough knowledge are seeing the 2 teraflops figure and jumping to that conclusion.
Saying the PS3 can do 2 teraflops isn't a lie. The question is, what kind of teraflops are they? The thing is a game machine, so it makes sense that they report the 1.8 teraflops on the graphics card. However, most of those are very limited operations. They're stuff like hardwired perspective correction or interpolation for looking up a texture. Those numbers are entirely relevant if you're doing graphics, not so relevant if you're computing fluid dynamics. In comparison, a supercomputer rated at 2 teraflops can handle 2 trillion general purpose douple-precision operations per second. Great for fluid dynamics calculations, but you don't need that kind of generality for graphics.
Again, I think most of the hype is in the minds of readers who don't really understand the information STI is putting out. IBM and Sony's info on Cell has been rather dry and technical. It's the "tech media" that's been drawing all sorts of conclusions from it.
First, I've never heard of the 170M number until now, the previous poster just made it up. Second, the 70M number doesn't need disproving. It's a perfectly valid number for what it is --- a measure of the console's flat-shading performance. It's like the "11 teraflops" number you see for a supercomputer running LinPACK. It allows people with a clue to gauge what the console will perform like.
Sony never claimed 160 million polygons per second. These numbers have become fabricated over time. They did claim 70M polygons per second, but nobody ever quotes the context. It was 70M unshaded polygons per second. The number for shaded/textured polygons (in the PS2 datasheet) is 20-30M, and the most recent games are indeed in the (lower end) of that range.
I'd also point out the "Toy Story in Real Time" thing was never a Sony claim either. As far as I've been able to track it down, it was some idiot journalist that made the claim, not a Sony spokesperson or any Sony marketing literature.
I think it'd be idiotic to say that the EE sucks. My PS2 plays games a sight better than my 2GHz p4 (with GeForce MX 440). That's quite impressive considering that the PS2's CPU is running at a clockrate nearly 7x slower, and its GPU is running at about half the clockrate. Of course, you're also ignoring the fact that after many years of work, game developers have been able to get the EE to perform reasonably close to its theoretical peak.
The "Cell is overhyped" thing is complete nonsense. These consoles will be more powerful than PCs on introduction (if only for the equivalent of a GeForce 7800GTX running at 550MHz!), and thanks to developers figuring out how to use the Cell's power over time, will still run games as well as contemporary PCs a few years down the line.
Touche.