Back in (Greek) Junior High we went for a one-day school trip to an island (Poros IIRC). The day plan consistend of letting a few dozens of 14-15 year-olds roam around the town unattended. Obviously, one of my group's first stops was at an Arcade, to spend some coins on things like Cadillacs and Dinosaurs or NBA Jam. Another group of friends called us to a back room to check something out. There was an electronic (electromechanic?) gambling machine that had a roullette (but with fewer segments than the classic) and you could bet on outcomes. I hadn't seen such a machine since my usual Arcades in the city were just, well, Arcades, so I was intrigued and stood by to watch some of my friends play. Born a geek, even before any formal statistics training, I started analyzing the game and the betting possibilities and I soon realized that - woops - the designers allowed for multiple bets per cycle, but had not calculated the combined odds properly, so there was a combination of bets that would give the player an advantage. So I told my friends to stop betting randomly and I showed them a winning strategy that would give them on average something like a quarter (the equivalent 20 or 50 drachma coin that arcades accepted back then) for every 4 or so (don't remember specifics) plays. After explaining, I oversaw a few rounds and then left. Yes, money was useful for me as well, but I had calculated the expected payout and judged that I had stayed enough at the arcade, so spending the rest of the day outside (and finding a good restaurant) was worth more for me than that amount. Sure enough, after several hours, we all met back at the boat. The 3-4 kids that had stayed at the arcade had each made around 2-3k drachma (I guess something like $20 of early 90's USD), which was not insubstantial for us back then (since I remember I felt about 20% regret for not staying myself, given the fact that I had not found such a great restaurant - hehe). Anyway, the moral of the story is that gambling games are not usually designed by PhD's, although I would have expected at least big lotteries like those in TFA to be designed by groups of excellent statisticians. And also that winning a gambling game is not always worth your time.
And he posted a video of a very lame chef "performance". Then, I assume someone from Benihana started posting very favorable reviews one after the other using different names but the exact same writing style (including full-capitalization of key terms). Then the blogger commented of how it is weird that all those posts have the same IP. Then this Benihana GM guy posts and says his company won't allow this libel, so they are trying to find out how to sue the blogger and tells the blogger to be brave and give his full name and contact details. The most hilarious point is when he closes said post with the phrase: BTW, are you Lebanese?
The results were not "FALSE" or "FAKE" in the sense you mean, Bing does not read the google result page and copy items, and managed to copy some fake items. These were perfectly legitimate sites that had a zero rank given the proposed query. When google engineers started clicking on those sites as a result of the - seemingly to Bing given the data at the time - unrelated queries, they fed Bing with user data connecting the search terms with the sites in question. Thus, the sites started ranking well. I would be really be surprised if Google is not doing the same thing - improving rankings based to what people click on. The only reason for Google not to use such data from users that prefer Bing, or Yahoo in cases where they have access to it, would be if they thought Bing and Yahoo is REALLY crap. I can't comment on that, since I use google myself - I haven't really tried anything else for years - but I wouldn't ever ignore my competitors. And when you talk about "fanbois" when you yourself appear biased beyond comprehension of TFA, it doesn't come out well.
Let me make this plain to you: Bing does not seem to care about which results Google returns. What it cares is the sites people choose to visit as a result of a search query. When people use bing for searching, bing can always get that piece of data and improve results, when people use their toolbar they can also get that information from queries running on other search engines.
The "fake results" were real sites that were just unrelated to the search. So bing was seing people after searching for something to be clicking on a website it thought irrelevant. In 7-9% of the cases that result got a good boost in Bing results, after all if people really do want to read that after doing that search, why not put it there? It is interesting that this worked for only a small percentage of fake results, when all the queries where strings that would not be typed naturally, so had no "real" results, although it is expected that Bing would not "blindly" use clickthrough data, but only along with other factors.
No, it is exactly what GP said. Bing seems to be "pushing up" results of searches that people click on. Yeah, the can be results of google, yahoo, bing or whatever, does it matter? Obviously they can run the algorithm more effectively on their own results, as they always have the data on what people clicked after a search, but why shouldn't they include such data from competitor queries when they can?
From the article it seems that MS is tracking which google (and I assume any other search engine, including Bing) search results people are clicking, and then trying to promote these in their results. It does sound like something very logical to do to improve search results, doesn't it?
Hmm, you honestly don't know? Strange for a/.er;) It is quite simple. Imagine you have a linux desktop, which boots directly into a Java VM. You can only use it to run compatible Java applications and don't have direct access to your machine. That is android. With Maemo/MeeGo you use it as your regular linux desktop, you have access to all of your hardware and software. Then, non-gui apps will compile mostly without any changes (gui apps depend on the app), and most useful ones are already compiled for you, so if you want cvs, you do "apt-get install cvs" (in Maemo) and there you go. Even if Maemo developers were very few compared to the Android developers, the fact that it is a full linux distro meant we got software that people are still struggling to port on Android (and at a performance penalty), e.g. I had dosbox, ScummVM and other great apps I use on linux, since day one of Maemo. I showed my N900 to a friend who seemed to like it and I told him that it can do many work-related things his linux laptop does. Well, he said, can it connect to my company's cisco VPN? Le't try, I told him, and a few minutes later we were connected with vpnc. Then he tells me, I also need remote desktop, and in a few seconds (apt-get install rdesktop) we were up and running! The list goes on. It is a bit harder to impress simple users, especially since Android is catching up, while it was very easy to impress (and mock) iphone users. I would just open a dozen web pages, which show up in their full glory (be it flash, javascript or whatever), and switch seamlessly between them. Then click on a video torrent link, download it with the bittorrent client and play it regardless if it was not prepared for mobile playback by itunes or whatever. What is interesting about MeeGo and Android though, is that if someone ports the latter's VM (Dalvik) to MeeGo, you will be able to have an android machine running as an app. Canonical was working on porting Dalvik to a regular linux kernel, but they seem to have stopped for some reason...
Android is a Java VM running on a linux kernel. MeeGo is a full linux distro. For simple users there might be no obvious advantage, for myself and some other people on the other hand the difference is that of day from night... And Nokia has been developing Maemo (MeeGo's predecessor) for more than 5 years, but they were stupid enough to devote very little resources and focus only on "internet tablets", since they believed (wanted) Symbian will always rule the smartphone world. Idiots.
All solutions I've heard so far require people calling international numbers. But do we know whether people have access to international numbers? And how are they going to learn about this service? And their whole problem is that they cannot coordinate their activities, being able to send tweets but not read them will not help much... Yeah, I can set up a dialup for Egyptian revolutionaries at my home. I can even post to twitter (ok, I have to make an account first, I'm not a twit myself) what they want to say if they call me at home. But even if they knew my number, it wouldn't really help them much. A slashdotter proposed we all start calling numbers at random in Egypt. While that is silly for many reasons (most obvious is the language issue), it is close to what I could consider a viable solution to the communication problem, provided inbound international calls are possible. So, instead of the people in Egypt using the internet to organize, they should use their family members abroad as proxies for organizing. You call your cousin in X place, he tells you about the planned activities and instead of trying to contact their inhabitants in X place, you contact their relatives in Queens... Anyway, just an idea, I do hope Mubarak relents soon and all this is not required.
The last time they had a serious error, they were trying to avoid it by saying that if you were not doing serious number theory research you would never notice. That didn't go well (and they later had to do the recall anyway), so they are not making such a mistake twice!
Actually this was news Friday morning... And it is not really a "surprising cameo" when SNL was trying to get him to come and we knew about it beforehand... That said, it is a shame that SNL has lost its mojo, it used to be an amazing show.
What is interesting is that those 2 SATA 6Gbps ports that the Intel boards have, are the ones unaffected by this! The problem is only with the other 4 ports. I bet they were thinking going with mostly the ol' 3Gbit ports will be safe and save some money... Woops!
According to Anand's coverage, Intel said that they started getting customer complaints after they had shipped about 100k units, and their engineers managed to duplicate the problem early last week, the cause of which they figured out in a couple of days.
The first thing that came to mind was: hey, Virtual Boy didn't look that bad after all! I was expecting something that looked closer to this: http://www.youtube.com/watch?v=DTXO7KGHtjI A blob, even holographic, is still a blob. There are many other things wrong with the video (the "acting", no metal bikinis, no second kinect camera at a different angle etc), so I was pretty underwhelmed...
At least it was not a bad year for animated features. Most notably Toy Story 3 lived up to the previous installments (hard feat) and I was very pleasantly surprised by "How to train your dragon". Speaking of Tron though, I don't understand how it is only nominated for Sound Editing. Visually it was an amazing experience (and hybrid 2D/3D depending on which world you were helped), but where it excelled was the soundtrack which was a masterful mixture of classical with electronic sound and was in complete harmony with what was going on in the picture. In fact, my wife who is quite the opposite of a computer geek (classical philologist) considered Tron the most entertaining movie of the year. Back to the soundtrack, it is one of the few which when I listen to I immediately "see" the scene that went with each track. And I am a big fan of Hans Zimmer, but the best moments of the Inception score are similar to just a small part of the Tron soundtrack, which builds much much more. Note that I am not accusing Daft Punk or Hans Zimmer for copying (Inception came out earlier, but Daft Punk had been working 2 years before on the Tron score so who knows if either heard the other), I am just saying that the Tron soundtrack was similar style but so much more, so I can't understand why Inception was nominated for original score instead of it. Well, ok, I do understand why, Daft Punk is not exactly "mainstream", so it was rather rhetorical... but it is still not fair.
I just finished the Civ 5 Mongol Scenario a few hours ago, which, as some of you know, is as accurate a simulation of the Genghis Khan conquests as a civ-based strategy game can be, which is damn more accurate than, say, any platform game I can think of. So, I can verify, re-enacting the whole Khan campaign, I did get the feeling of bliss after my efforts for intensive de-population and re-forestation were on their way. As I was reducing large cities to rumble and burning the population (to avoid their long term unhappiness - I told you Civ 5 is accurate), I would find real pleasure in contemplating the good I was doing for the planet, by replenishing its oxygen reserves, and exterminating the humans (at least the AI controlled ones) that is plaguing the earth...
They do have a history of interfering, probably with good intentions, but things never usually work out how you want them.
I guess you can define "good" as serving your own purposes with complete disregard of the rest of the world and absolutely no contemplation on the long run effects that could be worse than your short term gains.
Exactly. I have heard so many times, especially by Americans, that the solution is to overthrow the Iran government and establish a "democracy"... Read some friggin' history! The whole mess started BECAUSE the US overthrew the first good democratic government that Iran ever had, to protect the British petroleum interests (Operation Ajax). If I was an Iranian and had suffered the last 60 years because of that, I would be REALLY pissed, and possibly turn to religion, hate the West, need nuclear weapons to counter the ones provided to Israel, etc etc.
Sorry, but you are wrong. The PowerPC 970 core is not far behind modern cores in IPC, definately nowhere near than the more than 2x worse that you imply. Even though it is hard to compare them due to different ISA, you can take a look at common benchmarks such as Linpack etc. Looking at the latest Top-500 and comparing per core per clock, the only Power PC 970 system at #118 is anywhere from 40% slower (cmp to #33) to 20% faster (cmp to #39) to Nehalem-based systems depending the implementation (and on average 20-25% slower). And remember this is always per core, per clock, so at 3.2GHz the PowerPC 970 is not too shabby.
And yes, a modern midrange card is much more powerful than the RSX, but my whole point is that you can't tell that easily unless you crank up the resolution to more the usual 720p or 1080p of TVs. Yeah, a modern card would render most games at hundreds of FPS at 1280x720, so what?
There are two other important factors. First, the PS3 happened to be released at a point where CPU performance sort of maxed out per core. A 3.2GHz Cell core is not far from the cores that current flagship CPU's carry and the Cell was not alone either, it had 7 SPUs around it. Similarly the Xbox 360 had 3 of those 3.2GHz PPC cores. So a modern CPU with 4-8 symmetric cores is only faster than a PS3 or a Xbox 360 if you can use more than 3 of those cores in your game for most of the time. The second factor is that while the GPUs have become much, much faster than what the PS3 and Xbox 360 are using, the difference is easily visible in resolutions higher than the 720p or even 1080p that the consoles output to. Similarly, if the PSP2 had to output to say a 640x360 display, it wouldn't really need the PS3 GPU.
You are right, I did not notice that, the 24 pixel pipelines should be pulling the most FLOPS, so even in that metric it must be much faster. I did not doubt of course that Zacate is slower in CPU and GPU than the PS3, and it really is the upper limit (with current technology) of gaming on a netbook, the PS2 has to be slower than that as well.
But, yeah, the marketing dept must have found something they could "use".
Slashdot Mirror
Back in (Greek) Junior High we went for a one-day school trip to an island (Poros IIRC). The day plan consistend of letting a few dozens of 14-15 year-olds roam around the town unattended. Obviously, one of my group's first stops was at an Arcade, to spend some coins on things like Cadillacs and Dinosaurs or NBA Jam. Another group of friends called us to a back room to check something out. There was an electronic (electromechanic?) gambling machine that had a roullette (but with fewer segments than the classic) and you could bet on outcomes. I hadn't seen such a machine since my usual Arcades in the city were just, well, Arcades, so I was intrigued and stood by to watch some of my friends play. Born a geek, even before any formal statistics training, I started analyzing the game and the betting possibilities and I soon realized that - woops - the designers allowed for multiple bets per cycle, but had not calculated the combined odds properly, so there was a combination of bets that would give the player an advantage. So I told my friends to stop betting randomly and I showed them a winning strategy that would give them on average something like a quarter (the equivalent 20 or 50 drachma coin that arcades accepted back then) for every 4 or so (don't remember specifics) plays. After explaining, I oversaw a few rounds and then left. Yes, money was useful for me as well, but I had calculated the expected payout and judged that I had stayed enough at the arcade, so spending the rest of the day outside (and finding a good restaurant) was worth more for me than that amount.
Sure enough, after several hours, we all met back at the boat. The 3-4 kids that had stayed at the arcade had each made around 2-3k drachma (I guess something like $20 of early 90's USD), which was not insubstantial for us back then (since I remember I felt about 20% regret for not staying myself, given the fact that I had not found such a great restaurant - hehe).
Anyway, the moral of the story is that gambling games are not usually designed by PhD's, although I would have expected at least big lotteries like those in TFA to be designed by groups of excellent statisticians. And also that winning a gambling game is not always worth your time.
And he posted a video of a very lame chef "performance".
Then, I assume someone from Benihana started posting very favorable reviews one after the other using different names but the exact same writing style (including full-capitalization of key terms). Then the blogger commented of how it is weird that all those posts have the same IP. Then this Benihana GM guy posts and says his company won't allow this libel, so they are trying to find out how to sue the blogger and tells the blogger to be brave and give his full name and contact details. The most hilarious point is when he closes said post with the phrase:
BTW, are you Lebanese?
The results were not "FALSE" or "FAKE" in the sense you mean, Bing does not read the google result page and copy items, and managed to copy some fake items. These were perfectly legitimate sites that had a zero rank given the proposed query. When google engineers started clicking on those sites as a result of the - seemingly to Bing given the data at the time - unrelated queries, they fed Bing with user data connecting the search terms with the sites in question. Thus, the sites started ranking well.
I would be really be surprised if Google is not doing the same thing - improving rankings based to what people click on. The only reason for Google not to use such data from users that prefer Bing, or Yahoo in cases where they have access to it, would be if they thought Bing and Yahoo is REALLY crap. I can't comment on that, since I use google myself - I haven't really tried anything else for years - but I wouldn't ever ignore my competitors.
And when you talk about "fanbois" when you yourself appear biased beyond comprehension of TFA, it doesn't come out well.
Let me make this plain to you:
Bing does not seem to care about which results Google returns. What it cares is the sites people choose to visit as a result of a search query. When people use bing for searching, bing can always get that piece of data and improve results, when people use their toolbar they can also get that information from queries running on other search engines.
The "fake results" were real sites that were just unrelated to the search. So bing was seing people after searching for something to be clicking on a website it thought irrelevant. In 7-9% of the cases that result got a good boost in Bing results, after all if people really do want to read that after doing that search, why not put it there?
It is interesting that this worked for only a small percentage of fake results, when all the queries where strings that would not be typed naturally, so had no "real" results, although it is expected that Bing would not "blindly" use clickthrough data, but only along with other factors.
No, it is exactly what GP said. Bing seems to be "pushing up" results of searches that people click on. Yeah, the can be results of google, yahoo, bing or whatever, does it matter? Obviously they can run the algorithm more effectively on their own results, as they always have the data on what people clicked after a search, but why shouldn't they include such data from competitor queries when they can?
From the article it seems that MS is tracking which google (and I assume any other search engine, including Bing) search results people are clicking, and then trying to promote these in their results.
It does sound like something very logical to do to improve search results, doesn't it?
Hmm, you honestly don't know? Strange for a /.er ;)
It is quite simple. Imagine you have a linux desktop, which boots directly into a Java VM. You can only use it to run compatible Java applications and don't have direct access to your machine. That is android.
With Maemo/MeeGo you use it as your regular linux desktop, you have access to all of your hardware and software. Then, non-gui apps will compile mostly without any changes (gui apps depend on the app), and most useful ones are already compiled for you, so if you want cvs, you do "apt-get install cvs" (in Maemo) and there you go. Even if Maemo developers were very few compared to the Android developers, the fact that it is a full linux distro meant we got software that people are still struggling to port on Android (and at a performance penalty), e.g. I had dosbox, ScummVM and other great apps I use on linux, since day one of Maemo.
I showed my N900 to a friend who seemed to like it and I told him that it can do many work-related things his linux laptop does. Well, he said, can it connect to my company's cisco VPN? Le't try, I told him, and a few minutes later we were connected with vpnc. Then he tells me, I also need remote desktop, and in a few seconds (apt-get install rdesktop) we were up and running! The list goes on.
It is a bit harder to impress simple users, especially since Android is catching up, while it was very easy to impress (and mock) iphone users. I would just open a dozen web pages, which show up in their full glory (be it flash, javascript or whatever), and switch seamlessly between them. Then click on a video torrent link, download it with the bittorrent client and play it regardless if it was not prepared for mobile playback by itunes or whatever.
What is interesting about MeeGo and Android though, is that if someone ports the latter's VM (Dalvik) to MeeGo, you will be able to have an android machine running as an app. Canonical was working on porting Dalvik to a regular linux kernel, but they seem to have stopped for some reason...
Android is a Java VM running on a linux kernel. MeeGo is a full linux distro. For simple users there might be no obvious advantage, for myself and some other people on the other hand the difference is that of day from night...
And Nokia has been developing Maemo (MeeGo's predecessor) for more than 5 years, but they were stupid enough to devote very little resources and focus only on "internet tablets", since they believed (wanted) Symbian will always rule the smartphone world.
Idiots.
All solutions I've heard so far require people calling international numbers. But do we know whether people have access to international numbers? And how are they going to learn about this service? And their whole problem is that they cannot coordinate their activities, being able to send tweets but not read them will not help much...
Yeah, I can set up a dialup for Egyptian revolutionaries at my home. I can even post to twitter (ok, I have to make an account first, I'm not a twit myself) what they want to say if they call me at home. But even if they knew my number, it wouldn't really help them much.
A slashdotter proposed we all start calling numbers at random in Egypt. While that is silly for many reasons (most obvious is the language issue), it is close to what I could consider a viable solution to the communication problem, provided inbound international calls are possible.
So, instead of the people in Egypt using the internet to organize, they should use their family members abroad as proxies for organizing. You call your cousin in X place, he tells you about the planned activities and instead of trying to contact their inhabitants in X place, you contact their relatives in Queens...
Anyway, just an idea, I do hope Mubarak relents soon and all this is not required.
The last time they had a serious error, they were trying to avoid it by saying that if you were not doing serious number theory research you would never notice. That didn't go well (and they later had to do the recall anyway), so they are not making such a mistake twice!
Actually this was news Friday morning... And it is not really a "surprising cameo" when SNL was trying to get him to come and we knew about it beforehand...
That said, it is a shame that SNL has lost its mojo, it used to be an amazing show.
Source: http://news.cnet.com/8301-13577_3-20029974-36.html
What is interesting is that those 2 SATA 6Gbps ports that the Intel boards have, are the ones unaffected by this! The problem is only with the other 4 ports. I bet they were thinking going with mostly the ol' 3Gbit ports will be safe and save some money... Woops!
According to Anand's coverage, Intel said that they started getting customer complaints after they had shipped about 100k units, and their engineers managed to duplicate the problem early last week, the cause of which they figured out in a couple of days.
Source : http://www.anandtech.com/show/4142/intel-discovers-bug-in-6series-chipset-begins-recall
Or take the weight of the Library of Congress and divide it by the air speed velocity of an unladen swallow...
Oh, right I forgot. An African swallow.
The first thing that came to mind was: hey, Virtual Boy didn't look that bad after all!
I was expecting something that looked closer to this: http://www.youtube.com/watch?v=DTXO7KGHtjI
A blob, even holographic, is still a blob.
There are many other things wrong with the video (the "acting", no metal bikinis, no second kinect camera at a different angle etc), so I was pretty underwhelmed...
At least it was not a bad year for animated features. Most notably Toy Story 3 lived up to the previous installments (hard feat) and I was very pleasantly surprised by "How to train your dragon".
Speaking of Tron though, I don't understand how it is only nominated for Sound Editing. Visually it was an amazing experience (and hybrid 2D/3D depending on which world you were helped), but where it excelled was the soundtrack which was a masterful mixture of classical with electronic sound and was in complete harmony with what was going on in the picture. In fact, my wife who is quite the opposite of a computer geek (classical philologist) considered Tron the most entertaining movie of the year.
Back to the soundtrack, it is one of the few which when I listen to I immediately "see" the scene that went with each track. And I am a big fan of Hans Zimmer, but the best moments of the Inception score are similar to just a small part of the Tron soundtrack, which builds much much more. Note that I am not accusing Daft Punk or Hans Zimmer for copying (Inception came out earlier, but Daft Punk had been working 2 years before on the Tron score so who knows if either heard the other), I am just saying that the Tron soundtrack was similar style but so much more, so I can't understand why Inception was nominated for original score instead of it.
Well, ok, I do understand why, Daft Punk is not exactly "mainstream", so it was rather rhetorical... but it is still not fair.
I just finished the Civ 5 Mongol Scenario a few hours ago, which, as some of you know, is as accurate a simulation of the Genghis Khan conquests as a civ-based strategy game can be, which is damn more accurate than, say, any platform game I can think of.
So, I can verify, re-enacting the whole Khan campaign, I did get the feeling of bliss after my efforts for intensive de-population and re-forestation were on their way. As I was reducing large cities to rumble and burning the population (to avoid their long term unhappiness - I told you Civ 5 is accurate), I would find real pleasure in contemplating the good I was doing for the planet, by replenishing its oxygen reserves, and exterminating the humans (at least the AI controlled ones) that is plaguing the earth...
Funny how in the correction they name Qwest twice, and the actual perpetrator not at all.
They do have a history of interfering, probably with good intentions, but things never usually work out how you want them.
I guess you can define "good" as serving your own purposes with complete disregard of the rest of the world and absolutely no contemplation on the long run effects that could be worse than your short term gains.
Exactly. I have heard so many times, especially by Americans, that the solution is to overthrow the Iran government and establish a "democracy"... Read some friggin' history! The whole mess started BECAUSE the US overthrew the first good democratic government that Iran ever had, to protect the British petroleum interests (Operation Ajax). If I was an Iranian and had suffered the last 60 years because of that, I would be REALLY pissed, and possibly turn to religion, hate the West, need nuclear weapons to counter the ones provided to Israel, etc etc.
Hehe, so, you've read Forever Free I guess...
Actually, this is very good news.
It's great news. And perhaps he will have some more time to concentrate on how to bring one of the best sci-fi novels (Forever War) to the big screen.
Sorry, but you are wrong. The PowerPC 970 core is not far behind modern cores in IPC, definately nowhere near than the more than 2x worse that you imply. Even though it is hard to compare them due to different ISA, you can take a look at common benchmarks such as Linpack etc. Looking at the latest Top-500 and comparing per core per clock, the only Power PC 970 system at #118 is anywhere from 40% slower (cmp to #33) to 20% faster (cmp to #39) to Nehalem-based systems depending the implementation (and on average 20-25% slower). And remember this is always per core, per clock, so at 3.2GHz the PowerPC 970 is not too shabby.
And yes, a modern midrange card is much more powerful than the RSX, but my whole point is that you can't tell that easily unless you crank up the resolution to more the usual 720p or 1080p of TVs. Yeah, a modern card would render most games at hundreds of FPS at 1280x720, so what?
There are two other important factors.
First, the PS3 happened to be released at a point where CPU performance sort of maxed out per core. A 3.2GHz Cell core is not far from the cores that current flagship CPU's carry and the Cell was not alone either, it had 7 SPUs around it. Similarly the Xbox 360 had 3 of those 3.2GHz PPC cores. So a modern CPU with 4-8 symmetric cores is only faster than a PS3 or a Xbox 360 if you can use more than 3 of those cores in your game for most of the time.
The second factor is that while the GPUs have become much, much faster than what the PS3 and Xbox 360 are using, the difference is easily visible in resolutions higher than the 720p or even 1080p that the consoles output to. Similarly, if the PSP2 had to output to say a 640x360 display, it wouldn't really need the PS3 GPU.
You are right, I did not notice that, the 24 pixel pipelines should be pulling the most FLOPS, so even in that metric it must be much faster. I did not doubt of course that Zacate is slower in CPU and GPU than the PS3, and it really is the upper limit (with current technology) of gaming on a netbook, the PS2 has to be slower than that as well.
But, yeah, the marketing dept must have found something they could "use".