Disablying the cache is kinda dumb. What would be smarter is if you could "lock" a tiny portion of the cache (a couple pages) so it isn't swapped out and is then directly accessible as a virtual address to the CPU (with the prerequisite page protection). Then insure that the CPU clears the cache on initialization and that there is no easy way to access the cache without validating the memory page privileges for the accessing thread.
There are many CPU's with directly addressable caches. They're common in video game development dating all the way back to the caches for the two coprocessors on the Atari Jaguar. There's also the ScratchPad RAM on the PS2 and the VU0/VU1 memory areas. On the XBOX360, a portion of the L2 cache can be locked so it is directly addressable (even by the external GPU). On the PS3, the ram in the SPU's can be treated as a lockable cache (and there is a "secure" mode to run them securely for just such purposes).
I've used XP and Vista both. I have Vista at home and XP at work. When you first use Vista, it's much "clunkier" to get it set up (with all the UAC prompts and such). Your first week or two on Vista is painful and feels much worse than XP.
Also, your new Vista computer will run really slow for the first week or two while DL'ing updates, indexing the drive for Windows Search, Creating System Restore points (which is very slow and happens on each of the updates), etc.
However, after you pass this initial hump of pain, Vista actually seems to run much more smoothly and it looks considerably nicer than XP if you're running Aero (Vista Basic is just a waste compared to XP). By that time, enough preferences are configured and user programs installed that UAC doesn't up every 5 min (sometimes not at all for days or weeks). But a whole lot of reviews for Vista came after trying it out for a week or two and it's those first two weeks that are the most painful:-(
Newegg was hosting the SD1A firmware. I DL'd it from there (that's also where I bought the drive). However, they don't have a direct link to the firmware anymore and just have the instructions to get the firmware from Seagate.
I did the firmware update and it was pretty quick and painless.
I know it sounds outrageous to US but the terrorists are playing on a ball field with completely different rules when it comes to respecting the life of innocents.
Ok, sizzle chest. Next time you're on a plane and someone gets up and starts ranting about "sudden depressurization" and "water landings" you should jump on that person before the plane gets high jacked.
That's exactly what they're supposed to do: minimize risk to passengers and the flight. A person like that might not be actually high-jacking or bombing the plane -- they might be crazy and off-meds, overly drunk, or just an extreme asshole. However, they present a unknown yet real (and possibly lethal) risk to other passengers and to the safety of the flight. It's safer for the other 300 passengers on the plane if they are subdued. The acts of jumping around or screaming about horrible accidents (to frighten other passengers) are both illegal on airplanes. If they cause any serious disruption, the first thing that's going to happen is the plane is going to have to land at the nearest airport (which is not your destination airport) and it's at a minimum going to be an awful pain in the ass for the 299 people who behaved appropriately.
And yes, if a group of people were actively planning on how to deal with an upcoming "sudden depressurization" seriously, I'd probably be concerned enough to at least report it to the flight attendants. I'd certainly keep my eye on them for other suspicious behavior.
It doesn't necessarily mean they're guilty of anything but still, if you overhear any group seriously talking amongst themselves in a coordinated way about the impending destruction of the airplane you are on or are about to board, you probably should report them to the proper authorities whether they are white, black, asian, yellow, middle eastern, etc.
Three of the young adults make a remark about where would be the safest place to sit on the plane in the event of an accident or explosion.
Making *ANY* remark about a "possible explosion" on a plane loud enough that several other waiting passengers can hear you in the waiting area is likely to get you kicked off the flight. I'm all against profiling but they were pretty stupid to talk about a subject like that. You don't use the words "explosion" or "bomb" in an airport the same way you don't shout "fire" in a theater.
Microsoft uses an Irish tax haven to keep billions of their dollars out of reach of the American tax man.
There's a US tax law loophole that allow deferring taxes from profits made outside the US as long as the funds from the profits are not transferred into the US. This is one of the same loopholes that makes outsourcing to a foreign branch (i.e. cutting US jobs) tax-free compared to doing the same work in the US:-(
FWIW, I think 1 finger (one hand holding the device, the other using a finger), 2 thumbs (both hands holding using thumbs), or 1 thumb (same thumb as holding hand - i.e. 1-hand input) is better than 2 thumbs + 1 finger (which I can't figure out how that would be useful).
I'm an Omnikey man myself - been using one for about 20 years now. There is a company that makes an Omnikey clone but with windows keys. And when I say clone, I don't mean a cheap knock-off, the Avant Prime is near perfect down to using the same Alps switches. It's not cheap at $150 but I feel that mine was well worth it -- Keep in mind you will need a $15 PS2-to-USB adapter if you do not have a serial-PS2 keyboard input.
If you like the clicky feel, there is also the DAS Keyboard which is slightly cheaper at $129 and has USB (+ hub). They have a silly "Ultimate" version which has all blank keys to thwart anyone else from using your computer (err I mean to improve your typing skils). A friend of mine bought one (with the letters on it - not the blank keys). However, after he tested both his D.K. and my APrime, he was lusting over my APrime.
When they refer to "megawatts" in a power plant or storage, they're generally referring to the peak output capability. For example, the 20 Megawatts almost instantaneously would mean the batteries can supply a peak power of 20 Megawatts. The "almost instantaneously" means there is no significant time delay between the request for more power and actually getting the power out (which you might have with a coal fired plant for example).
The "megawatt hours" defines the total storage capability of the battery banks rather than the on-demand peak output.
FWIW it *WILL* cost you more in the long run for that cheaper PSU. If, as you claim, you *NEVER* power down your PC over a 3 year period of use, you will actually save money with a $50 PSU that is 80-90% efficient compared to a $18 PSU that is 60-70% efficient.
Actually, most of the SLC SSD drives are fairly immune to the random write stall issue that plagues MLC drives. For example, compare SLC and MLC drives from OCZ. The older OCZ Core SSD drives (SLC) have much faster random write access than newer OCZ Core V2 SSD drives (MLS) even though the Core II have much higher specified/published (sequential) write speeds.
OCZ's official line on the frightening performance problems with random writes on MLC drives (i.e. multi-second system stalls and random write throughput as low as 4 writes/second) is "we encourage potential customers to research this product and insure that it will fill their needs. These MLC based drives have extremely fast reads, and if you need a drive with fast sequential (frequent) writes, please check into our SLC based SATA II drive series."
At least OCZ is somewhat honest up front in acknowledging that their MLC drives are not for everyone. But FWIW, nearly all MLC SSD drives are orders of magnitude for real world performance (that includes writes) than their sequential performance specs would suggest.
Currently, the Intel drives are the only shipping MLC drives with good random write performance out-of-the-box. OCZ has announced (but is not yet shipping) a new "Vertex" series SSD that combines MLC with 64MB of RAM cache that speeds up random writes tremendously.
But in general, right now, it's buyer beware if you need fast random write access for higher system performance (i.e. a Windows user). Make sure you get either one of the Intel drives (MLC or SLC) or a well known SLC drive if you're concerned about anything other than strict read performance. Before you buy a MLC drive, follow OCZ's suggestion and do a lot of research on the drive first.
IBM is probably banking on the existence of people who want Cell processors in systems with more than 256megs of RAM. Other IBM value-adds would presumably include rack mountability, support for netbooting and other convenient management stuff, and so forth.
The newest Cell processors from IBM have two major features that customers wanted:
1) Support for normal DDR2 RAM. The original CELL (PS3 version) only supports RAMBUS XDR Memory. There is support for 16GB of DDR vs 256MB of XDR (PS3). 2) Support for pipelined double-precision operations on the SPU's. The original CELL (PS3 version) stalls the SPU's during all double-precision floating point operations and has no support for vectorized double-precision. The pipelining can increase the throughput of common double-precision operations by a factor of 5-8. There is still no support for vectorized double-precision floats.
20% less sodium by volume: Your Mom should try Kosher Salt Flakes. They're like salt flavored snowflakes (which fluffs up the volume) and they tend to stick to the outside of food easily so you get a salty "taste" with less (by mass) salt.
I'm assuming that if the moon was made from a fairly large amounts of ejected matter, that it would have formed from a gradual gravitational aquisition of the mass. Those other moons and planets that form this way have pretty large angular momentum.
Also important to note, the moon is pretty much tidally locked (the same side always faces the earth). It's inconceivable that Earth's gravitational field did not play an extremely key role in the current angular momentum of the moon. For more information, read the section Tidal Coupling in the Earth-Moon System in the linked article.
It also says that Mac OS will be able to optimize for SSD in the future better because OSX is a closed OS. (Which IIRC, the exact opposite is true: Windows is closed source and the lower levels of OSX are open source)
What sort of support do users expect from a $0.99 app. A big point of this article is there is so much shovelware (or as the articles call them "ringtone apps") at the $0.99 price range that users just buy and if it doesn't work, they don't care because $0.99 is cheap.
It's quite possible you will sell more than 5X the number of units at $0.99 than at a $4.99 price point. If that's the case, you will actually make more money selling the app at $0.99 than $4.99.
The big problem is there is no way for an iPhone user to tell which are the quality $0.99 apps and which are the crappy $0.99 apps. If there were a way for users to see which $0.99 apps were the highest quality, they would sell significantly better and increase the payoff for the developers.
FWIW, I had a T-Mobile G1 Android briefly before returning it. There are currently no pay applications in the Android market (or at least none that I could find -- they were all FREE). The girl at the T-Mobile store mentioned it as a plus that all the apps on Android are currently free.
Unfortunately, with the exception of a few ports from iPhone or other cell-phone apps (i.e. Shazam, Pacman, etc.), most of the G1 applications are lagging *FAR* behind their iPhone counterparts.
FWIW, I bought one of these at T-Mobile because the customer rep told me I could return it for a full refund (minus activation fee and plan usage) within 14 days. I returned it after 11 days.
I took a weeklong trip to CA and besides having to charge the phone 2-3 times a days (battery life sucks), the GPS somehow kept telling me I was in Maryland everytime I checked in when I was in San Francisco's Chinatown.
The phone also had a tendency to "drunk dial" people for me. It called one of my friends about 10 times for 1 second calls while it was in my pocket.
The keyboard is kinda clunky and I wish it used the accelerometer to automatically switch to landscape without opening the keytray. I also wish it had a virtual keypad for one-handed text entry (since with the tray open, you *NEED* two hands to hold and type). Finally, it's totally retarded not having at least an on-screen "number" pad for entering phone numbers. When you edit contacts, you have to enter their digits using the tiny little number keys at the top of the keypad. Selecting the phone number to edit in a contact should bring up a virtual dial-pad that uses the large screen in a standard 3x4 touch tone dialing configuration.
Another thing I didn't like was when you are in an area with a lot of WIFI networks (hotels), the available WIFI networks list keeps updating very quickly and doesn't "hold still" to allow you to easily select the one you want. I was trying to scroll down one screen to select one (out of about 15 available) network and before I scrolled and selected, it would refresh. I got it on the 8th or 9th try but it's just another example of clunky inelegance for the UI.
Oh, and I had some drawing glitches and the phone crashed about 4 times in the week that I had it.
I returned it. The G1 is worlds ahead of everything but the iPhone but it's still not currently any more usable than most Google "beta" programs. I'm guessing that in a year or two, when they've done a couple updates it will be more polished.
I would buy an iPhone but they majorly screwed up one important thing on the iPhone -- the volume. I can't hear anything when trying to make a call on the test iPhones at the store and the speaker phone goes to about 2.5 on a volume scale of 10. My buddies who have iPhones swear you can hear them fine if you use the headphones, or you hold the small hole firmly against your ear in exactly the right location (which has about a 1mm tolerance before it becomes unhearable). The G1 Google Phone had very good volume levels.
Right now, I'm just waiting for them to fix all the "beta-ness" of the G1 or double the volume of all the speakers on the iPhone before I pick up a new phone.
That's ironic because I have been programming the Cell processor at my job for the last three years. As a professional video game programmer, I do about half my coding on the PS3 Cell Processor:-)
I do most of the rest of my coding on the XBOX 360 which is also multicore but is a bit easier than the Cell to program.
If you want to learn Cell multiprocessor programming though, you can easily pick up a used PS3 for about $250 on Craigslist and set it up to run Linux. You don't get access to the GPU so hardware-accelerated graphics isn't possible but you can do a fair amount of Cell SPU multicore coding without too large an investment. If you plan on remotely targeting the PS3, I suggest setting up a simple VirtualBox VM on your home PC is you run Windows. Also, this site is a pretty good start for PS3 home brew performance optimizing for the Cell processor.
You won't run into any of the false sharing issues (which is a big deal on XBOX 360 performance) since there's only one PPC core and the SPU's are basically "DMA" driven for access to main RAM.
Slashdot Mirror
Disablying the cache is kinda dumb. What would be smarter is if you could "lock" a tiny portion of the cache (a couple pages) so it isn't swapped out and is then directly accessible as a virtual address to the CPU (with the prerequisite page protection). Then insure that the CPU clears the cache on initialization and that there is no easy way to access the cache without validating the memory page privileges for the accessing thread.
There are many CPU's with directly addressable caches. They're common in video game development dating all the way back to the caches for the two coprocessors on the Atari Jaguar. There's also the ScratchPad RAM on the PS2 and the VU0/VU1 memory areas. On the XBOX360, a portion of the L2 cache can be locked so it is directly addressable (even by the external GPU). On the PS3, the ram in the SPU's can be treated as a lockable cache (and there is a "secure" mode to run them securely for just such purposes).
I've used XP and Vista both. I have Vista at home and XP at work. When you first use Vista, it's much "clunkier" to get it set up (with all the UAC prompts and such). Your first week or two on Vista is painful and feels much worse than XP.
:-(
Also, your new Vista computer will run really slow for the first week or two while DL'ing updates, indexing the drive for Windows Search, Creating System Restore points (which is very slow and happens on each of the updates), etc.
However, after you pass this initial hump of pain, Vista actually seems to run much more smoothly and it looks considerably nicer than XP if you're running Aero (Vista Basic is just a waste compared to XP). By that time, enough preferences are configured and user programs installed that UAC doesn't up every 5 min (sometimes not at all for days or weeks). But a whole lot of reviews for Vista came after trying it out for a week or two and it's those first two weeks that are the most painful
150 WPM is certainly possible. The world record typist could maintain an average 150 WPM rate for 50 min. She had bursts as high as 212 WPM.
But claiming you can type as fast as the world record typist is like saying you can keep up lap swimming with Michael Phelps.
Newegg was hosting the SD1A firmware. I DL'd it from there (that's also where I bought the drive). However, they don't have a direct link to the firmware anymore and just have the instructions to get the firmware from Seagate.
I did the firmware update and it was pretty quick and painless.
D'Oh, I need a *WAY* better sarcasm filter :-)
HOW MANY TERRORISTS BRING THEIR CHILDREN TO A SUICIDE BOMBING?
Many suicide bombers are minors or women since they are less likely to draw attention. There have been several cases of mothers or parents using diaper bags to hide bombs (it's common in the middle east). One of the "liquid" plots to blow up an airplane that was stopped by Scotland yard was a mother and father who planned on using their baby's milk bottle to disguise the liquid explosive for a bomb.
I know it sounds outrageous to US but the terrorists are playing on a ball field with completely different rules when it comes to respecting the life of innocents.
Ok, sizzle chest. Next time you're on a plane and someone gets up and starts ranting about "sudden depressurization" and "water landings" you should jump on that person before the plane gets high jacked.
That's exactly what they're supposed to do: minimize risk to passengers and the flight. A person like that might not be actually high-jacking or bombing the plane -- they might be crazy and off-meds, overly drunk, or just an extreme asshole. However, they present a unknown yet real (and possibly lethal) risk to other passengers and to the safety of the flight. It's safer for the other 300 passengers on the plane if they are subdued. The acts of jumping around or screaming about horrible accidents (to frighten other passengers) are both illegal on airplanes. If they cause any serious disruption, the first thing that's going to happen is the plane is going to have to land at the nearest airport (which is not your destination airport) and it's at a minimum going to be an awful pain in the ass for the 299 people who behaved appropriately.
And yes, if a group of people were actively planning on how to deal with an upcoming "sudden depressurization" seriously, I'd probably be concerned enough to at least report it to the flight attendants. I'd certainly keep my eye on them for other suspicious behavior.
It doesn't necessarily mean they're guilty of anything but still, if you overhear any group seriously talking amongst themselves in a coordinated way about the impending destruction of the airplane you are on or are about to board, you probably should report them to the proper authorities whether they are white, black, asian, yellow, middle eastern, etc.
Three of the young adults make a remark about where would be the safest place to sit on the plane in the event of an accident or explosion.
Making *ANY* remark about a "possible explosion" on a plane loud enough that several other waiting passengers can hear you in the waiting area is likely to get you kicked off the flight. I'm all against profiling but they were pretty stupid to talk about a subject like that. You don't use the words "explosion" or "bomb" in an airport the same way you don't shout "fire" in a theater.
Microsoft uses an Irish tax haven to keep billions of their dollars out of reach of the American tax man.
There's a US tax law loophole that allow deferring taxes from profits made outside the US as long as the funds from the profits are not transferred into the US. This is one of the same loopholes that makes outsourcing to a foreign branch (i.e. cutting US jobs) tax-free compared to doing the same work in the US :-(
FWIW, I think 1 finger (one hand holding the device, the other using a finger), 2 thumbs (both hands holding using thumbs), or 1 thumb (same thumb as holding hand - i.e. 1-hand input) is better than 2 thumbs + 1 finger (which I can't figure out how that would be useful).
I removed Symantec AV from my computer (since it only protects against exploits nobody uses anymore and slows your PC down more than any virus)
I don't personally use Symantec anything but the word is for the 2009 version, they completely rewrote everything from scratch with an emphasis on speed that seems to have worked according to PCmag.
I'm an Omnikey man myself - been using one for about 20 years now. There is a company that makes an Omnikey clone but with windows keys. And when I say clone, I don't mean a cheap knock-off, the Avant Prime is near perfect down to using the same Alps switches. It's not cheap at $150 but I feel that mine was well worth it -- Keep in mind you will need a $15 PS2-to-USB adapter if you do not have a serial-PS2 keyboard input.
If you like the clicky feel, there is also the DAS Keyboard which is slightly cheaper at $129 and has USB (+ hub). They have a silly "Ultimate" version which has all blank keys to thwart anyone else from using your computer (err I mean to improve your typing skils). A friend of mine bought one (with the letters on it - not the blank keys). However, after he tested both his D.K. and my APrime, he was lusting over my APrime.
It was simple enough for me to understand.
When they refer to "megawatts" in a power plant or storage, they're generally referring to the peak output capability. For example, the 20 Megawatts almost instantaneously would mean the batteries can supply a peak power of 20 Megawatts. The "almost instantaneously" means there is no significant time delay between the request for more power and actually getting the power out (which you might have with a coal fired plant for example).
The "megawatt hours" defines the total storage capability of the battery banks rather than the on-demand peak output.
FWIW it *WILL* cost you more in the long run for that cheaper PSU. If, as you claim, you *NEVER* power down your PC over a 3 year period of use, you will actually save money with a $50 PSU that is 80-90% efficient compared to a $18 PSU that is 60-70% efficient.
Oh, FWIW, the Samsung drives that are fast use SLC. So far, the only shipping MLC drives that are fast come from Intel.
Actually, most of the SLC SSD drives are fairly immune to the random write stall issue that plagues MLC drives. For example, compare SLC and MLC drives from OCZ. The older OCZ Core SSD drives (SLC) have much faster random write access than newer OCZ Core V2 SSD drives (MLS) even though the Core II have much higher specified/published (sequential) write speeds.
OCZ's official line on the frightening performance problems with random writes on MLC drives (i.e. multi-second system stalls and random write throughput as low as 4 writes/second) is "we encourage potential customers to research this product and insure that it will fill their needs. These MLC based drives have extremely fast reads, and if you need a drive with fast sequential (frequent) writes, please check into our SLC based SATA II drive series."
At least OCZ is somewhat honest up front in acknowledging that their MLC drives are not for everyone. But FWIW, nearly all MLC SSD drives are orders of magnitude for real world performance (that includes writes) than their sequential performance specs would suggest.
Currently, the Intel drives are the only shipping MLC drives with good random write performance out-of-the-box. OCZ has announced (but is not yet shipping) a new "Vertex" series SSD that combines MLC with 64MB of RAM cache that speeds up random writes tremendously.
But in general, right now, it's buyer beware if you need fast random write access for higher system performance (i.e. a Windows user). Make sure you get either one of the Intel drives (MLC or SLC) or a well known SLC drive if you're concerned about anything other than strict read performance. Before you buy a MLC drive, follow OCZ's suggestion and do a lot of research on the drive first.
IBM is probably banking on the existence of people who want Cell processors in systems with more than 256megs of RAM. Other IBM value-adds would presumably include rack mountability, support for netbooting and other convenient management stuff, and so forth.
The newest Cell processors from IBM have two major features that customers wanted:
1) Support for normal DDR2 RAM. The original CELL (PS3 version) only supports RAMBUS XDR Memory. There is support for 16GB of DDR vs 256MB of XDR (PS3).
2) Support for pipelined double-precision operations on the SPU's. The original CELL (PS3 version) stalls the SPU's during all double-precision floating point operations and has no support for vectorized double-precision. The pipelining can increase the throughput of common double-precision operations by a factor of 5-8. There is still no support for vectorized double-precision floats.
20% less sodium by volume: Your Mom should try Kosher Salt Flakes. They're like salt flavored snowflakes (which fluffs up the volume) and they tend to stick to the outside of food easily so you get a salty "taste" with less (by mass) salt.
I'm assuming that if the moon was made from a fairly large amounts of ejected matter, that it would have formed from a gradual gravitational aquisition of the mass. Those other moons and planets that form this way have pretty large angular momentum.
Also important to note, the moon is pretty much tidally locked (the same side always faces the earth). It's inconceivable that Earth's gravitational field did not play an extremely key role in the current angular momentum of the moon. For more information, read the section Tidal Coupling in the Earth-Moon System in the linked article.
It also says that Mac OS will be able to optimize for SSD in the future better because OSX is a closed OS. (Which IIRC, the exact opposite is true: Windows is closed source and the lower levels of OSX are open source)
What sort of support do users expect from a $0.99 app. A big point of this article is there is so much shovelware (or as the articles call them "ringtone apps") at the $0.99 price range that users just buy and if it doesn't work, they don't care because $0.99 is cheap.
It's quite possible you will sell more than 5X the number of units at $0.99 than at a $4.99 price point. If that's the case, you will actually make more money selling the app at $0.99 than $4.99.
The big problem is there is no way for an iPhone user to tell which are the quality $0.99 apps and which are the crappy $0.99 apps. If there were a way for users to see which $0.99 apps were the highest quality, they would sell significantly better and increase the payoff for the developers.
FWIW, I had a T-Mobile G1 Android briefly before returning it. There are currently no pay applications in the Android market (or at least none that I could find -- they were all FREE). The girl at the T-Mobile store mentioned it as a plus that all the apps on Android are currently free.
Unfortunately, with the exception of a few ports from iPhone or other cell-phone apps (i.e. Shazam, Pacman, etc.), most of the G1 applications are lagging *FAR* behind their iPhone counterparts.
FWIW, I bought one of these at T-Mobile because the customer rep told me I could return it for a full refund (minus activation fee and plan usage) within 14 days. I returned it after 11 days.
I took a weeklong trip to CA and besides having to charge the phone 2-3 times a days (battery life sucks), the GPS somehow kept telling me I was in Maryland everytime I checked in when I was in San Francisco's Chinatown.
The phone also had a tendency to "drunk dial" people for me. It called one of my friends about 10 times for 1 second calls while it was in my pocket.
The keyboard is kinda clunky and I wish it used the accelerometer to automatically switch to landscape without opening the keytray. I also wish it had a virtual keypad for one-handed text entry (since with the tray open, you *NEED* two hands to hold and type). Finally, it's totally retarded not having at least an on-screen "number" pad for entering phone numbers. When you edit contacts, you have to enter their digits using the tiny little number keys at the top of the keypad. Selecting the phone number to edit in a contact should bring up a virtual dial-pad that uses the large screen in a standard 3x4 touch tone dialing configuration.
Another thing I didn't like was when you are in an area with a lot of WIFI networks (hotels), the available WIFI networks list keeps updating very quickly and doesn't "hold still" to allow you to easily select the one you want. I was trying to scroll down one screen to select one (out of about 15 available) network and before I scrolled and selected, it would refresh. I got it on the 8th or 9th try but it's just another example of clunky inelegance for the UI.
Oh, and I had some drawing glitches and the phone crashed about 4 times in the week that I had it.
I returned it. The G1 is worlds ahead of everything but the iPhone but it's still not currently any more usable than most Google "beta" programs. I'm guessing that in a year or two, when they've done a couple updates it will be more polished.
I would buy an iPhone but they majorly screwed up one important thing on the iPhone -- the volume. I can't hear anything when trying to make a call on the test iPhones at the store and the speaker phone goes to about 2.5 on a volume scale of 10. My buddies who have iPhones swear you can hear them fine if you use the headphones, or you hold the small hole firmly against your ear in exactly the right location (which has about a 1mm tolerance before it becomes unhearable). The G1 Google Phone had very good volume levels.
Right now, I'm just waiting for them to fix all the "beta-ness" of the G1 or double the volume of all the speakers on the iPhone before I pick up a new phone.
Cell CPUs also intrigue me a lot....
That's ironic because I have been programming the Cell processor at my job for the last three years. As a professional video game programmer, I do about half my coding on the PS3 Cell Processor :-)
I do most of the rest of my coding on the XBOX 360 which is also multicore but is a bit easier than the Cell to program.
If you want to learn Cell multiprocessor programming though, you can easily pick up a used PS3 for about $250 on Craigslist and set it up to run Linux. You don't get access to the GPU so hardware-accelerated graphics isn't possible but you can do a fair amount of Cell SPU multicore coding without too large an investment. If you plan on remotely targeting the PS3, I suggest setting up a simple VirtualBox VM on your home PC is you run Windows. Also, this site is a pretty good start for PS3 home brew performance optimizing for the Cell processor.
You won't run into any of the false sharing issues (which is a big deal on XBOX 360 performance) since there's only one PPC core and the SPU's are basically "DMA" driven for access to main RAM.