Unfortunately for you, you don't "tinker" with Cell. Since all the I/O is multi-GHz exotic Rambus signaling, you probably have to be an expert board designer to do anything with it. Not to mention that you have to get the processor, southbridge, and RAM from three different companies, probably signing a stack of NDAs in the process.
The SDK includes both GCC and XLC. GCC's autovectorization isn't the greatest, but Apple and IBM have been working on it. I think if you want fast SPE code you'll end up using intrinsics anyway.
This assumes (a) that the affiliates are not owned by the broadcaster and (b) that the affiliates are in a position of power.
Granted.
What's a better market for an advertiser : All of the viewers in one major market, or all of the users of iTunes?
It depends whether it's a local or national business.
The local, independant affiliate has lost market share in a big, big way over the years. They don't have the sway over the broadcasters that they once had. How many people get their TV off-air ( not via cable or satellite ) these days?
I don't see how this follows. The whole point of "must carry" is that cable and satellite must carry local affiliates, not a network feed. (Except in a few cases.)
Of course a blazing 33Kbps connection is cheaper than a ~1Mbps mesh. You ought to compare mesh against 3G or Starbucks-style 802.11, where every access point has dedicated backhaul; I suspect that mesh is much cheaper in that case.
I think most scientific publishers already sell articles by the article over the Web.
Conference proceedings and journals are a scam anyway. They cost an arm and a leg, and none of the money goes to the researchers or reviewers. Luckily many researchers route around this damage by posting their papers on the Web.
It's cool to have a row of proceedings on my shelf, but I just get the ones from the conferences I actually go to; I'd never buy proceedings separately.
The FBI has a whole Web site about CALEA, including details about cost recovery. It looks like they set aside $500M to cover the cost; I guess the money has all been spent by now, so the universities are left with an unfunded mandate.
That's nice, but are you suggesting that universities should be somehow above the law? CALEA sucks, but it applies to everyone equally. I don't see why universities deserve special treatment.
It probably doesn't work. Like everything else, the RF field is full of snake oil created by people who don't quite understand what they're doing. The stuff always demos well, but it never quite makes it into production.
You mean EFI. But as long as the Darwin bootloader is open source you can just modify it to do whatever you want. (For example, I think MOL uses a modified bootloader so that they don't have to fully emulate Open Firmware. Likewise I think ExPostFacto has a modified bootloader.) If the kernel is still talking to EFI after it's booted, then things could get tricky.
Mac OS X v10.4 Tiger: Developer Overview. "...there is only one version of the kernel for all Apple hardware." (Which must be 32-bit in order to run on older hardware.)
64-Bit Transition Guide. "Because 64-bit applications will be supported using a 32-bit kernel, this 64-bit support will have no impact on most device driver or kernel extension writers."
Maybe this experiment will eventually prove that TPM itself is impossible to achieve when more people are working to break your system than are employeed by Apple to defend it.
TPMs were never intended to be used for what Apple is using them for, thus the cracks only prove that a TPM isn't very useful for things it wasn't designed to do. The real TPM features like sealing and attestation still haven't been cracked.
Be nice to have some real competition versus Linux/OS X in terms of architecture.
It would, but I don't think the desktop/server market can support any new OSes. Cell phones and game consoles are another matter.
Were such a beast (research OS) ever to become a product, would it demonstrate a high level of backwards compatibility?
Singularity is totally incompatible with everything else. I suspect MS would use the Windows Hypervisor to run both Singularity and Windows at the same time if they want compatibility.
I doubt this will ever leave the lab. Singularity will be a test bed for MS researchers who want to play with various concepts.
My understanding is that Singularity supports DLLs, but they must be loaded when a process starts. If you need to load new code while a process is running, just spawn a new process and communicate with it.
Slashdot Mirror
Unfortunately for you, you don't "tinker" with Cell. Since all the I/O is multi-GHz exotic Rambus signaling, you probably have to be an expert board designer to do anything with it. Not to mention that you have to get the processor, southbridge, and RAM from three different companies, probably signing a stack of NDAs in the process.
The SDK includes both GCC and XLC. GCC's autovectorization isn't the greatest, but Apple and IBM have been working on it. I think if you want fast SPE code you'll end up using intrinsics anyway.
This assumes (a) that the affiliates are not owned by the broadcaster and (b) that the affiliates are in a position of power.
Granted.
What's a better market for an advertiser : All of the viewers in one major market, or all of the users of iTunes?
It depends whether it's a local or national business.
The local, independant affiliate has lost market share in a big, big way over the years. They don't have the sway over the broadcasters that they once had. How many people get their TV off-air ( not via cable or satellite ) these days?
I don't see how this follows. The whole point of "must carry" is that cable and satellite must carry local affiliates, not a network feed. (Except in a few cases.)
The affiliates probably wouldn't be happy about either of those options.
"Mesh hardware" is in fact an 802.11 access point with different firmware.
Of course a blazing 33Kbps connection is cheaper than a ~1Mbps mesh. You ought to compare mesh against 3G or Starbucks-style 802.11, where every access point has dedicated backhaul; I suspect that mesh is much cheaper in that case.
Even if someone does mess up their own home directory, they won't be able to touch system files?
Spyware doesn't necessarily need to modify system files to spy on users. The information in your home directory is the most valuable.
So theoretically one could log in as an admin and easily remove the unwanted warez.
Sure, but most Windows users don't even know they have spyware. That problem needs to be solved first.
Maybe FB-DIMM has licensing costs.
Microsoft is a monopoly with massive legacy lock-in. Apple doesn't have that advantage.
Ghz are not directly comparable between different processor architectures...
...and PowerPC has tended to deliver faster chips at the same Mhz as x86.
That's right, but...
Nope; the Opteron and Pentium M are faster per clock than the 970 in most tasks.
No pro user will rely on Rosetta.
:-)
These kids today and their universal binaries; back in the 68K->PPC transition we ran the entire OS under emulation -- and we liked it!
I think most scientific publishers already sell articles by the article over the Web.
Conference proceedings and journals are a scam anyway. They cost an arm and a leg, and none of the money goes to the researchers or reviewers. Luckily many researchers route around this damage by posting their papers on the Web.
It's cool to have a row of proceedings on my shelf, but I just get the ones from the conferences I actually go to; I'd never buy proceedings separately.
The FBI has a whole Web site about CALEA, including details about cost recovery. It looks like they set aside $500M to cover the cost; I guess the money has all been spent by now, so the universities are left with an unfunded mandate.
That's nice, but are you suggesting that universities should be somehow above the law? CALEA sucks, but it applies to everyone equally. I don't see why universities deserve special treatment.
Oops, I meant to say PathScale and Sun. With all the third-party compilers, it's not clear that AMD needs to write their own.
AMD-optimized compilers are available from PathScale and AMD, but AMD processors will run Intel-optimized code pretty well anyway.
It probably doesn't work. Like everything else, the RF field is full of snake oil created by people who don't quite understand what they're doing. The stuff always demos well, but it never quite makes it into production.
You mean EFI. But as long as the Darwin bootloader is open source you can just modify it to do whatever you want. (For example, I think MOL uses a modified bootloader so that they don't have to fully emulate Open Firmware. Likewise I think ExPostFacto has a modified bootloader.) If the kernel is still talking to EFI after it's booted, then things could get tricky.
Mac OS X v10.4 Tiger: Developer Overview. "...there is only one version of the kernel for all Apple hardware." (Which must be 32-bit in order to run on older hardware.)
64-Bit Transition Guide. "Because 64-bit applications will be supported using a 32-bit kernel, this 64-bit support will have no impact on most device driver or kernel extension writers."
You are screwed. Even universities are getting into the "patent everything" game.
Maybe this experiment will eventually prove that TPM itself is impossible to achieve when more people are working to break your system than are employeed by Apple to defend it.
TPMs were never intended to be used for what Apple is using them for, thus the cracks only prove that a TPM isn't very useful for things it wasn't designed to do. The real TPM features like sealing and attestation still haven't been cracked.
OTOH, JavaOS sucked so much that Sun would probably like to erase all evidence that it ever existed.
A SIP is like a Java/.NET sandbox. Because every SIP runs in a sandbox, hardware protection is not needed.
Be nice to have some real competition versus Linux/OS X in terms of architecture.
It would, but I don't think the desktop/server market can support any new OSes. Cell phones and game consoles are another matter.
Were such a beast (research OS) ever to become a product, would it demonstrate a high level of backwards compatibility?
Singularity is totally incompatible with everything else. I suspect MS would use the Windows Hypervisor to run both Singularity and Windows at the same time if they want compatibility.
I doubt this will ever leave the lab. Singularity will be a test bed for MS researchers who want to play with various concepts.
Absolutely. That's the point of research.
My understanding is that Singularity supports DLLs, but they must be loaded when a process starts. If you need to load new code while a process is running, just spawn a new process and communicate with it.