One note up front: I'm guessing based on knowledge of the PowerPC architecture...
> G5 Specific instruction usage sounds suspicious
This one would probably be the instructions that exist in a 64 bit PPC chip but not in a 32 bit version. Things like 64 bit loads/stores and 64 bit shifts/rotates.
> I really like the relaxed IEEE math operations.
The PPC floating point unit has a few instructions that don't follow the IEEE spec. In particular, there are combined multiply/add and multiply/subtract instructions that are much faster than doing the two operations separately. However, they don't round the intermediate results, so they don't always result in the exact same output. Because of this, they normally aren't allowed to be used in any program expecting exact IEEE compliant results.
I'll agree that the malloc thing does look somewhat suspicious. There's was a big flamewar over the performance of the default malloc on OSX on one of Apple's mailing lists not too long ago.
I know people here hate patents, but everyone should note that any modem driver written for this thing is going to infringe on a rather large number of patents. You've never seen such a mess of patents until you've looked at all the compression and error correction routines needed for a full software modem.
It's called subsidies. They were using profits from the games to offset losses in other groups. Because of this, the profits from the PlayStation business were actually larger than the profits for the whole company.
If you read the post you replied to, you'd notice that the last update was June 6 (about three weeks ago). I admit that I don't put regular effort into this project, but please make sure you're correct when you badmouth someone. I do this work for my own enjoyment. If you want it done faster, perhaps you could help out in some way...
I've found that almost everyone is happy to complain, but very few actually do anything about it. If people like you stop making it enjoyable, it won't be worth doing at all.
A better question is what happens to the Compaq Center at San Jose... This is HP's home turf, but Compaq was the only one willing to buy the naming rights to the San Jose Arena.
Well, I feel particularly qualified to answer this one, being the official maintainer of the linux-hfsplus project.:)
Honestly, I don't have an iPod, and I started working on HFS+ support for Linux about a year and a half ago, so unless someone feels like giving me an iPod as an incentive, the pace isn't likely to be affected much.
In theory I could look at the Darwin source, but I wanted to avoid any possible issues just in case some come up later. I've been working entirely from the official Apple documentation (which isn't too bad, but isn't perfect) and from drive images that I've created to try out various things.
My main stumbling block is strictly a lack of time. I finally found the time to get read support fairly stable (take a look at http://sf.net/projects/linux-hfsplus), but I haven't had time to write enough code to actually handle updating the filesystem properly. It's a mess because it's a very non-UNIX filesystem, and there is a lot of manipulation that has to happen to make it act like ext2 or ufs.
There are a ton of differences between HFS and HFS+. The disk block thing is just what was immediately obvious to end users back when HFS+ was introduced, because MacOS 8.1 didn't take advantage of the other features. Small list of new features: unicode filenames, longer filenames, unix style permissions, 64 bit file size, 32 bit block numbers, etc. I'm actually working on HFS+ support for linux, but it's not ready for users yet. (Check out the SourceForge project linux-hfsplus)
The modem is not a hardware modem, and in fact steals approx. 5% or so of the main CPU whenever it is in use. In fact, this is one of the major issues with poor network performance on the DC.
The serial cable in that URL does not attach to the modem port. There is actually a separate serial port on the Dreamcast that has nothing at all to do with the modem.
You're right about the analysis part, but terribly wrong on the timeframe. Anyone who cared from that angle got one when they were released in Japan, and Sony didn't change enough in the US release to make that much of a difference. I was a Sega employee back then, and we had a gigantic meeting just to let everyone in the company get a look at the competition. Trust me, it wouldn't be worth $15000, not even to Sega, just to find out what Sony changed for the US launch.
If you can be patient, there will be HFS+ support "Real Soon Now". There is code, but it is anything but reliable. If you really want to crash constantly, and not read the filesystem properly, you can patch your kernel and try. There was discussion of this on the linuxppc-dev mailing list recently, but noone seems to have time to really fix the bugs. (And I'm the biggest problem there, as I am nominally the one in charge of the hfsplus module) Help is appreciated, of course.
Slashdot Mirror
One note up front: I'm guessing based on knowledge of the PowerPC architecture...
> G5 Specific instruction usage sounds suspicious
This one would probably be the instructions that exist in a 64 bit PPC chip but not in a 32 bit version. Things like 64 bit loads/stores and 64 bit shifts/rotates.
> I really like the relaxed IEEE math operations.
The PPC floating point unit has a few instructions that don't follow the IEEE spec. In particular, there are combined multiply/add and multiply/subtract instructions that are much faster than doing the two operations separately. However, they don't round the intermediate results, so they don't always result in the exact same output. Because of this, they normally aren't allowed to be used in any program expecting exact IEEE compliant results.
I'll agree that the malloc thing does look somewhat suspicious. There's was a big flamewar over the performance of the default malloc on OSX on one of Apple's mailing lists not too long ago.
I know people here hate patents, but everyone should note that any modem driver written for this thing is going to infringe on a rather large number of patents. You've never seen such a mess of patents until you've looked at all the compression and error correction routines needed for a full software modem.
It's called subsidies. They were using profits from the games to offset losses in other groups. Because of this, the profits from the PlayStation business were actually larger than the profits for the whole company.
If you read the post you replied to, you'd notice that the last update was June 6 (about three weeks ago). I admit that I don't put regular effort into this project, but please make sure you're correct when you badmouth someone. I do this work for my own enjoyment. If you want it done faster, perhaps you could help out in some way...
I've found that almost everyone is happy to complain, but very few actually do anything about it. If people like you stop making it enjoyable, it won't be worth doing at all.
A better question is what happens to the Compaq Center at San Jose... This is HP's home turf, but Compaq was the only one willing to buy the naming rights to the San Jose Arena.
Well, I feel particularly qualified to answer this one, being the official maintainer of the linux-hfsplus project. :)
Honestly, I don't have an iPod, and I started working on HFS+ support for Linux about a year and a half ago, so unless someone feels like giving me an iPod as an incentive, the pace isn't likely to be affected much.
In theory I could look at the Darwin source, but I wanted to avoid any possible issues just in case some come up later. I've been working entirely from the official Apple documentation (which isn't too bad, but isn't perfect) and from drive images that I've created to try out various things.
My main stumbling block is strictly a lack of time. I finally found the time to get read support fairly stable (take a look at http://sf.net/projects/linux-hfsplus), but I haven't had time to write enough code to actually handle updating the filesystem properly. It's a mess because it's a very non-UNIX filesystem, and there is a lot of manipulation that has to happen to make it act like ext2 or ufs.
There are a ton of differences between HFS and HFS+. The disk block thing is just what was immediately obvious to end users back when HFS+ was introduced, because MacOS 8.1 didn't take advantage of the other features. Small list of new features: unicode filenames, longer filenames, unix style permissions, 64 bit file size, 32 bit block numbers, etc. I'm actually working on HFS+ support for linux, but it's not ready for users yet. (Check out the SourceForge project linux-hfsplus)
The modem is not a hardware modem, and in fact steals approx. 5% or so of the main CPU whenever it is in use. In fact, this is one of the major issues with poor network performance on the DC.
The serial cable in that URL does not attach to the modem port. There is actually a separate serial port on the Dreamcast that has nothing at all to do with the modem.
You're right about the analysis part, but terribly wrong on the timeframe. Anyone who cared from that angle got one when they were released in Japan, and Sony didn't change enough in the US release to make that much of a difference. I was a Sega employee back then, and we had a gigantic meeting just to let everyone in the company get a look at the competition. Trust me, it wouldn't be worth $15000, not even to Sega, just to find out what Sony changed for the US launch.
If you can be patient, there will be HFS+ support "Real Soon Now". There is code, but it is anything but reliable. If you really want to crash constantly, and not read the filesystem properly, you can patch your kernel and try. There was discussion of this on the linuxppc-dev mailing list recently, but noone seems to have time to really fix the bugs. (And I'm the biggest problem there, as I am nominally the one in charge of the hfsplus module) Help is appreciated, of course.