I read the article. My point is if he's trying to ACTUALLY determine whether the XBox is a viable platform for SOMEONE to buy it and use for clustering, it's pretty obvious what the result is going to be ahead of time.
ALSO: What "experimental results"? A single benchmark with very little accuracy or explanation that you can point at and say "I could do that cheaper with 3 PC's" before you even do the benchmark?
I can only assume he gave more results to the people who actually gave him the check, but I don't know.
I read the whole article. My point was why even do the experiment. It's pretty easy to tell you could spend less cash and get more raw processing power from a cheap CPU/MB combo.
This of course is not taking into consideration the video card's power (like others have pointed out) if you're trying to do some sort of vector processing that you can do faster on the GPU, or some sort of rendering. Why didn't he try to do something in that area and maybe make this a worthwhile experiment.
In summary, the whole experiment seemed sort of pointless when you can easily tell which way the experiment is going to go ahead of time. He does one benchmark and gives very little in the way of results.
Like I said, it might be a fun little project and all, but this seems sort of pointless from the viewpoint of useful information. It was a fun article to read, but you walk away from it saying to yourself "yep, figured that...".
Re:Quick Summary... and a Why?
on
Xbox Linux Cluster
·
· Score: 3, Interesting
I'm a bit confused as well...
Why would you want to buy an XBox for this in general? It's somewhat difficult to determine the cost factors here, but I would think you could buy just as powerful of a MB/CPU/RAM combo (actually, you could get a better CPU easily) cheaper. Heck, it's hard to even find people still selling 733 Celerons.
The only thing that might blow the budget is the video card you would need (it's a specialized G2, right?), and if you're clustering them, what's the point of having a good video card?
Seems like a fun little project, but the article seemed pretty light on actual performance data, etc.
The problem is these would be banner ads that no human ever saw, and they'd know that advertising on this service would be worthless (as no human eyes would ever see it).
What people really need for this service is more like a Web service that some daemon goes out and hits and grabs the info automatically. The problem again is nobody has a motivation to spend the money on bandwidth, maintenance, etc. and get zero money in return;).
The problem with this is it pulls from 3rd party sites (mainly tvguide.com and gist.com I believe) to get its programming information.
If LOTS of people started using this on their homemade PVR's, I'm guessing the programming information from tvguide.com and gist.com would suddenly change format to make this break:).
Although he seemed happy with 15 FPS, that seems a bit anemic to me. Has anyone pulled this off with 30 before? A lot of the Linux PVR implementations I've seen don't really mention what rate they achieve.
I'm just wondering how viable the performance would be if I went out and bought a capture card and mini-ITX MB etc. before I try something like this.
I've seen projects that rip the data out of various sources, but I've yet to see an open source one.
I guess nobody currently has the motivation to provide bandwidth and storage for free essentially and do all the footwork it would take to get the info from the networks.
Even at 50%, you're still talking about 5MB per minute, which is too big for "mainstream" users. God help them if they wanted to download that 2112 track or something;).
Of course I'm in the same boat as you. I downloaded 3 linux distros last week...
I would have to assume it's a lossy compression. It sounds like it was a WMA file.
Someone correct me if I'm wrong, but I don't think a lossless compression exists that will make the file size small enough for "most" people to download. That is to say make the audio about one tenth the size of the raw audio.
Slashdot Mirror
That's a good point. Maybe someone should do some tests in that area.
Perhaps using XBox's as a rendering farm would get you a lot more bang for the buck...
I read the article. My point is if he's trying to ACTUALLY determine whether the XBox is a viable platform for SOMEONE to buy it and use for clustering, it's pretty obvious what the result is going to be ahead of time.
ALSO: What "experimental results"? A single benchmark with very little accuracy or explanation that you can point at and say "I could do that cheaper with 3 PC's" before you even do the benchmark?
I can only assume he gave more results to the people who actually gave him the check, but I don't know.
I read the whole article. My point was why even do the experiment. It's pretty easy to tell you could spend less cash and get more raw processing power from a cheap CPU/MB combo.
This of course is not taking into consideration the video card's power (like others have pointed out) if you're trying to do some sort of vector processing that you can do faster on the GPU, or some sort of rendering. Why didn't he try to do something in that area and maybe make this a worthwhile experiment.
In summary, the whole experiment seemed sort of pointless when you can easily tell which way the experiment is going to go ahead of time. He does one benchmark and gives very little in the way of results.
Like I said, it might be a fun little project and all, but this seems sort of pointless from the viewpoint of useful information. It was a fun article to read, but you walk away from it saying to yourself "yep, figured that...".
I'm a bit confused as well...
Why would you want to buy an XBox for this in general? It's somewhat difficult to determine the cost factors here, but I would think you could buy just as powerful of a MB/CPU/RAM combo (actually, you could get a better CPU easily) cheaper. Heck, it's hard to even find people still selling 733 Celerons.
The only thing that might blow the budget is the video card you would need (it's a specialized G2, right?), and if you're clustering them, what's the point of having a good video card?
Seems like a fun little project, but the article seemed pretty light on actual performance data, etc.
"tried out his new submersible sea plane"
Not quite...it might feel like a plane in the water, but it's not a submersible sea plane.
The problem is these would be banner ads that no human ever saw, and they'd know that advertising on this service would be worthless (as no human eyes would ever see it).
;).
What people really need for this service is more like a Web service that some daemon goes out and hits and grabs the info automatically. The problem again is nobody has a motivation to spend the money on bandwidth, maintenance, etc. and get zero money in return
The problem with this is it pulls from 3rd party sites (mainly tvguide.com and gist.com I believe) to get its programming information.
:).
If LOTS of people started using this on their homemade PVR's, I'm guessing the programming information from tvguide.com and gist.com would suddenly change format to make this break
Although he seemed happy with 15 FPS, that seems a bit anemic to me. Has anyone pulled this off with 30 before? A lot of the Linux PVR implementations I've seen don't really mention what rate they achieve.
I'm just wondering how viable the performance would be if I went out and bought a capture card and mini-ITX MB etc. before I try something like this.
I've seen projects that rip the data out of various sources, but I've yet to see an open source one.
I guess nobody currently has the motivation to provide bandwidth and storage for free essentially and do all the footwork it would take to get the info from the networks.
I would guess it's not a cross browser type thing as much it's a they want to use ActiveX or encrypted javascript type thing.
Even at 50%, you're still talking about 5MB per minute, which is too big for "mainstream" users. God help them if they wanted to download that 2112 track or something ;).
Of course I'm in the same boat as you. I downloaded 3 linux distros last week...
I'm curious why it only works on IE.
Did they install any ActiveX controls on the page?
Or did they use encrypted javascript?
As far as I know, that would be the only REAL reasons they'd be that strict...
I would have to assume it's a lossy compression. It sounds like it was a WMA file.
Someone correct me if I'm wrong, but I don't think a lossless compression exists that will make the file size small enough for "most" people to download. That is to say make the audio about one tenth the size of the raw audio.