While we're on the topic of AppleTV hacks, I had an alternate theory on why the AppleTV doesn't include a DVD drive. As has been noted, dropping a DVD drive into would have added a pretty neglible additional cost to Apple. They already have a DVD player program so the work to actually add DVD playback would have been minimal.
My theory is that it was not implemented to lock out usage of a DVD player and try to steer people to purchase movies from iTunes. It also makes hacking the AppleTV to play DVD image files that much harder with nothing to start from. I know there are free ways to created DVD images, but a program like Flip4Mac's Drive-In makes the process extremely easy and dummy-proof. The idea of ripping the full DVD content to a central storage location is far more attractive to me than using HandBrake to rip just the movie since I lose nothing in the process (quality, extras, audio tracks, subtitles, etc.) although it would require a massive storage source.
Without DVD playback, the AppleTV locks out any such attempt.
My theory is that it was not implemented to lock out usage of a DVD player and try to steer people to purchase movies from iTunes.
You are falling for the very common incorrect assumption that iTunes Store sales are significant for Apple. They never have been; neither in profits nor in revenues. What has been significant is sales of hardware devices: iPods, Macs, and possibly now Apple TVs.
You are falling for the very common incorrect assumption that iTunes Store sales are significant for Apple. They never have been; neither in profits nor in revenues. What has been significant is sales of hardware devices: iPods, Macs, and possibly now Apple TVs.
It depends on what you think is small profit, I guess. The artist or label is getting about 42% of each purchase (which is 2% better than standard record store sales), which gives Apple about 58 cents per song. That adds up when you consider apple has sold over 750 million songs now... Of course, they've shelled out a lot of money developing iTunes, and in advertising, but their insanely low overhead on staff, warehouse space, transportation, etc. etc. etc. is vastly lower than your typical record chain. They also don't pay the licensing fee to the good people at Philips who own the patent rights to CDs and DVDs (?), which saves them another 1% against record stores...
It's been something of a mantra that Apple doesn't make significant money from iTunes store sales, but that's a line to develop buzz that it's a great bargain since they're not making so much. The truth is, like any investment, that after the initial amount of money spent investing in the whole system, it's starting to pay off now, and their profit margins have the potential to surpass any traditional record stores, even though they pay a higher percentage to the record company/independent artist...
Before you jump off a bridge, you better check to see how deep the water is.
Feh. We used to jump off the county bridges into whitewater below back home. This ain't nuthin.
Quote:
For your information, I used the exact quote from Mossberg in my original comment.
Then I fail to see what he was accused of saying that he didn't. You may disagree about how others interpreted the intent of his words, or the way that they were presented, but he did write those words. But really, that's immaterial at this point, since...
Quote:
My point was simply this, 'you may question what he said, but you needn't call him or imply that he's an frigg'n liar (my interpretation*) because he didn't include any supportive references.'
And that's the crux of the issue - I think your interpretation of 'talking out of his ass' is at odds with the rest of us. If they had said "He's lying" or even "He's just plain wrong", without information to back *themselves* up, I'd be giving them the hairy eyeball too. Instead, I think the rest of us were reading the phrase as most people would - as "where's his evidence? this is a meaningless assertion without it". Which it was. Right, wrong, doesn't matter. It's all about how it's presented. Supporting evidence = good, groundless assertions (even when correct) = lame. That's all this was, as I see it. A minor difference in interpretation.
I'd also point out that there is a *GULF* of difference between a professional environment with colleagues, and an online rumor board, as far as what is considered appropriate or acceptable behavior. I frequently disagree with my research colleagues, but I argue with them in a different way than I do on here. Language on here is a more colloquial, and vague. Ya gotta roll with the punches around here, and really, don't take it personally.
In other words, you assume that, while Apple doesn't use the GMA 950's decoding acceleration, and doesn't use the X1600's decoding acceleration either, or that of any other recent GPU in an Intel Mac, for that matter, they do use the one in the Go7300, even though ?*as evidenced by the successful Perian installation, the build numbers, etc. ?*the QuickTime decoding is virtually the same?
This seems to be wanting from me something slightly different than your previous post. This one seems to want proof that Apple is using the GeForce 7300 hardware features.
Here is the proof:
I have a MacBook with 1.83 GHz Core Duo
I downloaded a 1920 x 816 trailer from Apple
I re-compressed this down to 1280 x 720/24p, 5 Mbit/s H.264 (the maximum allowed by AppleTV) with 160 kbps stereo AAC
I downloaded Apple's CHUD tools
I used the CHUD tools to disable one of my Core Duo cores
I tried to play back the file.
It dropped a lot of frames (it went as low as 14 fps at some points).
If the standard QuickTime H.264 codec running on a 1.83 GHz Core Duo core cannot decode 1280 x 720/24p 5 Mbit/s H.264 in real-time, there's no way in hell that it can do it on AppeTV's underclocked Pentium-M. Not only is the CPU running at only just over 1/2 the clock speed of my CPU, it also has a slightly inferior SSE1/SSE2 implementation, and no SSE3 at all. For the AppleTV to be able to play back the H.264 files that it plays back, Apple must either be using the GeForce hardware decode features, or some kind of crazy compsci voodoo magic.
You are falling for the very common incorrect assumption that iTunes Store sales are significant for Apple. They never have been; neither in profits nor in revenues. What has been significant is sales of hardware devices: iPods, Macs, and possibly now Apple TVs.
Whilst I don't see the point of adding a DVD drive to an AppleTV (it's like adding a CD player to an iPod, and surely folks who are going to buy an AppleTV have a DVD player already). Surely the long-term aim of the AppleTV should be to replace the DVD player just like the iPod has replaced the walkman, and in many homes, the CD player too.
I've been trying to get my head around why Apple haven't done some sort of flip4mac drive-in thing with iTunes + AppleTV, and I don't get it. Maybe it does have something to do with the iTunes store. Whilst it's generally accepted that Apple make pretty much squat on music downloads, perhaps there is the potential there for them to actually make significant profits on the movie downloads?
Having said that, it still doesn't make sense to me. Surely the AppleTV would be more compelling if it (in conjunction with iTunes) offered an easy way to rip (unmodified a-la drive-in) all your DVDs you already own onto a server and then stream those files to AppleTV? Then Apple would sell more AppleTVs. Then, if Apple offered 1280 x 720/24p movies with 5.1 AC3 instead of the 640 x 360 ones with stereo that they do now, priced inbetween DVD and HD-DVD/blu-ray, there would be a compelling reason to buy any new content from the iTunes store rather than elsewhere (higher quality than DVD, but cheaper than HD-DVD/blu-ray).
Hey, would someone with an intel mac please email me the sshd binary please (it is in your /usr/sbin/ folder)? I have a PowerPC mac, and therefore don't have access to it. OpenSSH is open source, so distributing the binary is ok (ie, not warez or hacking). If you still don't feel comfortable distributing the binary from your OS X install, then please goto Openssh.com click on one of the mirror links. Then download openssh-4.6p1.tar.gz.
Then do the following in the terminal:
cd to the un-ziped folder
./configure --without-openssl-header-check
make
Then there should be a intel-os-x binary in that folder called sshd. Please email that to me. Thanks!
On the Mac Mini, all of this is done in software running on the Core2Duo, on the AppleTV all this is done in hardware on the GeForce Go 7300 (for H.264 content). The hardware implementations in the GeForce Go chip of each of these stages are superior, quality-wise, to the algorithms used on the Mac Mini when playing back the same files. Therefore AppleTV has higher picture quality
This seems to be the crux of your point. Which I do understand and should have been said in plain language earlier.
You also talk about the mini using the CPU to decode video when we just went through the whole thing where I said the mini is using the CPU for video in response to you complaining about the lackluster GMA950, you asserted I had no idea what I was talking about.
Quote:
My opinion is based on the fact that Apple have repeatedly implemented things poorly in QuickTime (both their MPEG-4 part 2 and part 10 (H.264) codecs suck relative to others in the industry*), but Nvidia on the other hand really know what they are doing. I am confident that the decoding, de-interlacing and scaling are better implemented by Nvidia than by Apple.
And now you admit much of this is your opinion and nothing you have empirical proof of. This sounds like technical nitpicking to me. In the long run when Joe and Jane Consumer are watching video on their computer will they care if the video is being decoded on superior Nvidia GPU hardware or on less optimal Quicktime in the CPU? Nope.
Whilst I don't see the point of adding a DVD drive to an AppleTV (it's like adding a CD player to an iPod, and surely folks who are going to buy an AppleTV have a DVD player already). Surely the long-term aim of the AppleTV should be to replace the DVD player just like the iPod has replaced the walkman, and in many homes, the CD player too.
I've been trying to get my head around why Apple haven't done some sort of flip4mac drive-in thing with iTunes + AppleTV, and I don't get it. Maybe it does have something to do with the iTunes store. Whilst it's generally accepted that Apple make pretty much squat on music downloads, perhaps there is the potential there for them to actually make significant profits on the movie downloads?
Having said that, it still doesn't make sense to me. Surely the AppleTV would be more compelling if it (in conjunction with iTunes) offered an easy way to rip (unmodified a-la drive-in) all your DVDs you already own onto a server and then stream those files to AppleTV? Then Apple would sell more AppleTVs. Then, if Apple offered 1280 x 720/24p movies with 5.1 AC3 instead of the 640 x 360 ones with stereo that they do now, priced inbetween DVD and HD-DVD/blu-ray, there would be a compelling reason to buy any new content from the iTunes store rather than elsewhere (higher quality than DVD, but cheaper than HD-DVD/blu-ray).
The problem is that they are trying to attract the very studio's that sold you those DVDs to sell their content on iTMS and there is a lot more resistance to this with video than there was with audio, plus a lot of competition beginning to crop up. Apple needs the content, so they need to get the studios and broadcasters to let them distribute it, and do so at a price that Apple feels it's customers will buy. Not $20/movie when you are not getting the physical media or the extras, but $14.99 for a new release. Apple needs that price low to attract customers to iTMS and add the value to Apple TV that will make it a success.
If the standard QuickTime H.264 codec running on a 1.83 GHz Core Duo core cannot decode 1280 x 720/24p 5 Mbit/s H.264 in real-time, there's no way in hell that it can do it on AppeTV's underclocked Pentium-M. Not only is the CPU running at only just over 1/2 the clock speed of my CPU, it also has a slightly inferior SSE1/SSE2 implementation, and no SSE3 at all.
You do realize that the OS X in the Apple TV has much, much less to do than the Mac OS X in your Mac mini? No. That would be too easy to consider. I'm sure you receive e-mails on your Apple TV all the time. Or compile ports in DarwinPorts. Yep. Gotta keep the machine busy. While watching 720p video.
The H.264 component on the Apple TV is indeed different. But your conclusions are still quite a stretch.
And now you admit much of this is your opinion and nothing you have empirical proof of. This sounds like technical nitpicking to me. In the long run when Joe and Jane Consumer are watching video on their computer will they care if the video is being decoded on superior Nvidia GPU hardware or on less optimal Quicktime in the CPU? Nope.
Some users will care if it results in poorer picture quality. I freely admit that I shouldn't have been nearly as unequivocal as I was in my first post to this thread. Hopefully I have demonstrated that there are plenty of reasons why the picture quality of a Mac Mini and AppleTV will not necessarily be equal. I'm really trying to point out that people shouldn't assume that the picture quality of all devices are equal. There are many factors that affect it.
When it comes to Joe and Jane Consumer, they have demonstrated often that they don't care all that much about quality, given that they buy 128 kbps audio downloads in huge quantities, and purchase some iTunes Movies. So for them I'm sure this debate doesn't matter.
You do realize that the OS X in the Apple TV has much, much less to do than the Mac OS X in your Mac mini? No. That would be too easy to consider. I'm sure you receive e-mails on your Apple TV all the time. Or compile ports in DarwinPorts. Yep. Gotta keep the machine busy. While watching 720p video.
The H.264 component on the Apple TV is indeed different. But your conclusions are still quite a stretch.
The AppleTV is running OS X 8N5107, so actually it bears a very striking resemblance to the OS on my MacBook (10.4.8 Intel was 8L2127, 10.4.9 Intel is 8P2137). Sure, there are lots of extra bits and pieces installed in my OS, but extra frameworks and the like take up no extra CPU resources just by being there. You have to be using them for that to happen. I should have said in my post that QuickTime and Finder where the only applications running when I did my test. There was nothing sucking up CPU cycles in the manner that you suggest.
Try this:
Quit all apps apart from Activity monitor, QuickTime and Finder.
On my MacBook, Activity monitor and pmTool (a utility that activity monitor uses) together take up % cpu, and various other components such as Window Server and the kernel take up a total of <1.5% CPU between them. This means that without Activity monitor or pmTool running, at least 98.5% of the CPU cycles are available.
Now, open up a 1280 x 720/24p 5 Mbit/s H.264 file in QuickTime and play it back. On my MacBook (with one core disabled), the WindowServer, kernel and QuickTime CPU usages all increase. The usages vary quite a bit, the peak seems to be around 4% for WindowServer and 3% for the kernel. So, without Activity Monitor running, there would be at least 93% processor power available to QuickTime. It is without Activity Monitor running that I performed my test that demonstrated a heavy rate of dropped frames. And this on a Core Duo core with improved (over the Pentium-M in the AppleTV) SSE1/2 and SSE3 (which the Pentium-M doesn't have at all) running at almost twice the clock speed.
How exactly do you propose that the OS on the AppleTV is structured? Even if you drive the WindowServer and kernel CPU usages down to zero, there still isn't anywhere near enough CPU power available for QuickTime to decode a 1280 x 720/24p 5 Mbit/s H.264 stream in real-time. The AppleTV must be using the GeForce hardware-decode features.
If you think the AppleTV is doing it all on the CPU, why do you think Apple put a GeForce in there? Wouldn't it have been much cheaper to use a GMA950? And if Apple in the process of developing the AppleTV worked out how to reduce the CPU usage of their QuickTime H.264 decoder by a factor of something like 4, why didn't they deliver that improved codec with the release of QuickTime 7.1.5?
The AppleTV is running OS X 8N5107, so actually it bears a very striking resemblance to the OS on my MacBook (10.4.8 Intel was 8L2127, 10.4.9 Intel is 8P2137).
I'm quite aware.
Though, technically, your build numbers are misleading. It's running 10.4.7 with Darwin 10.4.8 and a build between 10.4.8 and 10.4.9.
Quote:
Sure, there are lots of extra bits and pieces installed in my OS, but extra frameworks and the like take up no extra CPU resources just by being there. You have to be using them for that to happen. I should have said in my post that QuickTime and Finder where the only applications running when I did my test. There was nothing sucking up CPU cycles in the manner that you suggest.
Aside from dozens of agents and daemons, not to mention many additional kernel extensions.
Quote:
If you think the AppleTV is doing it all on the CPU,
I don't. I do think that you need to stop jumping to conclusions, as you have just done yet again.
Seems like Mr. H is just detailing the reasoning for what he believes to be the most likely scenario. This isn't jumping to conclusions in my book. If anything, his thorough reasoning is the exact opposite of jumping to conclusions.
Other than not speculating at all, there isn't a more reasonable avenue for discussion. But then again, not speculating would mean no discussion either.
Aside from dozens of agents and daemons, not to mention many additional kernel extensions.
You must have missed the bit where I detailed the CPU usage of 10.4.9 running on my MacBook with one core disabled. Yes, there are lots of agents and daemons resident in memory, no, they weren't using any measurable CPU time during my tests. Quit all your GUI apps, launch Activity monitor and see for yourself what the CPU usage at idle is like.
Quote:
Originally Posted by Chucker
I don't [think the AppleTV is doing it [decoding H.264] all on the CPU]. I do think that you need to stop jumping to conclusions, as you have just done yet again.
My statement was that there was "no way in hell it can do it [decode H.264] on AppeTV's underclocked Pentium-M." You then quoted this statement, and called it "quite a stretch". I'm sorry, but you are just contradicting yourself. If the AppleTV can't use its CPU to decode the H.264, where else is it going to do it?
this is apples big chance to again achieve cult status or blow it big time. should jobs decide to plug all the holes and make hacking this machine impossible or really hard, it will be like shooting yourself in the foot. if, on the other hand, they will allow the chip to fall where they may, this could be the start of a fantastic machine-running all codecs, using it as a ultra mini mac, whatever! let;s see what happens...
Word! It could be like Xbox Media Center -- except that, rather than modding your *existing* Xbox, folks'd *purchase* this device in order to mod it. With a larger HD and wider codec support, I'd be all over this! It seems like a device with a high Wifey Approval Factor. As it is, the Mac Mini in our bedroom pisses off my wife (and me) because FrontRow likes take a long time to scroll our Movies. The X360 MCE in our living room is great for streaming music -- but suck-ass video codec support (who wants to convert all their avi to wmv -- and don't tell me to transcode, that's a 0 on the WAF scale).
Anywho, all that to say this. If I can get unlimited video codec support on a fast Apple interface and not have my wife roll her eyes at me as I explain how to use it, I'll put one of these boxes on each of 3 different TVs around my house. How 'bout *that*, Mr Jobs?
Seems like Mr. H is just detailing the reasoning for what he believes to be the most likely scenario. This isn't jumping to conclusions in my book. If anything, his thorough reasoning is the exact opposite of jumping to conclusions.
Other than not speculating at all, there isn't a more reasonable avenue for discussion. But then again, not speculating would mean no discussion either.
Ditto that.
And I'd like to add, on balance, I think Mr. H is contributing useful technical arguments, hardware tests, and his thoughts to this thread. Many of the other contributors seem more concerned about his sentence anatomy rather than presenting their own detailed analysis for peer review.
Mr. H earnestly stuck out his neck to present even-toned thoughts and analysis. And naturally, they ought to be open for discussion. But why not present an equally long neck in response? Instead, it seems to me that on balance the collective retort to Mr. H is built on unconstructive critique (in a few instances, inflammatory critique) and small details of sentence nuance.
As we all know, snipper fire is easy work in anonymous fora.
Comments
My theory is that it was not implemented to lock out usage of a DVD player and try to steer people to purchase movies from iTunes. It also makes hacking the AppleTV to play DVD image files that much harder with nothing to start from. I know there are free ways to created DVD images, but a program like Flip4Mac's Drive-In makes the process extremely easy and dummy-proof. The idea of ripping the full DVD content to a central storage location is far more attractive to me than using HandBrake to rip just the movie since I lose nothing in the process (quality, extras, audio tracks, subtitles, etc.) although it would require a massive storage source.
Without DVD playback, the AppleTV locks out any such attempt.
My theory is that it was not implemented to lock out usage of a DVD player and try to steer people to purchase movies from iTunes.
You are falling for the very common incorrect assumption that iTunes Store sales are significant for Apple. They never have been; neither in profits nor in revenues. What has been significant is sales of hardware devices: iPods, Macs, and possibly now Apple TVs.
You are falling for the very common incorrect assumption that iTunes Store sales are significant for Apple. They never have been; neither in profits nor in revenues. What has been significant is sales of hardware devices: iPods, Macs, and possibly now Apple TVs.
It depends on what you think is small profit, I guess. The artist or label is getting about 42% of each purchase (which is 2% better than standard record store sales), which gives Apple about 58 cents per song. That adds up when you consider apple has sold over 750 million songs now... Of course, they've shelled out a lot of money developing iTunes, and in advertising, but their insanely low overhead on staff, warehouse space, transportation, etc. etc. etc. is vastly lower than your typical record chain. They also don't pay the licensing fee to the good people at Philips who own the patent rights to CDs and DVDs (?), which saves them another 1% against record stores...
It's been something of a mantra that Apple doesn't make significant money from iTunes store sales, but that's a line to develop buzz that it's a great bargain since they're not making so much. The truth is, like any investment, that after the initial amount of money spent investing in the whole system, it's starting to pay off now, and their profit margins have the potential to surpass any traditional record stores, even though they pay a higher percentage to the record company/independent artist...
Before you jump off a bridge, you better check to see how deep the water is.
Feh. We used to jump off the county bridges into whitewater below back home. This ain't nuthin.
For your information, I used the exact quote from Mossberg in my original comment.
Then I fail to see what he was accused of saying that he didn't. You may disagree about how others interpreted the intent of his words, or the way that they were presented, but he did write those words. But really, that's immaterial at this point, since...
My point was simply this, 'you may question what he said, but you needn't call him or imply that he's an frigg'n liar (my interpretation*) because he didn't include any supportive references.'
And that's the crux of the issue - I think your interpretation of 'talking out of his ass' is at odds with the rest of us. If they had said "He's lying" or even "He's just plain wrong", without information to back *themselves* up, I'd be giving them the hairy eyeball too. Instead, I think the rest of us were reading the phrase as most people would - as "where's his evidence? this is a meaningless assertion without it". Which it was. Right, wrong, doesn't matter. It's all about how it's presented. Supporting evidence = good, groundless assertions (even when correct) = lame. That's all this was, as I see it. A minor difference in interpretation.
I'd also point out that there is a *GULF* of difference between a professional environment with colleagues, and an online rumor board, as far as what is considered appropriate or acceptable behavior. I frequently disagree with my research colleagues, but I argue with them in a different way than I do on here. Language on here is a more colloquial, and vague. Ya gotta roll with the punches around here, and really, don't take it personally.
The artist or label is getting about 42% of each purchase
70.
which gives Apple about 58 cents per song.
After bandwidth, handling, etc., about 4 cents remain with Apple.
In other words, you assume that, while Apple doesn't use the GMA 950's decoding acceleration, and doesn't use the X1600's decoding acceleration either, or that of any other recent GPU in an Intel Mac, for that matter, they do use the one in the Go7300, even though ?*as evidenced by the successful Perian installation, the build numbers, etc. ?*the QuickTime decoding is virtually the same?
Here is the proof:
I have a MacBook with 1.83 GHz Core Duo
I downloaded a 1920 x 816 trailer from Apple
I re-compressed this down to 1280 x 720/24p, 5 Mbit/s H.264 (the maximum allowed by AppleTV) with 160 kbps stereo AAC
I downloaded Apple's CHUD tools
I used the CHUD tools to disable one of my Core Duo cores
I tried to play back the file.
It dropped a lot of frames (it went as low as 14 fps at some points).
If the standard QuickTime H.264 codec running on a 1.83 GHz Core Duo core cannot decode 1280 x 720/24p 5 Mbit/s H.264 in real-time, there's no way in hell that it can do it on AppeTV's underclocked Pentium-M. Not only is the CPU running at only just over 1/2 the clock speed of my CPU, it also has a slightly inferior SSE1/SSE2 implementation, and no SSE3 at all. For the AppleTV to be able to play back the H.264 files that it plays back, Apple must either be using the GeForce hardware decode features, or some kind of crazy compsci voodoo magic.
You are falling for the very common incorrect assumption that iTunes Store sales are significant for Apple. They never have been; neither in profits nor in revenues. What has been significant is sales of hardware devices: iPods, Macs, and possibly now Apple TVs.
Whilst I don't see the point of adding a DVD drive to an AppleTV (it's like adding a CD player to an iPod, and surely folks who are going to buy an AppleTV have a DVD player already). Surely the long-term aim of the AppleTV should be to replace the DVD player just like the iPod has replaced the walkman, and in many homes, the CD player too.
I've been trying to get my head around why Apple haven't done some sort of flip4mac drive-in thing with iTunes + AppleTV, and I don't get it. Maybe it does have something to do with the iTunes store. Whilst it's generally accepted that Apple make pretty much squat on music downloads, perhaps there is the potential there for them to actually make significant profits on the movie downloads?
Having said that, it still doesn't make sense to me. Surely the AppleTV would be more compelling if it (in conjunction with iTunes) offered an easy way to rip (unmodified a-la drive-in) all your DVDs you already own onto a server and then stream those files to AppleTV? Then Apple would sell more AppleTVs. Then, if Apple offered 1280 x 720/24p movies with 5.1 AC3 instead of the 640 x 360 ones with stereo that they do now, priced inbetween DVD and HD-DVD/blu-ray, there would be a compelling reason to buy any new content from the iTunes store rather than elsewhere (higher quality than DVD, but cheaper than HD-DVD/blu-ray).
Then do the following in the terminal:
cd to the un-ziped folder
./configure --without-openssl-header-check
make
Then there should be a intel-os-x binary in that folder called sshd. Please email that to me. Thanks!
On the Mac Mini, all of this is done in software running on the Core2Duo, on the AppleTV all this is done in hardware on the GeForce Go 7300 (for H.264 content). The hardware implementations in the GeForce Go chip of each of these stages are superior, quality-wise, to the algorithms used on the Mac Mini when playing back the same files. Therefore AppleTV has higher picture quality
This seems to be the crux of your point. Which I do understand and should have been said in plain language earlier.
You also talk about the mini using the CPU to decode video when we just went through the whole thing where I said the mini is using the CPU for video in response to you complaining about the lackluster GMA950, you asserted I had no idea what I was talking about.
My opinion is based on the fact that Apple have repeatedly implemented things poorly in QuickTime (both their MPEG-4 part 2 and part 10 (H.264) codecs suck relative to others in the industry*), but Nvidia on the other hand really know what they are doing. I am confident that the decoding, de-interlacing and scaling are better implemented by Nvidia than by Apple.
And now you admit much of this is your opinion and nothing you have empirical proof of. This sounds like technical nitpicking to me. In the long run when Joe and Jane Consumer are watching video on their computer will they care if the video is being decoded on superior Nvidia GPU hardware or on less optimal Quicktime in the CPU? Nope.
Whilst I don't see the point of adding a DVD drive to an AppleTV (it's like adding a CD player to an iPod, and surely folks who are going to buy an AppleTV have a DVD player already). Surely the long-term aim of the AppleTV should be to replace the DVD player just like the iPod has replaced the walkman, and in many homes, the CD player too.
I've been trying to get my head around why Apple haven't done some sort of flip4mac drive-in thing with iTunes + AppleTV, and I don't get it. Maybe it does have something to do with the iTunes store. Whilst it's generally accepted that Apple make pretty much squat on music downloads, perhaps there is the potential there for them to actually make significant profits on the movie downloads?
Having said that, it still doesn't make sense to me. Surely the AppleTV would be more compelling if it (in conjunction with iTunes) offered an easy way to rip (unmodified a-la drive-in) all your DVDs you already own onto a server and then stream those files to AppleTV? Then Apple would sell more AppleTVs. Then, if Apple offered 1280 x 720/24p movies with 5.1 AC3 instead of the 640 x 360 ones with stereo that they do now, priced inbetween DVD and HD-DVD/blu-ray, there would be a compelling reason to buy any new content from the iTunes store rather than elsewhere (higher quality than DVD, but cheaper than HD-DVD/blu-ray).
The problem is that they are trying to attract the very studio's that sold you those DVDs to sell their content on iTMS and there is a lot more resistance to this with video than there was with audio, plus a lot of competition beginning to crop up. Apple needs the content, so they need to get the studios and broadcasters to let them distribute it, and do so at a price that Apple feels it's customers will buy. Not $20/movie when you are not getting the physical media or the extras, but $14.99 for a new release. Apple needs that price low to attract customers to iTMS and add the value to Apple TV that will make it a success.
If the standard QuickTime H.264 codec running on a 1.83 GHz Core Duo core cannot decode 1280 x 720/24p 5 Mbit/s H.264 in real-time, there's no way in hell that it can do it on AppeTV's underclocked Pentium-M. Not only is the CPU running at only just over 1/2 the clock speed of my CPU, it also has a slightly inferior SSE1/SSE2 implementation, and no SSE3 at all.
You do realize that the OS X in the Apple TV has much, much less to do than the Mac OS X in your Mac mini? No. That would be too easy to consider. I'm sure you receive e-mails on your Apple TV all the time. Or compile ports in DarwinPorts. Yep. Gotta keep the machine busy. While watching 720p video.
The H.264 component on the Apple TV is indeed different. But your conclusions are still quite a stretch.
And now you admit much of this is your opinion and nothing you have empirical proof of. This sounds like technical nitpicking to me. In the long run when Joe and Jane Consumer are watching video on their computer will they care if the video is being decoded on superior Nvidia GPU hardware or on less optimal Quicktime in the CPU? Nope.
Some users will care if it results in poorer picture quality. I freely admit that I shouldn't have been nearly as unequivocal as I was in my first post to this thread. Hopefully I have demonstrated that there are plenty of reasons why the picture quality of a Mac Mini and AppleTV will not necessarily be equal. I'm really trying to point out that people shouldn't assume that the picture quality of all devices are equal. There are many factors that affect it.
When it comes to Joe and Jane Consumer, they have demonstrated often that they don't care all that much about quality, given that they buy 128 kbps audio downloads in huge quantities, and purchase some iTunes Movies. So for them I'm sure this debate doesn't matter.
You do realize that the OS X in the Apple TV has much, much less to do than the Mac OS X in your Mac mini? No. That would be too easy to consider. I'm sure you receive e-mails on your Apple TV all the time. Or compile ports in DarwinPorts. Yep. Gotta keep the machine busy. While watching 720p video.
The H.264 component on the Apple TV is indeed different. But your conclusions are still quite a stretch.
Try this:
Quit all apps apart from Activity monitor, QuickTime and Finder.
On my MacBook, Activity monitor and pmTool (a utility that activity monitor uses) together take up
Now, open up a 1280 x 720/24p 5 Mbit/s H.264 file in QuickTime and play it back. On my MacBook (with one core disabled), the WindowServer, kernel and QuickTime CPU usages all increase. The usages vary quite a bit, the peak seems to be around 4% for WindowServer and 3% for the kernel. So, without Activity Monitor running, there would be at least 93% processor power available to QuickTime. It is without Activity Monitor running that I performed my test that demonstrated a heavy rate of dropped frames. And this on a Core Duo core with improved (over the Pentium-M in the AppleTV) SSE1/2 and SSE3 (which the Pentium-M doesn't have at all) running at almost twice the clock speed.
How exactly do you propose that the OS on the AppleTV is structured? Even if you drive the WindowServer and kernel CPU usages down to zero, there still isn't anywhere near enough CPU power available for QuickTime to decode a 1280 x 720/24p 5 Mbit/s H.264 stream in real-time. The AppleTV must be using the GeForce hardware-decode features.
If you think the AppleTV is doing it all on the CPU, why do you think Apple put a GeForce in there? Wouldn't it have been much cheaper to use a GMA950? And if Apple in the process of developing the AppleTV worked out how to reduce the CPU usage of their QuickTime H.264 decoder by a factor of something like 4, why didn't they deliver that improved codec with the release of QuickTime 7.1.5?
I'm quite aware.
Though, technically, your build numbers are misleading. It's running 10.4.7 with Darwin 10.4.8 and a build between 10.4.8 and 10.4.9.
Sure, there are lots of extra bits and pieces installed in my OS, but extra frameworks and the like take up no extra CPU resources just by being there. You have to be using them for that to happen. I should have said in my post that QuickTime and Finder where the only applications running when I did my test. There was nothing sucking up CPU cycles in the manner that you suggest.
Aside from dozens of agents and daemons, not to mention many additional kernel extensions.
If you think the AppleTV is doing it all on the CPU,
I don't. I do think that you need to stop jumping to conclusions, as you have just done yet again.
Other than not speculating at all, there isn't a more reasonable avenue for discussion. But then again, not speculating would mean no discussion either.
Aside from dozens of agents and daemons, not to mention many additional kernel extensions.
You must have missed the bit where I detailed the CPU usage of 10.4.9 running on my MacBook with one core disabled. Yes, there are lots of agents and daemons resident in memory, no, they weren't using any measurable CPU time during my tests. Quit all your GUI apps, launch Activity monitor and see for yourself what the CPU usage at idle is like.
I don't [think the AppleTV is doing it [decoding H.264] all on the CPU]. I do think that you need to stop jumping to conclusions, as you have just done yet again.
My statement was that there was "no way in hell it can do it [decode H.264] on AppeTV's underclocked Pentium-M." You then quoted this statement, and called it "quite a stretch". I'm sorry, but you are just contradicting yourself. If the AppleTV can't use its CPU to decode the H.264, where else is it going to do it?
Seems like Mr. H is just detailing the reasoning for what he believes to be the most likely scenario.
And his scenario happens to be correct as well.
this is apples big chance to again achieve cult status or blow it big time. should jobs decide to plug all the holes and make hacking this machine impossible or really hard, it will be like shooting yourself in the foot. if, on the other hand, they will allow the chip to fall where they may, this could be the start of a fantastic machine-running all codecs, using it as a ultra mini mac, whatever! let;s see what happens...
-----------
the count
http://thecountsworld.blogspot.com/
Word! It could be like Xbox Media Center -- except that, rather than modding your *existing* Xbox, folks'd *purchase* this device in order to mod it. With a larger HD and wider codec support, I'd be all over this! It seems like a device with a high Wifey Approval Factor. As it is, the Mac Mini in our bedroom pisses off my wife (and me) because FrontRow likes take a long time to scroll our Movies. The X360 MCE in our living room is great for streaming music -- but suck-ass video codec support (who wants to convert all their avi to wmv -- and don't tell me to transcode, that's a 0 on the WAF scale).
Anywho, all that to say this. If I can get unlimited video codec support on a fast Apple interface and not have my wife roll her eyes at me as I explain how to use it, I'll put one of these boxes on each of 3 different TVs around my house. How 'bout *that*, Mr Jobs?
Seems like Mr. H is just detailing the reasoning for what he believes to be the most likely scenario. This isn't jumping to conclusions in my book. If anything, his thorough reasoning is the exact opposite of jumping to conclusions.
Other than not speculating at all, there isn't a more reasonable avenue for discussion. But then again, not speculating would mean no discussion either.
Ditto that.
And I'd like to add, on balance, I think Mr. H is contributing useful technical arguments, hardware tests, and his thoughts to this thread. Many of the other contributors seem more concerned about his sentence anatomy rather than presenting their own detailed analysis for peer review.
Mr. H earnestly stuck out his neck to present even-toned thoughts and analysis. And naturally, they ought to be open for discussion. But why not present an equally long neck in response? Instead, it seems to me that on balance the collective retort to Mr. H is built on unconstructive critique (in a few instances, inflammatory critique) and small details of sentence nuance.
As we all know, snipper fire is easy work in anonymous fora.
Ditto that.
And I'd like to add, on balance, I think Mr. H is contributing useful technical arguments, hardware tests, and his thoughts to this thread.
…
Mr. H earnestly stuck out his neck to present even-toned thoughts and analysis. And naturally, they ought to be open for discussion.
Thanks. I would say that I didn't help all that much right at the beginning. I was a bit too forthright.