| dijenerate | hi guys | 00:22 |
|---|---|---|
| dijenerate | GrueMaster: howdy | 00:23 |
| GrueMaster | hey. | 00:23 |
| dijenerate | question, is the tiny atom based prototype you have about the size of a nokia N800? | 00:24 |
| GrueMaster | very close. | 00:25 |
| dijenerate | and the power requirements are also close | 00:25 |
| GrueMaster | not sure. I have no specs on the system. | 00:25 |
| GrueMaster | Other than it uses atom and i'm running Linux on it. | 00:26 |
| dijenerate | so atom based systems can technically fit into the same footprint as 'arm' based PDAs | 00:26 |
| GrueMaster | getting there, yes. | 00:26 |
| dijenerate | 'almost' | 00:27 |
| GrueMaster | http://www.engadget.com/2008/04/02/video-lenovos-ideapad-u8-mid-teases-with-intels-atom/ | 00:28 |
| dijenerate | I just saw the Aigo MID unboxing video and realized the battery is a 3.7V 2700MaH that manages to last 2.5hrs on an x86 machine | 00:28 |
| GrueMaster | That's one sample. | 00:28 |
| dijenerate | I think I've found the point of the Centrino Atom! | 00:28 |
| dijenerate | only question now is, can they run completely enclosed like 'arm' systems can? | 00:40 |
| dijenerate | or do they still have 'stalker' fans | 00:41 |
| dijenerate | lol | 00:41 |
| GrueMaster | The unit I have has a very small fan on it that only kicks in under load. Yea, it's warmer than a PDA, but considering the technology, it's not bad. | 00:43 |
| dijenerate | does it support AMD64 instructions like the Core2 does? | 00:46 |
| dijenerate | CPU-Z reports the Atom as a Core2 Silverthorne | 00:46 |
| GrueMaster | no. YOu hardly need it for desktop, why would you want it now for a unit the size of a psp? | 00:47 |
| GrueMaster | The main benifit to x86-64 is memory addressing. SO-Dimms aren't that small yet. | 00:49 |
| dijenerate | ok | 00:55 |
| dijenerate | the CPU is really a lot of smoke and mirrors when compared to a desktop CPU | 00:55 |
| dijenerate | like being unable to handle rvmb encoded files at 644 x 488 881kbps without CPU usage hitting 100% | 00:56 |
| dijenerate | but an mpeg4 avi at 512x388 25fps 539kbps plays smoothly at 40~50% CPU usage | 00:58 |
| GrueMaster | what is rvmb? | 00:58 |
| dijenerate | that makes no sense mathematically considering what each format requires of the CPU but when you consider that there are dedicated mpeg4 decoding registers in the silverthorne | 00:59 |
| GrueMaster | And do you have any sample videos? | 00:59 |
| dijenerate | along with mp3, mpeg2 and mpeg1 among others, the plot thickens | 00:59 |
| dijenerate | real one HD format | 00:59 |
| GrueMaster | I've been playing some of the 1080i VC1 videos from Microsoft, and they look great. 10% cpu usage. | 00:59 |
| GrueMaster | Mainly for i/o | 00:59 |
| dijenerate | dedicated registers | 01:01 |
| GrueMaster | yep. | 01:01 |
| dijenerate | supporting flash etc | 01:01 |
| dijenerate | so the main CPU registers don't have to do the math | 01:01 |
| dijenerate | that is cheating tho' | 01:01 |
| GrueMaster | How? Most video cards do hardware decoding. | 01:02 |
| GrueMaster | nVidia and ATI both do that. | 01:02 |
| dijenerate | gives good looking results when running layman demos, that and using the Centrino name seems a bit dishonest | 01:02 |
| GrueMaster | Centrino Atom. There's a difference. | 01:02 |
| GrueMaster | Centrino is a name that defines a wireless mobile platform. | 01:03 |
| dijenerate | yes because it makes sense to save energy | 01:03 |
| dijenerate | but not because the video card's GPU can't do it | 01:03 |
| GrueMaster | And that is a bad thing? | 01:03 |
| dijenerate | but in this case the Atom can't | 01:04 |
| GrueMaster | Can't what? | 01:04 |
| GrueMaster | Last fall when I only had 2D drivers and software decoders, I could still do 1080i video, just with higher cpu utilization. | 01:04 |
| dijenerate | yeah, originally, but users of the old Centrino line have come to expect a certain amount of performance... intel is banking on that | 01:04 |
| dijenerate | it's not the graphics chipset I'm talking about | 01:05 |
| GrueMaster | Again, for what? YOu are assuming this is for high performance computing, which it is not. | 01:05 |
| dijenerate | the CPU can't handle it | 01:06 |
| dijenerate | if it had to do it conventionally, it would shoot up to 100% in no time | 01:06 |
| dijenerate | but there is a specific application I can think of that can benefit from the Atom | 01:06 |
| dijenerate | because of how close it is to the 'arm' platform | 01:06 |
| GrueMaster | With the wireless world, one of the design ideas for this is as a remote terminal. If you are travelling and you need to do some high powered number crunching, you can remote into your desktop system at home or work. | 01:07 |
| dijenerate | I know, you have said this before, but a celeron is not for high performance computing either and still outperforms this | 01:07 |
| dijenerate | which means that you could just as easily go 'arm' | 01:07 |
| GrueMaster | But how many Centrino platforms do you see today that can perform like a cell phone? | 01:07 |
| dijenerate | this is where I have the conflict... | 01:07 |
| dijenerate | it does everything 'arm' already does but can't do the things that x86 has done for ages without 'cheats' like dedicated registers | 01:08 |
| GrueMaster | Have you seen the video on moblin.org? While the systems in the video are mostly mockups, they are pretty close. | 01:08 |
| dijenerate | so if it isn't intended for high performance computing 'read: x86's domain' why not just use 'arm' | 01:09 |
| GrueMaster | One of the issues with arm is the total web experience. currently, there are ~12% of the plugins needed for most web sites available to arm, whereas there is ~80% available to x86 Linux. | 01:10 |
| dijenerate | http://www.youtube.com/watch?v=YJIymS3YdOo | 01:11 |
| GrueMaster | Also, try running Linux AND windows on an arm simultaneously. | 01:12 |
| dijenerate | this shows up the dedicated registers since there are none for the real video | 01:12 |
| GrueMaster | Of course, no one in this forum cares about windows. | 01:12 |
| dijenerate | we can see how weak the atom really is | 01:12 |
| dijenerate | that is a good point | 01:13 |
| dijenerate | so for x86 plugin support yes it is good | 01:13 |
| GrueMaster | But again, you are leaving power out of the equation. | 01:13 |
| dijenerate | but since the days when I did zaurus development, I wondered why plugins weren't just ported to 'arm' | 01:14 |
| GrueMaster | The lowest powered mobile processor is currently ~25w. This is ~3w. | 01:14 |
| dijenerate | they were ported to powerpc | 01:14 |
| dijenerate | my core solo is ~5W | 01:14 |
| dijenerate | what are you talking about | 01:14 |
| dijenerate | and the C7M is 7W | 01:14 |
| dijenerate | the AMD geode is somewhere in that vicinity as well | 01:15 |
| GrueMaster | The AMD is under powered in comparison. I have seen those benchmarks. | 01:15 |
| dijenerate | actually the Geode is a good example of what low power x86 should look like | 01:15 |
| dijenerate | it's also very old | 01:15 |
| GrueMaster | at any rate, it's beer time here on a hot Friday afternoon. | 01:16 |
| dijenerate | so unless the atom blows it away, the comparison is something for intel to be ashamed of | 01:16 |
| dijenerate | ok, np | 01:16 |
| dijenerate | thanks for the time to discuss this with me | 01:16 |
| dijenerate | ttyl | 01:16 |
| dijenerate | !taking | 01:16 |
| GrueMaster | Not necessarily. The Atom is a partial core redesign, which means it almost is completely new architecture. | 01:17 |
| GrueMaster | Not bad for a first attempt. | 01:17 |
| GrueMaster | And I've been working with Intel processors since the first P6. I know how they can be. | 01:18 |
| GrueMaster | later. | 01:18 |
| dijenerate | thing is intel hasn't really done a 'from scratch' core design since the Pentium M days, and now they licence AMD's tech, I have little respect for them these days | 01:19 |
| dijenerate | so most of these 'new' processors don't count as first attempts in my books | 01:19 |
| dijenerate | innovation begins with A these days not I | 01:19 |
| dijenerate | what intel has over AMD is way better marketing, they are so good they even believe their own hype | 01:20 |
| dijenerate | silly question, what's the difference between the menlow crownbeach and the standard menlow image | 02:07 |
| dijenerate | I'm having this random looping problem with menlow when i try to run it on a core solo GMA950 based device | 02:08 |
| === asac_ is now known as asac | ||
| === Zic_ is now known as Zic | ||
| Zic | lool: ping | 17:44 |
| Zic | (it does not concern directly Ubuntu Mobile, if you can query me when you are here, it will be cool :)) | 17:53 |
| === landley__ is now known as landley | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!