I primarily record shows via Over the Air digital broadcasts. I use MythTV, and record the HD versions of Office and Earl. So, I'm mostly self sufficient.
But, occasionally there is a problem with MythTV, my guide data, or the station's broadcast. In those cases, iTunes was a great secondary option. I could get missed episodes at a reasonable price, with decent viewing options (Laptop, iPod/iPhone, HTPC, AppleTV). I bought maybe four or five episodes last year. It's a shame to see these go.
I have tried the various streaming options from the broadcasters before. Both the pre-recorded programs, and live broadcast (e.g. NCAA final four games available from CBS streaming). I have found every instance of these to be too poor to actually watch. There are often problems even getting a video stream working, if it does start it doesn't last long before stopping, and in the rare case that I got a reliable stream - the video is a tiny lo-res thumbnail. No Thanks.
> any tuner device supported by linux should be usable with mythtv so if you find a USB tuner with linux support you can use it
MythTV is modular. You do the tuning/recording/processing on the backend, which talks to the frontend (display) via the network. These functions can be on the same box, but with the PS3 as a frontend, you would want a separate backend.
http://linuxtv.org/ has info on supported devices. My tuner of choice is the HDHomeRun, a network based dual HD tuner.
> The problem is I don't think there are any USB TV capture devices that work in MythTV.
The PS3 would make a good MythTV frontend (display) system. You're better off using a Linux box as the backend. That way you can still use the PS3 as a game system or Blue-Ray DVD player -- rather than leaving it in Linux mode all the time to record programs. As for tuners. the HDHomeRun is a good network-based HD tuner option.
> I have been using MythTV for something like 4 years now just because there isn't anything as good but I really hate it. It's so freaking slow and buggy. I mean once you have hundreds of recorded shows the thing just slows to a crawl. It takes forever just to delete a program. This all seems totally ridiculous considering the relatively tiny amount of data it is managing. Then there are the bugs like when it gets into some sort of bad state and the menus don't work (you can open the menu but can't select anything). Very poorly written piece of software if you ask me.
It's hard to say what the issue is from that description.
- Slow performance: All the standard Linux tuning applies. MythTV uses MySQL extensively, so tuning that performance is important. - Slow Delete: File system choice is important. ext* is very slow in deleting files. I use JFS because it deletes almost instantly. In newer MythTV versions, there is an option to do deletes in smaller chunks to avoid this problem on ext* systems. - I haven't seen bugs like those you described. I use 0.20.1, and it works well. MythTV is not ideal.. setup is difficult, as is configuration.. The menu systems could use some major improvements and configurability. It's definitely the worst DVR going, except for everything else.
The article did say that he was doing file transfers.
And, you simply cannot push 10,000 packets per second in any online game. Even if you assume a small packet size, you just don't have the bandwidth to do that many packets. Games are designed to use a modest amount of bandwidth, because that's all they can expect from the vast majority of their users. Also, because their servers act as an aggregation point for all that traffic. If they needed 10,000 packets/sec, their servers would have to be hosted in a site with multiple Terabit network feeds - not gonna happen.
Simply the only way to get up to the packet rates described are local LAN (Gigabit Ethernet. 100 Mb won't do it) under fairly high utilization. The only common app to do that is file transfer (xfer to NFS/SMB server, FTP/SCP copy, File Backup service, etc.)
This is a problem. It represents a bad decision by the OS designers, but it's not going to come into play when web browsing, playing online games, or anything else Internet based.
Okay. But, math doesn't match up with the numbers you typed.
1,500 Bytes is not the average packet, it's the maximum on most ethernet segments. But, the subject original subject is a stressful network environment effecting music playback. 10,000 packets per second is REALLY cranking the data.. so this isn't simple WWW browsing, etc. This is bulk file transfer. So, a large average packet size becomes more realistic in that environment.. say 1400 Bytes.
1,400 Bytes * 10,000 Packets per second = 14,000,000 BYTES / sec = 112,000,000 bits/sec = 112 Mbps
Obviously, that's not even possible on most common home networks, which are 100Mbps. But, an increasing number of people are doing Gig-E at home, in which case 112Mbps is well within the norms for bulk file transfer.
On modern fast multi-core systems, enforcing a pre-set cutoff for packet rate seems like a poor choice. As the linked article showed, the system had plenty of CPU left and didn't need to be throttled at that low a rate. There are also NIC and Driver factors in there.. others might be more or less efficient than the author's equipment -- offload of parts of packet processing and interrupt minimizing techniques can make a big difference.
In any case.. It's easy to say "that's what you get for using MS / Vista". But, really.. that's true in this case. Windows gives you the lowest common denominator. It's designed to be usable with any hardware, by users of any experience level, and to avoid problems by assuming a worst case scenario. So, that's the kind of solution you get given the assumptions MS uses. As we've seen in the Linux world, the solution is to take great pains to build a scheduler that holds up to ridiculous stresses.
I didn't realize WGA did periodic (constant?) checks on the system validity. I thought it was just a one-time check at installation (like entering the product ID on Win2K). That's hilarious that you can pay MS hundreds of dollar for the software, only to have it crippled when they screw up and accuse you of being a thief.
That's a misleading way of saying it has two digital tuners. 1080P is nonsense in that context, the tuners don't play any part in the decoding or display of the video. They just take the broadcast and convert it into a digital stream that the system saves to disk. Just like my HDHomeRun networked Tuners are "1080i or 720p tuners" because it grabs the broadcast and sends t to my MythTV box via UDP.
But, with the horsepower of the PS3, it could make a nice DVR. The cell processors can be utilized to decode the few 1080P H.264 channels available in Europe - or the much more common SD Widescreen channels.
Can it record TV programs to a network drive, or at least a USB drive? It can transcode for use on the PSP.. that's a nice use of the Cell power. Can that be done for iPods, iPhones, or Laptop use too?
Of course, it won't ever do some of the things MythTV does, like commercial flagging/skipping..
Following up on my own post... Based on various descriptions I have read on adobe.com, I think the hardware acceleration is limited to hardware overlay scaling. So, when you display a window larger than the actual video size, or full-screen, the application uses a video overlay to draw the video on. That overlay can be scaled up/down, and the GPU takes care of the scaling of the content, rather than relying on the software to scale & re-draw the frame at the desired size.
This is nice. But, it is a small fraction of the CPU overhead required for H.264 decoding. I think we're still a ways away from widespread GPU acceleration (unfortunately.. many apps could really use this).
Yeah, because Google throws in all those Flash ads and animated images to their search results.. Oh wait.
No, they can actually generate a ton of revenue from unobtrusive text ads placed in the WWW page. Hmmm..
I'll admit that I was never a big fan of YouTube before. The quality was horrible, and the content was mostly garbage. But, placing video advertisements before the desired content guarantees that my viewing goes from infrequent to almost never. It also opens the door to competitors who find a way to be less annoying in delivering video.
Yes, I see the same thing. That's one reason I switched to Camino as my WWW browser, it has flash-block functionality built-in. I have complained to a Safari developer about this. His response was basically "tell Adobe.. we can't do anything about it".
Hardware accel would help for H.264 video. But, I think Mac FlashPlayer just needs to be optimized, because I get that same high cpu load when running seemingly simple flash apps. It's really annoying when some flash animated advertisement is running on a www page, and my system fans kick into high gear to cool it. That sucks for such a lame purpose.
The article claims that that Adobe said it will use hardware acceleration for H.264.. Are there any more details on this?
Is it Windows-only? Probably. Does it use DirectX video acceleration APIs (do they handle H.264) or maybe OpenGL shader (GLSL) offload? If it's the second, it would have a chance for Mac and Linux support too.
I think this is fairly common in Universities that have a stand-alone campus (as opposed to one inter-mixed in a city environment).
My alma mater has an extensive system of steam tunnels(*) that run throughout the huge campus. These have been used for communications links for a long time. When I was there, we had an FDDI ring running to major buildings for a high speed backbone. I'm sure they've continued to upgrade the equipment on that fiber through the years. Having your own fiber offers a lot of interesting possibilities for great interconnect speed, and distributed services or data center decentralization.
> Converting a stock Debian Etch install to a Xen dom0 takes about 5 minutes, including the reboot. Creating a new domU takes about 2 minutes, from deciding to do it until I have an up-and-running virtual server.
Sure, getting a Xen-capable Linux going is simple. In recent Linux distributions it's just a matter of selecting a couple packages for installation.
Installing client VMs (DomU in the really intuitive Xen nomenclature) can be easy, and can be a MAJOR pain in the ass.
Installing the trivial 'ttylinux' was painless. Specify the ISO, start the VM, and it's done before it starts.
Installing CentOS was not so simple. Install failed completely and silently. Dig around for support info.. there are hundreds of sources, each very superficial and fragmented. Finally find a random user report on a forum recommending to set the OS type to Solaris for the install. Finally works.
Installing Windows was also not a simple process. The first stage of the install was easy. Point it to the ISO, and let it fly. But, after the system reboot, re-mounting the ISO and getting the VM to find it proves to be very difficult. Finding obscure command-line options gets it moving in the right direction (although, the documented parameters fail to actually mount the ISO), after much more searching and experimentation, I was able to get it to recognize the physical disk in a CD drive and complete the install.
Installing a pre-built Xen "virtual appliance" also proves to be near impossible. This is partly because the suppliers of those appliances have little or no Xen documentation (one can only assume this reflects their userbase.. VMWare documentation is more plentiful). And, it is partly because of the configuration issues in Xen. Is this a paravirtualized image or a full hardware virtualized image? What type of disk image does it use? Again, back to hunting down configuration information all over the Internet. Many suggestions, few actually work to do anything useful. I still haven't gotten these working.
Then, with the OS's that do successfully install, using them can be troublesome. In both Windows and Linux GUI environments, there is some quirk with the mouse pointer that causes the shown VNC pointer to be offset from the GUI pointer by varying amounts (yes, I did disable mouse acceleration. Problem decreases a bit, but is still quite bad). There are many other quirks, which often don't respond well to the documented fixes: making a VM actually reboot - rather than just stopping when rebooted, requiring command line intervention to restart the VM; automatically starting VMs on reboot of the hypervisor host (Dom0). Getting a hung VM to correctly respond with status incormation or actually quit when told to. And many more issues...
Managing and monitoring multiple VMs on a host is also quite weak on Xen, and the strength of VMWare ESX server. It's all done through an intuitive GUI, tons of status and monitoring information is available, and there are bunches of APIs to extend beyond what ESX does itself.
> When I compare that to the hell that was setting up (and remotely administering) VMWare, I realise I'd never want to go back. And that's without even getting into VMWare's habit of eventually swallowing up all RAM and swap on the host until everything grinds to a clanking halt.
The only way I could see Xen, in its current state, being superior is if you absolutely had to use only text based console management. You can, and often must, manage xen from the terminal. If you want to use a GUI, to ease management and hide the details of all those command-line tools, Xen just doesn't measure up.
I have seen some third party projects that aim to make better Xen management tools. Red Hat has a Python based GUI that is better than nothing. And, as I mentioned, XenSource's proprietary tools are not bad. The good news is, that Xen could easily be very competitive.. the hard work is done, the virtual machine works. Wrap a GUI around it, and it becomes available to 10x as many users.
It all depends on how it is packaged. Subscribing to shows, like you can do with podcasts in iTunes, would work very well as multicast. They might re-multicast the show several times, and weight it towards off-hours when utilization is lowest.
A Tivo Season Pass type model also works well for this.
It will obviously not work for 100% of the cases. True on-demand viewing, like browsing through YouTube, won't benefit much from multicast. But, they could cut down a huge percentage of bandwidth usage by using multicast where practical.
Depending on the type of service, users could be encouraged to use the pre-scheduled model by pricing it more attractively. If the multicast season pass was 1/3 the price of instant on-demand, most people would choose the cheaper option.
Xen has a lot of potential. The basic virtualization capabilities are on par with VMWare or anybody else.
What Xen _really_ blows at is usability / manageability. Setting up Xen is a pain in the ass, especially if you're on something other than 32bit x86. Figuring out obscure command line options and text config file syntax won't take them very far.
XenSource has a closed source, functionally limited GUI management tool in their free (as in beer) XenExpress. It makes managing Xen VMs more realistic, but the limitations are too severe (maximum of 4 VMs, missing some features).
If they want to compete with VMWare, and fend off KVM, they'll need a lot more traction. They only way they'll get it is to start building the user-base.
They need to open source their management tools, and make Xen as easy to use as VMWare. Maybe they need to hold back a few enterprise-grade features, so that they can still sell product at the high end. But, the common linux users, and low-end business users could still be enticed away from VMWare, to a more open solution, if it was available. If they continue their half-open approach, they even compete with themselves, in Xen on Ubuntu/Suse/RedHat.
If they don't open up, VMWare continues to dominate. Microsoft's upcoming hypervisor expands to the strong number 2 option, and other wildcards might crop up.. KVM with a good mgmnt too.
They don't necessarily need to. But, if they are useful, Intel could either adopt them or make their equivalent.
Adopt: x86-64 (AMD Created, Intel adopted it when the Itanium sunk)
Co-existing features: SIMD: MMX/SSE and 3DNow! (SSE eventually won out, but they co-existed for a long time).
Virtualization: Intel VT and AMD-V co-exist today, and both are used by virtualization projects like Xen.
Wasn't AMD also talking about licenses or agreements with other companies to allow for different types of coprocessor chips to be used alongside their processors?
There is some interesting potential in that realm.. Crypto accelerators for VPN, SAN, or other devices. Multimedia encode/decode accelerators (encode 1080P H.264 in real time?). Inevitable video game acceleration devices (physics co-processor, accelerated NIC chip, 3D GPU offload processor?).
Those would be even more interesting in home-user oriented Athlon64 boards. Multi-socket opteron boards are out of my price range.
This would make an excellent MythTV frontend.. Flash-based OS or Net-Boot. Small size, VGA or DVI output.
The 1GHz VIA C7 would not handle HD decoding on its own.. it's not fast enough.
But, the Unichrome features an MPEG2 decoder which offloads the CPU so that even the C7 could handle HD playback.
The question is: Is the version of the Unichrome GPU in this thing HD capable? Unfortunately, most of the Unichrome GPUs are limited to 1024x1024, which is obviously not going to cut it for HD.
As others mentioned, similar things can be done now -- an IDE/Flash boot into a minimal hypervisor Linux for Xen or KVM. That would also allow some flexibility, to maybe run a few things directly on the hardware. I would be very interested in an approach like this for my home Linux server.
For larger enterprise uses, the really simple hypervisor is nice. Just slap another box in there, and it is quickly added to your compute cluster. If they do it right, that system could even net-boot and auto-install the latest hypervisor image when it's first added. Factor in VMWare's "VMotion" stuff, where VMs can be moved among compute nodes in a cluster, and that simple compute node, along with a big NAS, is really slick.
After reading the blurb on this, it sounds an awful lot like "Solaris Zones" -- which is similar to BSD Jails or OpenVZ on Linux.
It's a kernel level partitioning of resources, to create virtualized hosts with low overhead. They all use the same kernel (so you couldn't have Linux/Windows/Mac virtual machines), but each system/app is unaware of the others.
That way, you can have two virtual instances, each running Apache, but with different/conflicting middleware below it -- and no worries about them crapping on one another.
The example they give in the article is being able to run Office 2003 and Office 2007 on the same machine. The concept behind it is cool. But, doesn't that example illustrate a lot of what is wrong with Windows -- they need an all new virtualization technology just to install two versions of Office on your PC?!?
I just grabbed the executive summary version, and didn't see any mention of virtualization..
To me, this seems like one of the more important aspects of power efficiency. Individual server efficiency is important, but the gains from higher utilization could be even more significant. Adding another core to a hypervisor will always be more efficient than adding a new system (CPU, Power Supply, disks, video, etc..). The energy efficient hardware can also be applied to the hypervisor hosts. Build efficient servers, and use as few of them as practical.
Many data centers are already greatly decreasing their server count using virtualization. This should be part of any data center energy efficiency discussion.
Years ago, I bought a very expensive Sony sub-notebook (PCG-SR7K, I think). It was great for portability.. very light and small. But, it had two big problems: The 10" screen was way too small for everyday use, and the keyboard was too small for decent typing.
I like the concept of this Asus.. small, light, flash storage, Linux OS. Great for many users primarily interested in web access. But, the 10" screen is too limiting.
For the $100 increase in price from 7" to 10", they could surely afford a better LCD panel. A 17" LCD is on sale at Best Buy this week for $129 retail (i.e. including the retailer markup, packaging, standalone monitor components and power supply, etc..). They could have easily done a display large enough to fill the top of the laptop that $100 markup. If they had done that, it would have been very powerful/usable.
As it is now, with a 10" display, it seems like a toy imitation of a real laptop.
I think the main problem Tivo has is getting people to go for more recurring monthly costs..
Their previous HD units were VERY expensive ($1,000 for the initial HD DirecTivo, $800 for Series3 Tivo, some decreases & rebates after that). I think a lot of people are not willing to pay several hundred dollars for a product, then keep paying significant monthly fees.
With a one year agreement, the monthly fee is $17. $15 for 2 years, and $13 for 3 years. That adds up quickly.
It's even more distressing when you see all the kickbacks and advertisements Tivo throws in to the service you're paying $17/month for
- Click Thumbs up for more info on this commercial - Click thumbs up to record the program being advertised - Here's an item in your main menu -- click here to view a commercial for our partner's product - Check out these programs we think you'll like -- Fox payed us a lot to make sure we reccomended their new shows - We kept track of all your viewing habits - commercials watched/skipped - and we're selling that info.. Thanks!
I realize they haven't made a profit, and they need to increase revenue. But, this junk is going about it in the wrong way.
The only way their service will be palatable is to work with Cable/Satellite providers and include the service price into the monthly service bill. A $5 monthly surcharge is manageable.. $60-80+ per month for cable or satellite, plus $17/month for Tivo guide data is not.
Slashdot Mirror
I primarily record shows via Over the Air digital broadcasts. I use MythTV, and record the HD versions of Office and Earl. So, I'm mostly self sufficient.
But, occasionally there is a problem with MythTV, my guide data, or the station's broadcast. In those cases, iTunes was a great secondary option. I could get missed episodes at a reasonable price, with decent viewing options (Laptop, iPod/iPhone, HTPC, AppleTV). I bought maybe four or five episodes last year. It's a shame to see these go.
I have tried the various streaming options from the broadcasters before. Both the pre-recorded programs, and live broadcast (e.g. NCAA final four games available from CBS streaming). I have found every instance of these to be too poor to actually watch. There are often problems even getting a video stream working, if it does start it doesn't last long before stopping, and in the rare case that I got a reliable stream - the video is a tiny lo-res thumbnail. No Thanks.
> any tuner device supported by linux should be usable with mythtv so if you find a USB tuner with linux support you can use it
MythTV is modular. You do the tuning/recording/processing on the backend, which talks to the frontend (display) via the network. These functions can be on the same box, but with the PS3 as a frontend, you would want a separate backend.
http://linuxtv.org/ has info on supported devices. My tuner of choice is the HDHomeRun, a network based dual HD tuner.
> The problem is I don't think there are any USB TV capture devices that work in MythTV.
The PS3 would make a good MythTV frontend (display) system. You're better off using a Linux box as the backend. That way you can still use the PS3 as a game system or Blue-Ray DVD player -- rather than leaving it in Linux mode all the time to record programs. As for tuners. the HDHomeRun is a good network-based HD tuner option.
> I have been using MythTV for something like 4 years now just because there isn't anything as good but I really hate it. It's so freaking slow and buggy. I mean once you have hundreds of recorded shows the thing just slows to a crawl. It takes forever just to delete a program. This all seems totally ridiculous considering the relatively tiny amount of data it is managing. Then there are the bugs like when it gets into some sort of bad state and the menus don't work (you can open the menu but can't select anything). Very poorly written piece of software if you ask me.
It's hard to say what the issue is from that description.
- Slow performance: All the standard Linux tuning applies. MythTV uses MySQL extensively, so tuning that performance is important.
- Slow Delete: File system choice is important. ext* is very slow in deleting files. I use JFS because it deletes almost instantly. In newer MythTV versions, there is an option to do deletes in smaller chunks to avoid this problem on ext* systems.
- I haven't seen bugs like those you described. I use 0.20.1, and it works well. MythTV is not ideal.. setup is difficult, as is configuration.. The menu systems could use some major improvements and configurability. It's definitely the worst DVR going, except for everything else.
The article did say that he was doing file transfers.
And, you simply cannot push 10,000 packets per second in any online game. Even if you assume a small packet size, you just don't have the bandwidth to do that many packets. Games are designed to use a modest amount of bandwidth, because that's all they can expect from the vast majority of their users. Also, because their servers act as an aggregation point for all that traffic. If they needed 10,000 packets/sec, their servers would have to be hosted in a site with multiple Terabit network feeds - not gonna happen.
Simply the only way to get up to the packet rates described are local LAN (Gigabit Ethernet. 100 Mb won't do it) under fairly high utilization. The only common app to do that is file transfer (xfer to NFS/SMB server, FTP/SCP copy, File Backup service, etc.)
This is a problem. It represents a bad decision by the OS designers, but it's not going to come into play when web browsing, playing online games, or anything else Internet based.
> Let's do some math
Okay. But, math doesn't match up with the numbers you typed.
1,500 Bytes is not the average packet, it's the maximum on most ethernet segments. But, the subject original subject is a stressful network environment effecting music playback. 10,000 packets per second is REALLY cranking the data.. so this isn't simple WWW browsing, etc. This is bulk file transfer. So, a large average packet size becomes more realistic in that environment.. say 1400 Bytes.
1,400 Bytes * 10,000 Packets per second = 14,000,000 BYTES / sec = 112,000,000 bits/sec = 112 Mbps
Obviously, that's not even possible on most common home networks, which are 100Mbps. But, an increasing number of people are doing Gig-E at home, in which case 112Mbps is well within the norms for bulk file transfer.
On modern fast multi-core systems, enforcing a pre-set cutoff for packet rate seems like a poor choice. As the linked article showed, the system had plenty of CPU left and didn't need to be throttled at that low a rate. There are also NIC and Driver factors in there.. others might be more or less efficient than the author's equipment -- offload of parts of packet processing and interrupt minimizing techniques can make a big difference.
In any case.. It's easy to say "that's what you get for using MS / Vista". But, really.. that's true in this case. Windows gives you the lowest common denominator. It's designed to be usable with any hardware, by users of any experience level, and to avoid problems by assuming a worst case scenario. So, that's the kind of solution you get given the assumptions MS uses. As we've seen in the Linux world, the solution is to take great pains to build a scheduler that holds up to ridiculous stresses.
I didn't realize WGA did periodic (constant?) checks on the system validity. I thought it was just a one-time check at installation (like entering the product ID on Win2K). That's hilarious that you can pay MS hundreds of dollar for the software, only to have it crippled when they screw up and accuse you of being a thief.
Why do people use that crap?
That's a misleading way of saying it has two digital tuners. 1080P is nonsense in that context, the tuners don't play any part in the decoding or display of the video. They just take the broadcast and convert it into a digital stream that the system saves to disk. Just like my HDHomeRun networked Tuners are "1080i or 720p tuners" because it grabs the broadcast and sends t to my MythTV box via UDP.
But, with the horsepower of the PS3, it could make a nice DVR. The cell processors can be utilized to decode the few 1080P H.264 channels available in Europe - or the much more common SD Widescreen channels.
Can it record TV programs to a network drive, or at least a USB drive?
It can transcode for use on the PSP.. that's a nice use of the Cell power. Can that be done for iPods, iPhones, or Laptop use too?
Of course, it won't ever do some of the things MythTV does, like commercial flagging/skipping..
Following up on my own post... Based on various descriptions I have read on adobe.com, I think the hardware acceleration is limited to hardware overlay scaling. So, when you display a window larger than the actual video size, or full-screen, the application uses a video overlay to draw the video on. That overlay can be scaled up/down, and the GPU takes care of the scaling of the content, rather than relying on the software to scale & re-draw the frame at the desired size.
This is nice. But, it is a small fraction of the CPU overhead required for H.264 decoding. I think we're still a ways away from widespread GPU acceleration (unfortunately.. many apps could really use this).
Yeah, because Google throws in all those Flash ads and animated images to their search results.. Oh wait.
No, they can actually generate a ton of revenue from unobtrusive text ads placed in the WWW page. Hmmm..
I'll admit that I was never a big fan of YouTube before. The quality was horrible, and the content was mostly garbage. But, placing video advertisements before the desired content guarantees that my viewing goes from infrequent to almost never. It also opens the door to competitors who find a way to be less annoying in delivering video.
Yes, I see the same thing. That's one reason I switched to Camino as my WWW browser, it has flash-block functionality built-in. I have complained to a Safari developer about this. His response was basically "tell Adobe.. we can't do anything about it".
Hardware accel would help for H.264 video. But, I think Mac FlashPlayer just needs to be optimized, because I get that same high cpu load when running seemingly simple flash apps. It's really annoying when some flash animated advertisement is running on a www page, and my system fans kick into high gear to cool it. That sucks for such a lame purpose.
The article claims that that Adobe said it will use hardware acceleration for H.264.. Are there any more details on this?
Is it Windows-only? Probably.
Does it use DirectX video acceleration APIs (do they handle H.264) or maybe OpenGL shader (GLSL) offload? If it's the second, it would have a chance for Mac and Linux support too.
I think this is fairly common in Universities that have a stand-alone campus (as opposed to one inter-mixed in a city environment).
t
My alma mater has an extensive system of steam tunnels(*) that run throughout the huge campus. These have been used for communications links for a long time. When I was there, we had an FDDI ring running to major buildings for a high speed backbone. I'm sure they've continued to upgrade the equipment on that fiber through the years. Having your own fiber offers a lot of interesting possibilities for great interconnect speed, and distributed services or data center decentralization.
(*) The MSU steam tunnels are the source of the Dungeons and Dragons tunnel games folklore, because of an incident with a disturbed child prodigy: http://en.wikipedia.org/wiki/Steam_tunnel_inciden
> Converting a stock Debian Etch install to a Xen dom0 takes about 5 minutes, including the reboot. Creating a new domU takes about 2 minutes, from deciding to do it until I have an up-and-running virtual server.
Sure, getting a Xen-capable Linux going is simple. In recent Linux distributions it's just a matter of selecting a couple packages for installation.
Installing client VMs (DomU in the really intuitive Xen nomenclature) can be easy, and can be a MAJOR pain in the ass.
Installing the trivial 'ttylinux' was painless. Specify the ISO, start the VM, and it's done before it starts.
Installing CentOS was not so simple. Install failed completely and silently. Dig around for support info.. there are hundreds of sources, each very superficial and fragmented. Finally find a random user report on a forum recommending to set the OS type to Solaris for the install. Finally works.
Installing Windows was also not a simple process. The first stage of the install was easy. Point it to the ISO, and let it fly. But, after the system reboot, re-mounting the ISO and getting the VM to find it proves to be very difficult. Finding obscure command-line options gets it moving in the right direction (although, the documented parameters fail to actually mount the ISO), after much more searching and experimentation, I was able to get it to recognize the physical disk in a CD drive and complete the install.
Installing a pre-built Xen "virtual appliance" also proves to be near impossible. This is partly because the suppliers of those appliances have little or no Xen documentation (one can only assume this reflects their userbase.. VMWare documentation is more plentiful). And, it is partly because of the configuration issues in Xen. Is this a paravirtualized image or a full hardware virtualized image? What type of disk image does it use? Again, back to hunting down configuration information all over the Internet. Many suggestions, few actually work to do anything useful. I still haven't gotten these working.
Then, with the OS's that do successfully install, using them can be troublesome. In both Windows and Linux GUI environments, there is some quirk with the mouse pointer that causes the shown VNC pointer to be offset from the GUI pointer by varying amounts (yes, I did disable mouse acceleration. Problem decreases a bit, but is still quite bad). There are many other quirks, which often don't respond well to the documented fixes: making a VM actually reboot - rather than just stopping when rebooted, requiring command line intervention to restart the VM; automatically starting VMs on reboot of the hypervisor host (Dom0). Getting a hung VM to correctly respond with status incormation or actually quit when told to. And many more issues...
Managing and monitoring multiple VMs on a host is also quite weak on Xen, and the strength of VMWare ESX server. It's all done through an intuitive GUI, tons of status and monitoring information is available, and there are bunches of APIs to extend beyond what ESX does itself.
> When I compare that to the hell that was setting up (and remotely administering) VMWare, I realise I'd never want to go back. And that's without even getting into VMWare's habit of eventually swallowing up all RAM and swap on the host until everything grinds to a clanking halt.
The only way I could see Xen, in its current state, being superior is if you absolutely had to use only text based console management. You can, and often must, manage xen from the terminal. If you want to use a GUI, to ease management and hide the details of all those command-line tools, Xen just doesn't measure up.
I have seen some third party projects that aim to make better Xen management tools. Red Hat has a Python based GUI that is better than nothing. And, as I mentioned, XenSource's proprietary tools are not bad. The good news is, that Xen could easily be very competitive.. the hard work is done, the virtual machine works. Wrap a GUI around it, and it becomes available to 10x as many users.
It all depends on how it is packaged. Subscribing to shows, like you can do with podcasts in iTunes, would work very well as multicast. They might re-multicast the show several times, and weight it towards off-hours when utilization is lowest.
A Tivo Season Pass type model also works well for this.
It will obviously not work for 100% of the cases. True on-demand viewing, like browsing through YouTube, won't benefit much from multicast. But, they could cut down a huge percentage of bandwidth usage by using multicast where practical.
Depending on the type of service, users could be encouraged to use the pre-scheduled model by pricing it more attractively. If the multicast season pass was 1/3 the price of instant on-demand, most people would choose the cheaper option.
Xen has a lot of potential. The basic virtualization capabilities are on par with VMWare or anybody else.
What Xen _really_ blows at is usability / manageability. Setting up Xen is a pain in the ass, especially if you're on something other than 32bit x86. Figuring out obscure command line options and text config file syntax won't take them very far.
XenSource has a closed source, functionally limited GUI management tool in their free (as in beer) XenExpress. It makes managing Xen VMs more realistic, but the limitations are too severe (maximum of 4 VMs, missing some features).
If they want to compete with VMWare, and fend off KVM, they'll need a lot more traction. They only way they'll get it is to start building the user-base.
They need to open source their management tools, and make Xen as easy to use as VMWare. Maybe they need to hold back a few enterprise-grade features, so that they can still sell product at the high end. But, the common linux users, and low-end business users could still be enticed away from VMWare, to a more open solution, if it was available. If they continue their half-open approach, they even compete with themselves, in Xen on Ubuntu/Suse/RedHat.
If they don't open up, VMWare continues to dominate. Microsoft's upcoming hypervisor expands to the strong number 2 option, and other wildcards might crop up.. KVM with a good mgmnt too.
They don't necessarily need to. But, if they are useful, Intel could either adopt them or make their equivalent.
Adopt: x86-64 (AMD Created, Intel adopted it when the Itanium sunk)
Co-existing features: SIMD: MMX/SSE and 3DNow! (SSE eventually won out, but they co-existed for a long time).
Virtualization: Intel VT and AMD-V co-exist today, and both are used by virtualization projects like Xen.
Wasn't AMD also talking about licenses or agreements with other companies to allow for different types of coprocessor chips to be used alongside their processors?
There is some interesting potential in that realm.. Crypto accelerators for VPN, SAN, or other devices. Multimedia encode/decode accelerators (encode 1080P H.264 in real time?). Inevitable video game acceleration devices (physics co-processor, accelerated NIC chip, 3D GPU offload processor?).
Those would be even more interesting in home-user oriented Athlon64 boards. Multi-socket opteron boards are out of my price range.
> Have fun transcoding your HD content to MPEG2. A better solution for your goal would be a real processor or a dedicated IC
Huh? The U.S. broadcast standard for digital TV is MPEG2. So, ALL the HD content on my MythTV system is already MPEG2.
This would make an excellent MythTV frontend.. Flash-based OS or Net-Boot. Small size, VGA or DVI output.
The 1GHz VIA C7 would not handle HD decoding on its own.. it's not fast enough.
But, the Unichrome features an MPEG2 decoder which offloads the CPU so that even the C7 could handle HD playback.
The question is: Is the version of the Unichrome GPU in this thing HD capable? Unfortunately, most of the Unichrome GPUs are limited to 1024x1024, which is obviously not going to cut it for HD.
Have you not been paying attention? Funding "faith based organizations" was the first thing Bush did after he got elected in 2000.
Congress initially denied funding, citing the separation of church and state, and Bush bypassed them via an "Executive Order".
Welcome to the theocracy.
I guess they see how well all those middle-eastern governments are working, and want to bring the same thing to the U.S.
As others mentioned, similar things can be done now -- an IDE/Flash boot into a minimal hypervisor Linux for Xen or KVM. That would also allow some flexibility, to maybe run a few things directly on the hardware. I would be very interested in an approach like this for my home Linux server.
For larger enterprise uses, the really simple hypervisor is nice. Just slap another box in there, and it is quickly added to your compute cluster. If they do it right, that system could even net-boot and auto-install the latest hypervisor image when it's first added. Factor in VMWare's "VMotion" stuff, where VMs can be moved among compute nodes in a cluster, and that simple compute node, along with a big NAS, is really slick.
After reading the blurb on this, it sounds an awful lot like "Solaris Zones" -- which is similar to BSD Jails or OpenVZ on Linux.
It's a kernel level partitioning of resources, to create virtualized hosts with low overhead. They all use the same kernel (so you couldn't have Linux/Windows/Mac virtual machines), but each system/app is unaware of the others.
That way, you can have two virtual instances, each running Apache, but with different/conflicting middleware below it -- and no worries about them crapping on one another.
The example they give in the article is being able to run Office 2003 and Office 2007 on the same machine. The concept behind it is cool. But, doesn't that example illustrate a lot of what is wrong with Windows -- they need an all new virtualization technology just to install two versions of Office on your PC?!?
I just grabbed the executive summary version, and didn't see any mention of virtualization..
To me, this seems like one of the more important aspects of power efficiency. Individual server efficiency is important, but the gains from higher utilization could be even more significant. Adding another core to a hypervisor will always be more efficient than adding a new system (CPU, Power Supply, disks, video, etc..). The energy efficient hardware can also be applied to the hypervisor hosts. Build efficient servers, and use as few of them as practical.
Many data centers are already greatly decreasing their server count using virtualization. This should be part of any data center energy efficiency discussion.
Years ago, I bought a very expensive Sony sub-notebook (PCG-SR7K, I think). It was great for portability.. very light and small.
But, it had two big problems: The 10" screen was way too small for everyday use, and the keyboard was too small for decent typing.
I like the concept of this Asus.. small, light, flash storage, Linux OS. Great for many users primarily interested in web access. But, the 10" screen is too limiting.
For the $100 increase in price from 7" to 10", they could surely afford a better LCD panel. A 17" LCD is on sale at Best Buy this week for $129 retail (i.e. including the retailer markup, packaging, standalone monitor components and power supply, etc..). They could have easily done a display large enough to fill the top of the laptop that $100 markup. If they had done that, it would have been very powerful/usable.
As it is now, with a 10" display, it seems like a toy imitation of a real laptop.
I think the main problem Tivo has is getting people to go for more recurring monthly costs..
Their previous HD units were VERY expensive ($1,000 for the initial HD DirecTivo, $800 for Series3 Tivo, some decreases & rebates after that). I think a lot of people are not willing to pay several hundred dollars for a product, then keep paying significant monthly fees.
With a one year agreement, the monthly fee is $17. $15 for 2 years, and $13 for 3 years. That adds up quickly.
It's even more distressing when you see all the kickbacks and advertisements Tivo throws in to the service you're paying $17/month for
- Click Thumbs up for more info on this commercial
- Click thumbs up to record the program being advertised
- Here's an item in your main menu -- click here to view a commercial for our partner's product
- Check out these programs we think you'll like -- Fox payed us a lot to make sure we reccomended their new shows
- We kept track of all your viewing habits - commercials watched/skipped - and we're selling that info.. Thanks!
I realize they haven't made a profit, and they need to increase revenue. But, this junk is going about it in the wrong way.
The only way their service will be palatable is to work with Cable/Satellite providers and include the service price into the monthly service bill. A $5 monthly surcharge is manageable.. $60-80+ per month for cable or satellite, plus $17/month for Tivo guide data is not.