GE, 3M, IBM, Wal-mart, Nike, Exxon, AT&T, J&J and most of the finance industry disagree with you. What was your metric for "successful" again? Smartphone production or something?
How about business metrics like:
Market Cap: Apple (546B), Exxon (400B), AT&T (209B), WalMart (236B), GE (220B) etc etc. Apple is the largest US company and amongst the top of the world Annual profit: Apple (25.9B), Exxon (41B), WMT (15.7B), AT&T (4B) etc etc
With the exception of Exxon Mobile Apple's the most profitable company with largest market cap and one of the highest growth rates in both market cap and profits. I could have also listed revenue here where again Apple would be one of the top.
So I think you should think before you post about success or not. For any investor Apple has been THE success story and it's not over.
Did you even see the keynote? The reason Apple unveiled the phone in January and sent on sale in June was said by Steve during the keynote. He said the regulations for a new company to start selling a new phone required them to submit the phone for a long 3-4 month review by FCC or what not. However they didn't want the phone details to be leaked this way so they made the keynote introducing the whole phone and THEN submitted it to FCC when everyone already knew about it.
Except Apple made the iPad first, then when they had gotten the touch interface and screen to work decently they decided that they can make a phone too and decided to pull that first. Interesting that they could do ALL of that in a measly four months.
Are you keeping your apps open all the time? I found one of the major battery hoggers to be background apps and especially those you'd never suspect (like facebook). Especially bad is Viber. If you get a message in Viber you better read it and kill the bloody thing fast as otherwise it'll run your battery dry and your phone hot in under an hour. And I've seen that on multiple phones so it's a software issue of Viber. Try to see what happens if you after exiting the apps always kill them. Might be in for a surprise in battery life. And after you've confirmed that you can start to look which ones are the idiots hogging all the juice by keeping some running and killing others...
Wouldn't you be better off to use a tablet for that? Firstly the bigger screen makes it more comfortable, but you'd also not have to rush the battery as an iPad for example can last ca 10-11h playing movies so a 4h commute would leave you with a good 6h of work time left on the tablet possibly as you don't use it the whole time you'll have plenty even to watch another movie in the evening in bed:)
Seriously? You'd be satisfied with a phone that lasts a full day? I just recently was in supermarket reading e-mail in the checkout queue and a guy was speaking with the cashier about phones and the cashier asked if he's doing something wrong because he bought a smartphone (looked like a HTC or smth) and it lasts at best a full day even if he just calls. The other guy who seemed to be a phone guy or smth said that's normal and that his lasts approximately a day if all goes well, but if he uses it more he has to plug it in and that's normal. All smartphones do that... I just shook my head and kept reading e-mails on my iPhone 4 that I had taken from the charger the previous morning at 8 AM (it was late afternoon when this happened, so ca 30h later) and it still had 76% of charge left. I use the phone quite substantially, I use it for personal hotspot at times, I play games on it and constantly browse web and facebook. I also use Waze for traffic information and that's the only bigger battery drainer due to constant GPS usage. But the phone easy lasts 2-3 days, sometimes 4 days. And it's running latest iOS 5.1.1 and is almost 2 years old (bought 2010 october).
So one of the things I think is THE main failure of Android is the phones have crappy battery and the OS doesn't seem to be optimized for really running conservatively on the battery. And I haven't tweaked the OS in any way really to sustain the phone longer (about 50% of the time I'm on 3G, not wifi). I've only turned off location services for a few things (including notifications as I don't use location fencing right now), but most notifications and location services are still used and that has only a small impact. I've found some third party apps that if left in the background do consume a lot of battery even though they shouldn't (Facebook, Waze, Viber are extreme drainers if they are in background, not while in foreground). So killing those apps after I've finished using them expanded my battery life by a day and that's an issue with the 3rd party app and possibly somewhat also the OS.
Oh they can. It'll be enough if the device utterly sucks on the hardware end. For example if constant lockups like the one during presentation keep occurring or if the fans don't work properly after say 1 week of usage and it heats up to unbearable levels or if the battery lasts only the distance between two power sockets at best.
Considering the iPhone's still supported start from 3Gs onwards and that iPhone sales have grown close to exponentially I'd say the difference between the two denominators is smaller than you assume it is. But I think it's 80% of devices still in use (i.e. reporting something home through carriers for debugging purposes) and I'd not be surprised if the number is about the whole install base as the recent Apple presentation did show also OS versions of 3.x that have to be early iPhone's too, not just latest models.
I think the "surprise" part was only to the ATLAS controller. The fact that tidal forces affect the beam has been known since LEP times as is the water level in lake Geneva and also the TGV schedule that used to have huge electricity spikes as they "grounded" their lines and said grounding happened to reach LHC 100m below causing the orbit of the beam to jump. Hell, I've been working in the CMS experiment since 2004 and I knew about the tidal force compensation need and that's why the LHC control room guy was calmly saying he has to adjust for moon's tide as it's part of their operating procedure. Same goes to monitoring the rain quantity and snow melting to knowing what the water level is in lake Geneva so that they can compensate for that too.
Damn hit submit and then remembered your other parts of the question. With regard to safety and redundancy it's actually in favor of SpaceX. As an example from Falcon Heavy wiki: "The structural safety margins are 40% above flight loads, higher than the 25% margins of other rockets.[10]". They've designed the whole things from ground up for manned flight with extended safety margins so this is not the reason for cheapness...
And with regard to Musk and his motives for the company. Look at the recent 60 minutes interview with him. When Neil Armstrong etc claim that one shouldn't push for commercialization and NASA should do everything themselves (sounds like SLS lobbying to get the old guys out and get support without due consideration what they're doing) Elon's basically in tears. You can see that those guys were/are his heroes and motivation to do the same kind of stuff and it really hurt him. So he's not a fat CEO waiting for a fat paycheck, but this IS his vision and hope and that defines his priorities.
I think a lot of this is indeed coming from the fact that SpaceX is operating like a startup. They also have claimed a lot of the cost reduction comes from the fact that they manufacture every single component themselves in their own factory where the raw materials come in from one door and spaceships/rockets come out the other. They also re-use a lot of the tooling etc because of the design (Falcon 9 is basically 9x Falcon 1 and Falcon Heavy is basically 3x Falcon 9 with inter-exchange of fuel). Due to their relative size as a company their corporate overhead has to be an order of magnitude smaller than say Boeing or Lockheed-Martin. And their main interest is first off to get to be a major player in the space industry and to compete they have to be unique in some way and their way is the cost. I'm guessing their cross margin per launch is quite small while I'd not be surprised if others have a 100% or more cross margin. Musk's vision is to help make humans a space faring civilization with cost to orbit heavily reduced and allowing exploration of the solar system including colonization. As I've understood his hope is to retire on Mars so he's pushing with the company towards making that a reality within his lifetime (assuming he retires at 60/65 he's got 20/25 years left, it's not fully utopic).
And having worked for a private company and government contract and a large EU project I can tell you that the costs shown in a large EU project per person are far higher than you'd normally associate with the tasks done just because you can. And considering that there aren't too many players around I'd not be surprised if they've kept the launch device costs somewhat bloated because they can and it's being disrupter right now by an independent player that has different motives:)
Oh forgot to add. Space X took the risk on themselves when the did the Falcon 1 and it failed 3x in row. They bet the company on the 4th launch and it worked and have been delivering cargo to LEO and now to ISS since. The Falcon 9 works (it has had 3/3 successes) and Falcon Heavy is basically re-using a lot of the Falcon 9 material so it's likely they'll make it work too. The maiden flight of SLS is planned in 2017 while I'd assume the Falcon Heavy or it's successor might easily be providing the service in 2017 for a lot cheaper (they'll probably have ironed out the reusability part of the first stages by then).
You forget the alternative. If the public funds the development of new rocket systems by NASA and keeps using them at a cost of 5x that of the private company, then after N launches the total cost for the public is higher than had they developed it in cooperation with a private enterprise that then operates the technology under contracts at a much lower cost. As an example, Falcon 9 of Space X has lift capacity of 10 tons for a cost of $56M. The Delta-IV rocket had 22 tons and $300M so the cost per kg is 3x higher. The Ariane 5 rocket by ESA has 21 ton lift to LEO and cost of $200M. The Atlas V (earlier one) had lift on 9t at a cost of $125M.
Now if Space X can pull off Falcon Heavy (the first launch is planned already in 2013) then the planned cost of 53t to LEO is $80-125M. That gives it a per kg cost of 5.8x less than Delta-IV, 4x less than Ariane 5 etc. And that's assuming the high end of the price range. It's also a rocket that can deliver cargo to trans-lunar-orbit or even to Mars (14t to Mars, 16t to trans lunar). Why the hell would we need an SLS with 50-130t capacity with an outrageous price tag when we can just launch two Falcon 9 heavy's for the same capacity and probably less than the equivalent launch cost and if need be assemble the final inter-plantery spacecraft in orbit...
I'd like to point out that Space X did first prove they're capable of real life feats with Musk's own money. Yes the latest developments and the docking with ISS has been co-funded (or mostly funded) by NASA and they'll pay them more for additional launches, but they're getting a real deal out of it as SpaceX is doing it way cheaper than any alternative and they actually deliver already today, not far fetched promises.
So the word means perfectly what I meant. We DESERVE to die if we're still stuck only on this rock... Actually I think that's true even if we reduce the timeframe to say 1000 years.
I guess my irony detector is malfunctioning or you really misunderstood the post you responded to. The impressive part is measuring the actual Andromeda galaxy lateral motion that is so tiny form our perspective and timescale that doing that accurately enough in 8 years is the impressive part. The fact that NVIDIA made an animation of it realtime or not is fully besides the point.
Though on a side note I agree that GPU's are becoming crazy fast...
Virtualization in any real scale is a cluster. HPC is a cluster. There's no need to cluster a cluster, it's pointless. Virtualization has no business in HPC, but in almost every other computing aspect it makes real sense.
Yes it does make sense. Software release management. It's far more comfortable to keep a dedicated hardnode OS running and deploy user needed software VM's in containers on the hardnode as the user needs arise. The container based virtualization has close to 0 overhead in all aspects (including I/O) and allows for live migration (low priority) as well as easy deployment of new OS's. Think about 200 physical servers that have to be reinstalled over PXE asking IPMI to make sure the boot devices are right, the PXE configs etc. There are plenty of things that can go wrong in those installations even if they work 90% of the time. If however it's host based authentication and you run parallel ssh to destroy VM's and deploy new ones you can do it nicely from one single central host on all HN's at the same time without downtime at all (reserve a VM, wait until it's free and redeploy without affecting any other VM on the node). The cost of 1-2% overhead max is well worth the management advantage.
We run a 1200 core HPC cluster with 900TB of storage. Our workflows are mostly CERN analysis jobs that come through Grid so we don't really have a control on what exactly the user is doing in his/her job. We used to allow up to the number of cores of jobs in any worknode, but had issues with misbehaving jobs sometimes killing off the whole node taking the other jobs with it when some code ran amok on memory allocations or what not. We attempted various resource controls, but still at times they managed to make the node unresponsive.
Since about 1 month we run 100% virtualized OpenVZ cluster with every compute core as a VM (and we actually over-provision by 10% as a lot of jobs are data intense and not CPU intense). So far no single node has died due to bad user behavior. Single VM's have ran into OOM, but as they are virtual containers the jobs are killed inside and the VM recovers nicely allowing the next job to execute. As far as we've seen studies and tested the OpenVZ container based virtualization doesn't have the drawbacks of local disk I/O and network overheads and as it's just a container the overhead on CPU is negligible. Also what it allows us is that if the experiments declare a new needed workernode software content we can just deploy one such instance, clone it and redeploy all VM's far faster than we could reinstall the nodes. (The HN's run CentOS 6 while the WN's require SL5 for example so as long as you remain with Linux you're fine).
We also run our central scheduling servers and Grid CE's and information systems and what not in the same OpenVZ containers allowing for example to live migrate between nodes to allow for service work and allowing 24 core / 128GB ram nodes to be used for 30-40 services while officially those services require exclusive OS instances because of conflicting binary and library versions. Since we went virtual we've risen in availability ratings and the management overhead has significantly dropped. We're going to expand the cluster in the next 1-2 months to 5900 job slots and 2PB of storage running on 10GbE non-blocking interconnect and expect the whole thing to be a real nice boost in our computing capacity and availability.
Of course our model relies on the fact that CERN doesn't use multi core jobs, but instead splits the computational tasks into subtasks each working quite nicely on a single core therefore allowing such virtualization to work perfectly. For MPI applications one would just have to use different kind of VM configs and possibly auto-deploy based on job requirements, but I don't see that you'd lose much unless your each job requires a full hardnode or even multiple of them. And actually even in those cases running one VM per HN might not be a bad idea for software maintenance as well as starting from a fresh clean VM every time a job starts if you really want to (we've not done an auto-deploy system as our jobs all use the same requirements, but it's not hard considering a VM deploy on OpenVZ usually takes less than a minute from image to running VM).
Or you could use containers like OpenVZ assuming your workloads are Linux based. In which case the access to local disk is native and you're limited to what you'd be getting from the physical node also. This way you can virtualize for example a hardnode with loads of SATA disks for biggers storage and some SSD's for database and just bind_mount the databases from the SSD drives. Gives you the benefit of hybrid setup of big files and fast database. Also OpenVZ has close to zero overhead while providing excellent protection against single VM idiocy.
We're operating our whole HPC cluster on OpenVZ and managing it using OpenNode. The compute nodes have 24 cores and run 27 VM's and though we used to run the jobs in single hardnodes and did lose some nodes due to some bad behaving jobs that ate all the RAM this is no longer true because those bad jobs would be killed inside their VM once they exhaust the resources keeping everything else nicely running. We also operate our central nodes for job scheduling, WAN scheduling of jobs in Grid etc there and those employ databases (basically MySQL for state information and the main load comes from SQL). We used to run them on KVM virtualization, but that was extremely slow due to lots of IOPS from the DB. Ever since we moved to OpenVZ the bottleneck disappeared (at least beyond the virtualization, we're still somewhat limited by the underlying hardware as we've not yet implemented SSD's for those). I've also seen network performance tests and OpenVZ can do 10G connectivity close to line speed while you'd be hard pressed to get the same from KVM, Xen or VMWare.
Of course if you need to run Windows or any other stuff you're shit out of luck on those features and have to rely on full virtualization where indeed you have to seriously consider the workload as any kind of I/O (disk and net) has serious overhead and needs to be planned for accordingly.
You do know, that you can get basically all of Gutenberg Project books on iBook store for free, right? Just the same place where you can get the newer books with still working copyright also, but have to pay for them. I've got about 40 or so books bought in the last year of which about 3-4 are the free gutenberg classics while the rest are newer books. I think I've already saved close to the price of the iPad by buying them in the iBook store instead of paper books. I usually save 7-8 eur per book, sometimes way more in comparison to the local brick and mortar stores. Also, I can get close to any book I want unlike the local bookstore contents that is highly biased.
You do know that self publishing authors get 70%, right? They set the price, Apple takes 30% cut and the rest goes to the entity publishing on iBook store. You could write a book and publish it. If you set the price at $0.99 or $99 it's up to you, you'll still get 70% of it. I'm guessing the limits on what you can buy from iBook store are quite relaxed. Probably graphic porn being the only one possibly rejected, I doubt other books would be.
Slashdot Mirror
GE, 3M, IBM, Wal-mart, Nike, Exxon, AT&T, J&J and most of the finance industry disagree with you. What was your metric for "successful" again? Smartphone production or something?
How about business metrics like:
Market Cap: Apple (546B), Exxon (400B), AT&T (209B), WalMart (236B), GE (220B) etc etc. Apple is the largest US company and amongst the top of the world
Annual profit: Apple (25.9B), Exxon (41B), WMT (15.7B), AT&T (4B) etc etc
With the exception of Exxon Mobile Apple's the most profitable company with largest market cap and one of the highest growth rates in both market cap and profits. I could have also listed revenue here where again Apple would be one of the top.
So I think you should think before you post about success or not. For any investor Apple has been THE success story and it's not over.
Or you had a jailbroken phone with Cydia installer and installed tons of apps through there...
Did you even see the keynote? The reason Apple unveiled the phone in January and sent on sale in June was said by Steve during the keynote. He said the regulations for a new company to start selling a new phone required them to submit the phone for a long 3-4 month review by FCC or what not. However they didn't want the phone details to be leaked this way so they made the keynote introducing the whole phone and THEN submitted it to FCC when everyone already knew about it.
Except Apple made the iPad first, then when they had gotten the touch interface and screen to work decently they decided that they can make a phone too and decided to pull that first. Interesting that they could do ALL of that in a measly four months.
Are you keeping your apps open all the time? I found one of the major battery hoggers to be background apps and especially those you'd never suspect (like facebook). Especially bad is Viber. If you get a message in Viber you better read it and kill the bloody thing fast as otherwise it'll run your battery dry and your phone hot in under an hour. And I've seen that on multiple phones so it's a software issue of Viber. Try to see what happens if you after exiting the apps always kill them. Might be in for a surprise in battery life. And after you've confirmed that you can start to look which ones are the idiots hogging all the juice by keeping some running and killing others...
Wouldn't you be better off to use a tablet for that? Firstly the bigger screen makes it more comfortable, but you'd also not have to rush the battery as an iPad for example can last ca 10-11h playing movies so a 4h commute would leave you with a good 6h of work time left on the tablet possibly as you don't use it the whole time you'll have plenty even to watch another movie in the evening in bed :)
Seriously? You'd be satisfied with a phone that lasts a full day? I just recently was in supermarket reading e-mail in the checkout queue and a guy was speaking with the cashier about phones and the cashier asked if he's doing something wrong because he bought a smartphone (looked like a HTC or smth) and it lasts at best a full day even if he just calls. The other guy who seemed to be a phone guy or smth said that's normal and that his lasts approximately a day if all goes well, but if he uses it more he has to plug it in and that's normal. All smartphones do that... I just shook my head and kept reading e-mails on my iPhone 4 that I had taken from the charger the previous morning at 8 AM (it was late afternoon when this happened, so ca 30h later) and it still had 76% of charge left. I use the phone quite substantially, I use it for personal hotspot at times, I play games on it and constantly browse web and facebook. I also use Waze for traffic information and that's the only bigger battery drainer due to constant GPS usage. But the phone easy lasts 2-3 days, sometimes 4 days. And it's running latest iOS 5.1.1 and is almost 2 years old (bought 2010 october).
So one of the things I think is THE main failure of Android is the phones have crappy battery and the OS doesn't seem to be optimized for really running conservatively on the battery. And I haven't tweaked the OS in any way really to sustain the phone longer (about 50% of the time I'm on 3G, not wifi). I've only turned off location services for a few things (including notifications as I don't use location fencing right now), but most notifications and location services are still used and that has only a small impact. I've found some third party apps that if left in the background do consume a lot of battery even though they shouldn't (Facebook, Waze, Viber are extreme drainers if they are in background, not while in foreground). So killing those apps after I've finished using them expanded my battery life by a day and that's an issue with the 3rd party app and possibly somewhat also the OS.
I'm not sure even Microsoft could screw this up.
Oh they can. It'll be enough if the device utterly sucks on the hardware end. For example if constant lockups like the one during presentation keep occurring or if the fans don't work properly after say 1 week of usage and it heats up to unbearable levels or if the battery lasts only the distance between two power sockets at best.
Considering the iPhone's still supported start from 3Gs onwards and that iPhone sales have grown close to exponentially I'd say the difference between the two denominators is smaller than you assume it is. But I think it's 80% of devices still in use (i.e. reporting something home through carriers for debugging purposes) and I'd not be surprised if the number is about the whole install base as the recent Apple presentation did show also OS versions of 3.x that have to be early iPhone's too, not just latest models.
I think he has a serious problem between the iPad and the couch.
That's one way to solve the overpopulation I guess...
I think the "surprise" part was only to the ATLAS controller. The fact that tidal forces affect the beam has been known since LEP times as is the water level in lake Geneva and also the TGV schedule that used to have huge electricity spikes as they "grounded" their lines and said grounding happened to reach LHC 100m below causing the orbit of the beam to jump. Hell, I've been working in the CMS experiment since 2004 and I knew about the tidal force compensation need and that's why the LHC control room guy was calmly saying he has to adjust for moon's tide as it's part of their operating procedure. Same goes to monitoring the rain quantity and snow melting to knowing what the water level is in lake Geneva so that they can compensate for that too.
Damn hit submit and then remembered your other parts of the question. With regard to safety and redundancy it's actually in favor of SpaceX. As an example from Falcon Heavy wiki: "The structural safety margins are 40% above flight loads, higher than the 25% margins of other rockets.[10]". They've designed the whole things from ground up for manned flight with extended safety margins so this is not the reason for cheapness...
And with regard to Musk and his motives for the company. Look at the recent 60 minutes interview with him. When Neil Armstrong etc claim that one shouldn't push for commercialization and NASA should do everything themselves (sounds like SLS lobbying to get the old guys out and get support without due consideration what they're doing) Elon's basically in tears. You can see that those guys were/are his heroes and motivation to do the same kind of stuff and it really hurt him. So he's not a fat CEO waiting for a fat paycheck, but this IS his vision and hope and that defines his priorities.
I think a lot of this is indeed coming from the fact that SpaceX is operating like a startup. They also have claimed a lot of the cost reduction comes from the fact that they manufacture every single component themselves in their own factory where the raw materials come in from one door and spaceships/rockets come out the other. They also re-use a lot of the tooling etc because of the design (Falcon 9 is basically 9x Falcon 1 and Falcon Heavy is basically 3x Falcon 9 with inter-exchange of fuel). Due to their relative size as a company their corporate overhead has to be an order of magnitude smaller than say Boeing or Lockheed-Martin. And their main interest is first off to get to be a major player in the space industry and to compete they have to be unique in some way and their way is the cost. I'm guessing their cross margin per launch is quite small while I'd not be surprised if others have a 100% or more cross margin. Musk's vision is to help make humans a space faring civilization with cost to orbit heavily reduced and allowing exploration of the solar system including colonization. As I've understood his hope is to retire on Mars so he's pushing with the company towards making that a reality within his lifetime (assuming he retires at 60/65 he's got 20/25 years left, it's not fully utopic).
And having worked for a private company and government contract and a large EU project I can tell you that the costs shown in a large EU project per person are far higher than you'd normally associate with the tasks done just because you can. And considering that there aren't too many players around I'd not be surprised if they've kept the launch device costs somewhat bloated because they can and it's being disrupter right now by an independent player that has different motives :)
Oh forgot to add. Space X took the risk on themselves when the did the Falcon 1 and it failed 3x in row. They bet the company on the 4th launch and it worked and have been delivering cargo to LEO and now to ISS since. The Falcon 9 works (it has had 3/3 successes) and Falcon Heavy is basically re-using a lot of the Falcon 9 material so it's likely they'll make it work too. The maiden flight of SLS is planned in 2017 while I'd assume the Falcon Heavy or it's successor might easily be providing the service in 2017 for a lot cheaper (they'll probably have ironed out the reusability part of the first stages by then).
You forget the alternative. If the public funds the development of new rocket systems by NASA and keeps using them at a cost of 5x that of the private company, then after N launches the total cost for the public is higher than had they developed it in cooperation with a private enterprise that then operates the technology under contracts at a much lower cost. As an example, Falcon 9 of Space X has lift capacity of 10 tons for a cost of $56M. The Delta-IV rocket had 22 tons and $300M so the cost per kg is 3x higher. The Ariane 5 rocket by ESA has 21 ton lift to LEO and cost of $200M. The Atlas V (earlier one) had lift on 9t at a cost of $125M.
Now if Space X can pull off Falcon Heavy (the first launch is planned already in 2013) then the planned cost of 53t to LEO is $80-125M. That gives it a per kg cost of 5.8x less than Delta-IV, 4x less than Ariane 5 etc. And that's assuming the high end of the price range. It's also a rocket that can deliver cargo to trans-lunar-orbit or even to Mars (14t to Mars, 16t to trans lunar). Why the hell would we need an SLS with 50-130t capacity with an outrageous price tag when we can just launch two Falcon 9 heavy's for the same capacity and probably less than the equivalent launch cost and if need be assemble the final inter-plantery spacecraft in orbit...
I'd like to point out that Space X did first prove they're capable of real life feats with Musk's own money. Yes the latest developments and the docking with ISS has been co-funded (or mostly funded) by NASA and they'll pay them more for additional launches, but they're getting a real deal out of it as SpaceX is doing it way cheaper than any alternative and they actually deliver already today, not far fetched promises.
So the word means perfectly what I meant. We DESERVE to die if we're still stuck only on this rock... Actually I think that's true even if we reduce the timeframe to say 1000 years.
If after 3-4 billion years we're still stuck on this rock we deserve to burn up...
I guess my irony detector is malfunctioning or you really misunderstood the post you responded to. The impressive part is measuring the actual Andromeda galaxy lateral motion that is so tiny form our perspective and timescale that doing that accurately enough in 8 years is the impressive part. The fact that NVIDIA made an animation of it realtime or not is fully besides the point.
Though on a side note I agree that GPU's are becoming crazy fast...
Virtualization in any real scale is a cluster. HPC is a cluster. There's no need to cluster a cluster, it's pointless. Virtualization has no business in HPC, but in almost every other computing aspect it makes real sense.
Yes it does make sense. Software release management. It's far more comfortable to keep a dedicated hardnode OS running and deploy user needed software VM's in containers on the hardnode as the user needs arise. The container based virtualization has close to 0 overhead in all aspects (including I/O) and allows for live migration (low priority) as well as easy deployment of new OS's. Think about 200 physical servers that have to be reinstalled over PXE asking IPMI to make sure the boot devices are right, the PXE configs etc. There are plenty of things that can go wrong in those installations even if they work 90% of the time. If however it's host based authentication and you run parallel ssh to destroy VM's and deploy new ones you can do it nicely from one single central host on all HN's at the same time without downtime at all (reserve a VM, wait until it's free and redeploy without affecting any other VM on the node). The cost of 1-2% overhead max is well worth the management advantage.
We run a 1200 core HPC cluster with 900TB of storage. Our workflows are mostly CERN analysis jobs that come through Grid so we don't really have a control on what exactly the user is doing in his/her job. We used to allow up to the number of cores of jobs in any worknode, but had issues with misbehaving jobs sometimes killing off the whole node taking the other jobs with it when some code ran amok on memory allocations or what not. We attempted various resource controls, but still at times they managed to make the node unresponsive.
Since about 1 month we run 100% virtualized OpenVZ cluster with every compute core as a VM (and we actually over-provision by 10% as a lot of jobs are data intense and not CPU intense). So far no single node has died due to bad user behavior. Single VM's have ran into OOM, but as they are virtual containers the jobs are killed inside and the VM recovers nicely allowing the next job to execute. As far as we've seen studies and tested the OpenVZ container based virtualization doesn't have the drawbacks of local disk I/O and network overheads and as it's just a container the overhead on CPU is negligible. Also what it allows us is that if the experiments declare a new needed workernode software content we can just deploy one such instance, clone it and redeploy all VM's far faster than we could reinstall the nodes. (The HN's run CentOS 6 while the WN's require SL5 for example so as long as you remain with Linux you're fine).
We also run our central scheduling servers and Grid CE's and information systems and what not in the same OpenVZ containers allowing for example to live migrate between nodes to allow for service work and allowing 24 core / 128GB ram nodes to be used for 30-40 services while officially those services require exclusive OS instances because of conflicting binary and library versions. Since we went virtual we've risen in availability ratings and the management overhead has significantly dropped. We're going to expand the cluster in the next 1-2 months to 5900 job slots and 2PB of storage running on 10GbE non-blocking interconnect and expect the whole thing to be a real nice boost in our computing capacity and availability.
Of course our model relies on the fact that CERN doesn't use multi core jobs, but instead splits the computational tasks into subtasks each working quite nicely on a single core therefore allowing such virtualization to work perfectly. For MPI applications one would just have to use different kind of VM configs and possibly auto-deploy based on job requirements, but I don't see that you'd lose much unless your each job requires a full hardnode or even multiple of them. And actually even in those cases running one VM per HN might not be a bad idea for software maintenance as well as starting from a fresh clean VM every time a job starts if you really want to (we've not done an auto-deploy system as our jobs all use the same requirements, but it's not hard considering a VM deploy on OpenVZ usually takes less than a minute from image to running VM).
Or you could use containers like OpenVZ assuming your workloads are Linux based. In which case the access to local disk is native and you're limited to what you'd be getting from the physical node also. This way you can virtualize for example a hardnode with loads of SATA disks for biggers storage and some SSD's for database and just bind_mount the databases from the SSD drives. Gives you the benefit of hybrid setup of big files and fast database. Also OpenVZ has close to zero overhead while providing excellent protection against single VM idiocy.
We're operating our whole HPC cluster on OpenVZ and managing it using OpenNode. The compute nodes have 24 cores and run 27 VM's and though we used to run the jobs in single hardnodes and did lose some nodes due to some bad behaving jobs that ate all the RAM this is no longer true because those bad jobs would be killed inside their VM once they exhaust the resources keeping everything else nicely running. We also operate our central nodes for job scheduling, WAN scheduling of jobs in Grid etc there and those employ databases (basically MySQL for state information and the main load comes from SQL). We used to run them on KVM virtualization, but that was extremely slow due to lots of IOPS from the DB. Ever since we moved to OpenVZ the bottleneck disappeared (at least beyond the virtualization, we're still somewhat limited by the underlying hardware as we've not yet implemented SSD's for those). I've also seen network performance tests and OpenVZ can do 10G connectivity close to line speed while you'd be hard pressed to get the same from KVM, Xen or VMWare.
Of course if you need to run Windows or any other stuff you're shit out of luck on those features and have to rely on full virtualization where indeed you have to seriously consider the workload as any kind of I/O (disk and net) has serious overhead and needs to be planned for accordingly.
You do know, that you can get basically all of Gutenberg Project books on iBook store for free, right? Just the same place where you can get the newer books with still working copyright also, but have to pay for them. I've got about 40 or so books bought in the last year of which about 3-4 are the free gutenberg classics while the rest are newer books. I think I've already saved close to the price of the iPad by buying them in the iBook store instead of paper books. I usually save 7-8 eur per book, sometimes way more in comparison to the local brick and mortar stores. Also, I can get close to any book I want unlike the local bookstore contents that is highly biased.
You do know that self publishing authors get 70%, right? They set the price, Apple takes 30% cut and the rest goes to the entity publishing on iBook store. You could write a book and publish it. If you set the price at $0.99 or $99 it's up to you, you'll still get 70% of it. I'm guessing the limits on what you can buy from iBook store are quite relaxed. Probably graphic porn being the only one possibly rejected, I doubt other books would be.