This is pretty close to what I did for a long time... but then I got engaged. When you have TONS of shared passwords, and she is particularly bad at remembering any of them, 1Password is the answer.
The "shared vaults" are awesome. We can both add passwords / logins / credit cards / whatever there... and it shows up on all of our collective devices.
Has revolutionized the way I do things. Yeah: I have to trust 1Password... but the alternative is just non-functioning.
My home is full of smart stuff. My fiance has full access to that smart stuff. If she leaves... I can easily revoke her access with one (ok, two) touches of a button in the settings of my iPhone (to revoke her access to Homekit). She won't be able to do anything with my house past that point.
This has absolutely NOTHING to do with "crappy IOT security"... or any such scare mongering thing. All that's wrong here is that people don't know how their own devices work.
If you're not solely using Homekit then there is always another simple solution:
Hue bulbs that someone who has left the house can control? First: unplug the hub. The lights will still function perfectly as "dumb" lights that go on and off with the switch... but not a soul will be able to remotely control them. Want to turn them back into smart lights? Reset the hub and plug it back in and set it up as a new device... the person that's left the house will not be able to control them whatsoever.
That's pretty much the same for any smart stuff: unplug the hub first... then later if you want that capability back - just reset the Hub and set it up as a new device.
It works the same way for Caseta lighting, for Ecobee thermostats (just reset it and re-set it up), for Google Home, Alexa devices, etc.
I don't understand why people feel like they are somehow "at the mercy" of these devices. Just freaking unplug them and no one will be able to control your house... and when you have time to set them back up again - do it.
Why does every damn IOT story on here have to be followed with 1000 luddites screaming "I told you they were insecure!". This is freaking Slashdot! We're not afraid of technology here! We can talk about how to mitigate technological problems and steps to take to solve issues like these.
I started and have headed a large open source project for the last 10 years: (plug!) http://mooseframework.org/ (quick description: think open source COMSOL on steroids).
We have a few thousand users spread across the world... but only a fraction of them contribute monetarily back to the project. We are lucky though that we are based at a US national laboratory where we are able to use government programs in support of energy research to pay for our project.
Going open source was a measured decision that took at lot of time to come to. Ultimately, we decided that creating an open platform for science was better than trying to charge license fees... and instead of taking money from our users our model is to partner with them to write proposals for joint funding. That model is working out ok so far (some years better than others!).
However: we not only create an open source library... we rely on many as well. The two biggest ones that we use are libMesh ( http://libmesh.github.io/ ) and PETSc ( https://www.mcs.anl.gov/petsc/ ). In both cases we have paid for full-time developers on those libraries for pretty much the entirety of our project. Sometimes we ask them to complete certain tasks for us - but for the most part the money is given with minimal strings attached so they can maintain their software and continue to make it better (not just for us, but for everyone).
For some of the smaller libraries we use we often fund work at universities associated with those projects. Sometimes it's a small amount of money - but we try to give _something_.
Everyone that is making money (for-profit or non-profit) while using open-source software should try to fund the projects you directly rely on as much as possible. Like many other things: even a little bit goes a long way. Open-source has never meant "you should use it and not monetarily support it"... people need $ to keep going.
These machines are still "distributed memory" supercomputers. It's rare to see a true "shared memory" cluster in HPC these days.
Infiniband works off of a RDMA process (Remote Direct Memory Access) - but you wouldn't consider it to be "shared" memory (and programmers don't typically interact with the RDMA calls directly - most often still using MPI... but MPI then uses RDMA to achieve the transfer).
That said: you are correct that interconnect is one of the things that makes a supercomputer "super". The price of the interconnect can be a significan percentage of the purchase price of the machine. The number of network cards, number of switches (and hence topology) and length of cabling all makes a difference in the price... and the performance.
Yep! CPUs are definitely easier to program and sometimes GPUs are exactly wrong for certain workloads.
BTW: The current #1 (which will surely be supplanted in the soon to be refreshed Top500) is an all "CPU" machine (somewhat close to what Intel Phi was): https://www.top500.org/system/...
10M actual "CPU" cores. But - they are clocked lower and of quite a bit different architecture from your normal Xeon...
ARM's rise is definitely interesting because it should give us another option for good flops/watt while remaining simple to program.
The flops/socket is still better than BlueGene procs do - and I suspect the flops/watt will be a LOT better than the Xeon system you pointed out.
An exascale computer can't simply use 10M Xeons... you would need to build a small nuclear reactor next to it to power it. And while GPUs are useful for generating flops... not all workloads map well to them. These cores are general purpose: they can run anything a Xeon can run... but should use a lot less power.
With the death of Intel Phi... the HPC community really only has GPUs to offer good flops/watt. The problem with that? Not all workloads map well to GPUs and you often can't rewrite millions of dollars of software that doesn't use GPUs.
ARM offers another alternative: it can run anything an x86 processor can at better flops/watt.
The rise of ARM in HPC is _definitely_ an interesting development!
You are naive. That's how you make a really crappy supercomputer.
This machine will have more than 100,000 cores. At that scale there are many things that must be carefully thought out. Even just _launching_ a job at 100k procs presents challenges (enough so that people who do it well put out press releases about it: http://mvapich.cse.ohio-state.... ). Beyond choosing the processor (obvious) here are some of the things that must be thought about / balanced:
1. Power - for machines this large you often have to make special deals with local utility companies to power it efficiently. 2. Cooling - The heat load will be immense, deciding how to cool it is incredibly important 3. Interconnect - There are many options here (although fewer than in the past). Choosing e.g. Infiniband vs Ethernet, etc. comes with different tradeoffs and can depend on what your average application will be doing (many short messages vs large messages, etc.) 4. Switching - How many switches are needed? What topology will you use (fat-tree, hypercube, etc.). It depends somewhat on how much you want to spend on switches and somewhat on what your typical application workload looks like. 5. RAM - RAM is currently incredibly expensive (thanks to cell-phones using so much of it!). How much RAM, what type, how fast can greatly tip the scales in price / performance 6. OS - Most of these machines these days run Linux - but there are many different flavors. Things get optimized all the way down to exactly which Kernel version to use - and everything is hand-tuned 7. Job Scheduler - Several options here from PBS to Slurm and proprietary vendor specific options. How good your job scheduler is can have a HUGE impact in the usability of the machine. 8. Filesystem - Most of these machines have at least two types of filesystems: "home" and "scratch"... where "home" can be something reliable - maybe even using NFS and "scratch" is typically some highspeed filesystem (Lustre, Panasas, etc.). Choosing the balance between the two is critical. Note that 100,000 processes reading/writing simultaneously can take even carefully crafted filesystems to their knees. 9. Local disk - for a long time it was in voguge to run a "diskless" system - but now "disks" are making a come back (in the way of NV-RAM). Depending on what your applications look like this can provide huge speedups.
(I'm sure I missed something - but these are the big ones)
Anyway: It's not simple. Purchasing for these machines typically takes at least a year just in the phase where you're defining the requirements and then another 6 months or so to put out bids and go through the selection process.
In case you're wondering - I do work in the national lab system, I use these machines daily and am part of procurement decisions for them...
Agreed - so instead of talking about the "7M Gallons, the horror!"... let's talk about the environmental protection controls and oversight that are going to ensure that this operation is safe.
Just looked through the article - they estimate that ~40% of the water will be evaporated - with 60% going back. So that means this is only going to "drain" 2.8M Gallons per Day... and how much of that evaporated water will fall back into the lake as rain too?
We simply can't have it both ways: we have to find some middle ground with manufacturing if we want the jobs. As long as they are using the natural resources responsibly and not polluting them or making a long-term impact... we need to allow them to do their thing.
If you want manufacturing jobs - then you have to let them do manufacturing here. Manufacturing takes water and power... no way around it.
I'm sure that the water is not so much "used" (as in it disappears)... I'm sure they have a method for returning most of it. I would be more interested in what their controls are for the re-release of that water.
I work on a console all day long too - and I think the font there matters _even more_ than in GUIs. Retina screens make a big difference on eye strain when staring at the console all day - and a nice monospaecd font (I like Monaco) can make a big different over the long haul.
That said: I don't care _that_ much about fonts in apps / on websites. But for the console I spend _all_ day looking at it... so I want it to be pleasing to look at.
It's not just math when the inputs are from the real world. You could say that humans learning is only chemistry / quantum mechanics - that doesn't make it less true that we learn.
Each Roomba (the more expensive ones anyway) use sensor data to grow an "idea" of how your house is laid out and use that to clean more efficiently. What that is that they do will be individual to each Roomba.
I call that learning: taking sensor data over time and forming a series of actions based on it.
Then you're basically arguing for the _owner_ being responsible... and not the manufacturer.
I think the businesses selling these things would be just fine with that.
I personally think it probably just falls under your home-owners/renters insurance. We might see a rising need to detail all robotic entities within your home when you get insurance: and your rate will be set accordingly...
As I pointed out above... high-end Roomba's from iRobot already have fairly sophisticated self-learning (the cheap ones are just random - but the high-end ones "learn" as they go). They can also do some damage if they make a mistake (example I gave above is knocking over a candle and burning down a house).
Where do draw the line? How much "personality" does a robot need to have?
It doesn't have to be a "humanoid" robot that is capable of many tasks: many households today already have "robots" of varying degree that can do damage based on their AI.
One simple example is Roomba vacuums. Based on their sensors/AI they vacuum your house... but they could mess up and bump into a table... knocking over a candle and burning the house down.
So who's liable? Is it iRobot because they made it? The owners because they should have been overseeing it? Or the robot itself because it's AI/sensors/training wasn't good enough?
Similarly: lawn-mowing robots are becoming popular these days. I don't think I need to explain to you all the various things that can go wrong with a roaming robot with huge spinning blades underneath it...
As we roboticize various pieces of our lives we're going to continue running into this time and again... it does make sense to start asking the questions...
I might actually have a use for this. I'm part of a team developing scientific simulation software... we work a lot at the command-line and it is handy to be able to visualize solutions/inputs quickly. We have many tools for visualization - but having something built right into the command-line _might_ be useful (huge emphasis on *might*!).
Like you say: Netflix has plenty of clout and money to buy fast lanes (or make deals for them).
In _addition_... the inability for a new upstart streaming service to be able to procure the same deals (because it is small) means that Netflix can effectively shutout upcoming competition.
THAT is what net neutrality is really about. About creating a level playing field for anyone who wants to start a new service. Without it all of the big players can collude to lock everyone in again (just like we have been with Cable/NBC/CBS/Disney/Etc.).
Netflix _used_ to be the "plucky upstart"... but it is very much a media juggxrnaught these days!
I did look into Backblaze a few years ago. At the time it seemed like they made it difficult to select just _one_ directory to backup... it _really_ wanted to backup your whole computer.
I use them for personal backup... this is the first I've heard of this! I just went and searched my email and didn't find anything.
This is definitely a shame! One of the things I liked most about their service is that it was easy to setup with black/white lists on what to back up. I really only wanted a backup for my photography hobby... everything else is backed up fine via Time Machine (and I rotate a drive offsite). Crashplan dealt well with this.
Anyway - nothing to do about it now. I'll start shopping around...
I'm in my 30s and have already had a large amount of success by having a little bit of math, computer science and engineering knowledge. I've received many awards for my work (even one from President Obama at the White House)... but I'm incredibly intimidated by my peers who all specialized in either math OR computer science OR engineering. While I'm always able to put the pieces together in a novel way... which solves interesting problems... I always feel out of my depth when it comes to conversation.
I'm currently back at school doing a PhD in yet another interdisciplinary field: Computational Science and Engineering. But this time I'm specializing in applying it to nuclear energy production. It feels good to specialize a bit and really learn something about _one_ field in particular. I still won't be the world's greatest nuclear engineer... but at least I can hold my own in conversations now.
In addition to just feeling like I don't know much I must admit that publishing is always difficult. Journals tend to be very specialized and deciding where to send my papers or even what audience to target can be tough.
That said: there are not too many generalists out there, so I know that my interdisciplinary skillset will always be valuable... you just have to push past that feeling of knowing "nothing"
Slashdot Mirror
This is pretty close to what I did for a long time... but then I got engaged. When you have TONS of shared passwords, and she is particularly bad at remembering any of them, 1Password is the answer.
The "shared vaults" are awesome. We can both add passwords / logins / credit cards / whatever there... and it shows up on all of our collective devices.
Has revolutionized the way I do things. Yeah: I have to trust 1Password... but the alternative is just non-functioning.
Why is Slashdot suddenly full of luddites?
My home is full of smart stuff. My fiance has full access to that smart stuff. If she leaves... I can easily revoke her access with one (ok, two) touches of a button in the settings of my iPhone (to revoke her access to Homekit). She won't be able to do anything with my house past that point.
This has absolutely NOTHING to do with "crappy IOT security"... or any such scare mongering thing. All that's wrong here is that people don't know how their own devices work.
If you're not solely using Homekit then there is always another simple solution:
Hue bulbs that someone who has left the house can control? First: unplug the hub. The lights will still function perfectly as "dumb" lights that go on and off with the switch... but not a soul will be able to remotely control them. Want to turn them back into smart lights? Reset the hub and plug it back in and set it up as a new device... the person that's left the house will not be able to control them whatsoever.
That's pretty much the same for any smart stuff: unplug the hub first... then later if you want that capability back - just reset the Hub and set it up as a new device.
It works the same way for Caseta lighting, for Ecobee thermostats (just reset it and re-set it up), for Google Home, Alexa devices, etc.
I don't understand why people feel like they are somehow "at the mercy" of these devices. Just freaking unplug them and no one will be able to control your house... and when you have time to set them back up again - do it.
Why does every damn IOT story on here have to be followed with 1000 luddites screaming "I told you they were insecure!". This is freaking Slashdot! We're not afraid of technology here! We can talk about how to mitigate technological problems and steps to take to solve issues like these.
Jebus!
I started and have headed a large open source project for the last 10 years: (plug!) http://mooseframework.org/ (quick description: think open source COMSOL on steroids).
We have a few thousand users spread across the world... but only a fraction of them contribute monetarily back to the project. We are lucky though that we are based at a US national laboratory where we are able to use government programs in support of energy research to pay for our project.
Going open source was a measured decision that took at lot of time to come to. Ultimately, we decided that creating an open platform for science was better than trying to charge license fees... and instead of taking money from our users our model is to partner with them to write proposals for joint funding. That model is working out ok so far (some years better than others!).
However: we not only create an open source library... we rely on many as well. The two biggest ones that we use are libMesh ( http://libmesh.github.io/ ) and PETSc ( https://www.mcs.anl.gov/petsc/ ). In both cases we have paid for full-time developers on those libraries for pretty much the entirety of our project. Sometimes we ask them to complete certain tasks for us - but for the most part the money is given with minimal strings attached so they can maintain their software and continue to make it better (not just for us, but for everyone).
For some of the smaller libraries we use we often fund work at universities associated with those projects. Sometimes it's a small amount of money - but we try to give _something_.
Everyone that is making money (for-profit or non-profit) while using open-source software should try to fund the projects you directly rely on as much as possible. Like many other things: even a little bit goes a long way. Open-source has never meant "you should use it and not monetarily support it"... people need $ to keep going.
Sounds good - except AMD Epyc is 32 cores / socket today... and 64 cores / socket next year.
Personally: I would love a 128 core, dual-socket AMD Epyc based Mac Pro next year...
That said: I do still think that an ARM based Mac Pro would be fun... but hopefully we'll see chips with more than 28 cores in them...
These machines are still "distributed memory" supercomputers. It's rare to see a true "shared memory" cluster in HPC these days.
Infiniband works off of a RDMA process (Remote Direct Memory Access) - but you wouldn't consider it to be "shared" memory (and programmers don't typically interact with the RDMA calls directly - most often still using MPI... but MPI then uses RDMA to achieve the transfer).
That said: you are correct that interconnect is one of the things that makes a supercomputer "super". The price of the interconnect can be a significan percentage of the purchase price of the machine. The number of network cards, number of switches (and hence topology) and length of cabling all makes a difference in the price... and the performance.
Yep! CPUs are definitely easier to program and sometimes GPUs are exactly wrong for certain workloads.
BTW: The current #1 (which will surely be supplanted in the soon to be refreshed Top500) is an all "CPU" machine (somewhat close to what Intel Phi was): https://www.top500.org/system/...
10M actual "CPU" cores. But - they are clocked lower and of quite a bit different architecture from your normal Xeon...
ARM's rise is definitely interesting because it should give us another option for good flops/watt while remaining simple to program.
The flops/socket is still better than BlueGene procs do - and I suspect the flops/watt will be a LOT better than the Xeon system you pointed out.
An exascale computer can't simply use 10M Xeons... you would need to build a small nuclear reactor next to it to power it. And while GPUs are useful for generating flops... not all workloads map well to them. These cores are general purpose: they can run anything a Xeon can run... but should use a lot less power.
Well - ease of programming for one thing.
With the death of Intel Phi... the HPC community really only has GPUs to offer good flops/watt. The problem with that? Not all workloads map well to GPUs and you often can't rewrite millions of dollars of software that doesn't use GPUs.
ARM offers another alternative: it can run anything an x86 processor can at better flops/watt.
The rise of ARM in HPC is _definitely_ an interesting development!
You are naive. That's how you make a really crappy supercomputer.
This machine will have more than 100,000 cores. At that scale there are many things that must be carefully thought out. Even just _launching_ a job at 100k procs presents challenges (enough so that people who do it well put out press releases about it: http://mvapich.cse.ohio-state.... ). Beyond choosing the processor (obvious) here are some of the things that must be thought about / balanced:
1. Power - for machines this large you often have to make special deals with local utility companies to power it efficiently.
2. Cooling - The heat load will be immense, deciding how to cool it is incredibly important
3. Interconnect - There are many options here (although fewer than in the past). Choosing e.g. Infiniband vs Ethernet, etc. comes with different tradeoffs and can depend on what your average application will be doing (many short messages vs large messages, etc.)
4. Switching - How many switches are needed? What topology will you use (fat-tree, hypercube, etc.). It depends somewhat on how much you want to spend on switches and somewhat on what your typical application workload looks like.
5. RAM - RAM is currently incredibly expensive (thanks to cell-phones using so much of it!). How much RAM, what type, how fast can greatly tip the scales in price / performance
6. OS - Most of these machines these days run Linux - but there are many different flavors. Things get optimized all the way down to exactly which Kernel version to use - and everything is hand-tuned
7. Job Scheduler - Several options here from PBS to Slurm and proprietary vendor specific options. How good your job scheduler is can have a HUGE impact in the usability of the machine.
8. Filesystem - Most of these machines have at least two types of filesystems: "home" and "scratch"... where "home" can be something reliable - maybe even using NFS and "scratch" is typically some highspeed filesystem (Lustre, Panasas, etc.). Choosing the balance between the two is critical. Note that 100,000 processes reading/writing simultaneously can take even carefully crafted filesystems to their knees.
9. Local disk - for a long time it was in voguge to run a "diskless" system - but now "disks" are making a come back (in the way of NV-RAM). Depending on what your applications look like this can provide huge speedups.
(I'm sure I missed something - but these are the big ones)
Anyway: It's not simple. Purchasing for these machines typically takes at least a year just in the phase where you're defining the requirements and then another 6 months or so to put out bids and go through the selection process.
In case you're wondering - I do work in the national lab system, I use these machines daily and am part of procurement decisions for them...
Lol - did you slip a copypasta in there about "docker"?
Agreed - so instead of talking about the "7M Gallons, the horror!"... let's talk about the environmental protection controls and oversight that are going to ensure that this operation is safe.
Just looked through the article - they estimate that ~40% of the water will be evaporated - with 60% going back. So that means this is only going to "drain" 2.8M Gallons per Day... and how much of that evaporated water will fall back into the lake as rain too?
We simply can't have it both ways: we have to find some middle ground with manufacturing if we want the jobs. As long as they are using the natural resources responsibly and not polluting them or making a long-term impact... we need to allow them to do their thing.
If you want manufacturing jobs - then you have to let them do manufacturing here. Manufacturing takes water and power... no way around it.
I'm sure that the water is not so much "used" (as in it disappears)... I'm sure they have a method for returning most of it. I would be more interested in what their controls are for the re-release of that water.
I work on a console all day long too - and I think the font there matters _even more_ than in GUIs. Retina screens make a big difference on eye strain when staring at the console all day - and a nice monospaecd font (I like Monaco) can make a big different over the long haul.
That said: I don't care _that_ much about fonts in apps / on websites. But for the console I spend _all_ day looking at it... so I want it to be pleasing to look at.
It's not just math when the inputs are from the real world. You could say that humans learning is only chemistry / quantum mechanics - that doesn't make it less true that we learn.
Each Roomba (the more expensive ones anyway) use sensor data to grow an "idea" of how your house is laid out and use that to clean more efficiently. What that is that they do will be individual to each Roomba.
I call that learning: taking sensor data over time and forming a series of actions based on it.
Then you're basically arguing for the _owner_ being responsible... and not the manufacturer.
I think the businesses selling these things would be just fine with that.
I personally think it probably just falls under your home-owners/renters insurance. We might see a rising need to detail all robotic entities within your home when you get insurance: and your rate will be set accordingly...
As I pointed out above... high-end Roomba's from iRobot already have fairly sophisticated self-learning (the cheap ones are just random - but the high-end ones "learn" as they go). They can also do some damage if they make a mistake (example I gave above is knocking over a candle and burning down a house).
Where do draw the line? How much "personality" does a robot need to have?
It doesn't have to be a "humanoid" robot that is capable of many tasks: many households today already have "robots" of varying degree that can do damage based on their AI.
One simple example is Roomba vacuums. Based on their sensors/AI they vacuum your house... but they could mess up and bump into a table... knocking over a candle and burning the house down.
So who's liable? Is it iRobot because they made it? The owners because they should have been overseeing it? Or the robot itself because it's AI/sensors/training wasn't good enough?
Similarly: lawn-mowing robots are becoming popular these days. I don't think I need to explain to you all the various things that can go wrong with a roaming robot with huge spinning blades underneath it...
As we roboticize various pieces of our lives we're going to continue running into this time and again... it does make sense to start asking the questions...
I might actually have a use for this. I'm part of a team developing scientific simulation software... we work a lot at the command-line and it is handy to be able to visualize solutions/inputs quickly. We have many tools for visualization - but having something built right into the command-line _might_ be useful (huge emphasis on *might*!).
I'll definitely check it out...
Funny - I'm nearly the opposite... I don't discriminate against any sport: I like them all for the same reason (they're all sports!).
Even with some kids playing tag... I would be yelling and routing for the underdog :-)
Nope - look again.
If you look closely at the waffle fins you can see the slightly different perspective of each camera.
They really were two different feeds... just perfectly choreographed.
It's actually even more insidious than that.
Like you say: Netflix has plenty of clout and money to buy fast lanes (or make deals for them).
In _addition_... the inability for a new upstart streaming service to be able to procure the same deals (because it is small) means that Netflix can effectively shutout upcoming competition.
THAT is what net neutrality is really about. About creating a level playing field for anyone who wants to start a new service. Without it all of the big players can collude to lock everyone in again (just like we have been with Cable/NBC/CBS/Disney/Etc.).
Netflix _used_ to be the "plucky upstart"... but it is very much a media juggxrnaught these days!
I did look into Backblaze a few years ago. At the time it seemed like they made it difficult to select just _one_ directory to backup... it _really_ wanted to backup your whole computer.
Has that changed?
I use them for personal backup... this is the first I've heard of this! I just went and searched my email and didn't find anything.
This is definitely a shame! One of the things I liked most about their service is that it was easy to setup with black/white lists on what to back up. I really only wanted a backup for my photography hobby... everything else is backed up fine via Time Machine (and I rotate a drive offsite). Crashplan dealt well with this.
Anyway - nothing to do about it now. I'll start shopping around...
I'm in my 30s and have already had a large amount of success by having a little bit of math, computer science and engineering knowledge. I've received many awards for my work (even one from President Obama at the White House)... but I'm incredibly intimidated by my peers who all specialized in either math OR computer science OR engineering. While I'm always able to put the pieces together in a novel way... which solves interesting problems... I always feel out of my depth when it comes to conversation.
I'm currently back at school doing a PhD in yet another interdisciplinary field: Computational Science and Engineering. But this time I'm specializing in applying it to nuclear energy production. It feels good to specialize a bit and really learn something about _one_ field in particular. I still won't be the world's greatest nuclear engineer... but at least I can hold my own in conversations now.
In addition to just feeling like I don't know much I must admit that publishing is always difficult. Journals tend to be very specialized and deciding where to send my papers or even what audience to target can be tough.
That said: there are not too many generalists out there, so I know that my interdisciplinary skillset will always be valuable... you just have to push past that feeling of knowing "nothing"