Critics watch far more movies than the average person does. As a result they tend to see patterns in movies far more often than average viewers do. This means that movies that basically just do a good job in execution but don't break any new ground tend to be seen very negatively by critics but very well by the average person. It is mostly a problem of saturation.
I just did a quick look of some of the movies I like that score really badly on rotten tomatoes. Some of these are even below 10% but I still enjoy them.
Ultraviolet Resident evil movies Netflix Iron Fist The Boondock Saints 1 and 2
You can just do a search on google for movies that audiences love that critics hate and get a LOT of results.
Iron Fist and Boondock Saints I understand the least why critics hated them so much. Iron Fist is doing VERY well with regular people and was hated by critics. Boonddock saints are good action movies and I don't get what the problem with those are at all.
Oh one really important thing I forget is the Windows Subsystem for Linux.
With that I can run most linux binaries under windows without changes. This has made it far easier to work with linux systems since you have a real bash and ssh to work with. I can even compile my simulations under WSL and run them without problems.
WSL has made doing linux development much faster and easier and I look forward to the updated WSL in the april windows 10 update and integration with VS 2017.
Windows 10 runs faster on multicore processors. On a quad core i7 I get about 15% faster performance with MATLAB or my own HPC threaded applications on Windows 10 vs Windows 7 due to the newer scheduler. It was added in Windows 8 but refined a lot in Windows 10.
For my simulations I now get almost identical performance on the same hardware with Windows 10 and Linux and it used to be that Windows was always slower. This has made it easier to do testing on other systems. I mostly deploy on Linux but Windows is also a supported platform.
I also get a bit better battery life under Windows 10 compared to Windows 7 or 8.1.
With the HPC C++ project I am working on even with the Intel tools loaded and after using stuff like Vtune and a debugger I am getting about 200MB of usage with VS2017 and under VS2015 I end up with about 500MB of usage. I also do not see a bunch of children processes. At least from my experience VS2015 is faster and lower resource usage than Eclipse or PyCharm.
At least with C++ the resource usage seems pretty minimal.
The IDE is 32bit. The compile, debug, profile etc chain are 32bit and 64bit.
There is probably no reason for the IDE to be 64bit since it does not come even close to use enough memory to justify that. I have opened a few visual studio projects in 2017 and most of them don't use more than 200 MB. Resource usage so far is about half that of VS 2015.
For highly cpu and memory intensive applications I don't see how this would work. The vast majority of software runs at only a few percent of what a CPU is capable of mostly waits on IO. However for HPC applications the memory and processing usage are often tuned down to the cache line level and that pretty much requires low level access to the system.
Even more extreme are things like BLAS which is tuned to different cpu archs. Running an optimized BLAS vs unoptimized BLAS is usually 10x the performance or more and having a simulation go from a month to 10 months would not be acceptable.
Very large doses of some vitamins can actually make you very sick or significantly increase your cancer risk. The most common thing is people that think that LOTS of antioxidants are good for them. The problem is that if you take too many your body can no longer destroy cancer cells with free radicals generated by macrophages. If you take antioxidants too far you significantly increase your risk of cancer and your risk for bacterial infection.
There are also some fat soluble vitamins that are toxic if you take too much (like D).
You do need vitamins and minerals and as our technology advances we can figure out how much you truly need but for now it is still important to use moderation.
It would not take 3-6 months it would take 2-3 years.Some of these newer fields are combining knowledge of chemistry, manufacturing and computer science. It is just not that easy to teach someone to do it.
Efforts have been made to have different people do each part of it and mostly that has failed. The issues are complex enough that you really do need one mind to at least know enough about all of those that the problem can be fully developed and specialized tasks can then be divided out.
There are new fields the USA does not have yet. I went to Germany because of a research group here that is working on something that only a few other groups in the entire world are working in. Some US companies would love to hire the entire team so we could teach more people how to do what we do.
There are all kinds of advances in biotech, nanotech, quantum computing, genetic engineering etc and the fields are so complex and so specialized that it is easy to have a situation where the there are certain skills the USA just does not have yet. It takes a while for universities to catch up with the education required and meanwhile work still needs to be done. Some of the research I am working on will likely end up in the textbooks.
A heck of a lot of problems are easier to solve with people in the same room together. I have worked with some colleagues remotely trying to solve some incredibly difficult problems and what it finally took was everyone in the same room together for a few days to get it figured out.
Lots of paper, stuff written on the white board and discussions with far too many math equations but the problem was solved.
If you work for a company that makes a physical product you often have material engineers, electrical engineers, chemical engineers, programmers etc all working together and people able to be in the same place together really helps.
I see this as a separate issue. If you make it to a US university and graduate your diploma should come with a visa and a path to a greencard. There should be no H1-B or anything else. In my graduating class we had chemical engineers that had to return to their home countries in a field where unemployment is around 0.5%. They should have all been able to stay.
One company I knew took 2 years to fill a position and finally hired someone that was not qualified and is trying to train them to do the job but the problem is that they don't really know how to train someone to do the job. There where some people in other countries that could have done the job but they where never able to get them to the USA.
All H1-B visa requests should go into a pool and from that entries should be selected in descending order from the highest wages. Then the companies that are seriously trying to bring someone in that they really can't get would get the people they need. I have seen companies try to bring a single person in and where offering a LOT for the job but never won the lottery for the H1-B slot.
This seems like it would almost entirely address the current problems of H1-B being used to drive wages down. It is hard to drive wages down when the slots are essentially auctioned.
There are companies outside of the tech companies that do pull in highly qualified people with H1-B and don't screw their workers over.
I am sick and tired of the system being abused to lower wages and treat people like servants as so many of the tech companies do. Most of them where even involved in agreements with each other in silicon valley to drive wages down. The system needs to be fundamentally fixed and the companies abusing it needed to be fined MORE than what the H1-B system abuse saved them.
I know people in certain VERY new fields where american companies have more job openings than there are qualified people on earth for those jobs. These jobs would all pay well in excess of $150K/year and would not cost any american a job. It would actually create a lot more job for americans since each engineer/scientist hired usually results in more support jobs at the company (due to being able to make more products).
However due to how the H1-B system is abused it is really hard to get the people you need. I would prefer the H1-B visa system becomes a bidding system where it goes from highest to lowest paid until the slots are used up. This would ensure that those people we truly need and can't get locally are the ones brought in and not people that depress wages in the area.
The USA is working on these new fields but there are a very small number that the USA is behind on by probably 5-10 years. Eventually we will have more qualified people but it is going to be a while before the current positions can be filled.
I have had AV software have false positives on software I was compiling and it would delete it immediately. I even tried to mark that area as safe but to no avail. I ended up getting rid of the AV software since I could not get work done that way.
I have also run into AV software where a bad update went through and the software ended up attacking the OS and did quite a lot of damage in terms of downtime.
At the end of the day it is easier for me to avoid viruses than it is to deal with most AV software I have encountered and the one built into windows seems to do a good enough job. The email I get comes through gmail and it seems to do an excellent job of filtering away the spam, malware etc and the software I download from intel, mathworks, nvidia lawrence livermore labs etc is very unlikely to be infected.
From what I could see in a few different tests windows defender is about 97% and there are a few scanners that go to 99.9% but the higher the detection rate the more likely it also is to suffer from false positives and impact the system negatively while running.
I had bitdefender installed on my machine about a year ago and I was writing c++ HPC software. Everything was compiled with the Intel compiler and mkl with profile guided optimizations. Bitdefender started detecting my binaries as virus infected and deleting them. This happened a few times and I disabled it for a month and later turned it back on with newer virus definitions and the same issue kept happening. It even detected some of the binaries I had on a shared drive and deleted them also.
The false positive rate on some of these scanners is just too high.
I will just stay with windows defender since it has not interfered with any of my debugging or profiling and has never deleted the software I am compiling.
If you have the professional version windows will ask if you want to reboot and you can delay it and then keep delaying it for as long as you want. If I am in the middle of something I will normally tell it to delay for 4 hours and the system already does not check for updates at all during the normal working hours I set.
All of these issues I have run into are people using the home version for work. You can do this but it does come with drawbacks as a result.
Microsoft has been working quite hard to make windows a good development platform for linux. Between WSL and the changes to Visual Studio it has gotten pretty easy to do writing, compiling and debugging of linux software from windows.
For me this is really important since linux has never run well on this laptop. I have optimus which means I have a dedicated gpu + integrated gpu and with windows it seamlessly switches between them and everything works. Under linux there are commands to make one or the other run but it is not remotely seamless and it is really buggy. I have also run into problems with ubuntu and fedora where an update will sometimes break x entirely where the default output gets set to the device that is not activated and then having to deal with debugging that.
I also write C++ simulation software and I have found no better IDE that VisualStudio so far. With eclipse under linux once I upgraded to an SSD I sometimes had issues to compile multiple times to compile without errors about files not being found. If I compiled from the command line that never happened. Debugging is MUCH worse in eclipse vs visual studio. The worse thing though is profiling. I have no idea what happened to it on linux since I have done linux development for almost 20 years now and we used to have some of the best profilers out there but no it seems most of them just do a horrible job. Trying to profile a program that uses shared libraries in linux mostly ends up with no, poor or inconsistent results even when the program behavior is highly consistent. I ended up trying the proprietary vtune from intel and that worked great on linux and windows.
In the end it is easier to do development on windows where all the desktop type stuff works and get the software running completely correctly and debugged and then deploy it to linux servers, clusters, supercomputers etc for actual running. At this point I pretty much use windows for desktop work and linux for all the server work and the WSL system has made life much simpler.
Overall I have been screwed over FAR more commonly from a small business than a large one. What I have found is that large businesses don't screw you over on small contracts since it is not worth the time to do it. It gets them more negative press and for no reason at all. I have also found that small businesses tend to be the most abusive and deceptive in what they want to do.
I once had a contract to build an electronic voting system for a professional ethics body and they wanted the ability to tamper with the election, They where PISSED when they found out the system notified everyone voting when they tried to temper with the system. I have just never seen a large business do that.
The USA also uses limited lifetime landmines so they don't stick around after a conflict to keep killing people. I may not like landmines but I understand why they are used and having ones that self destruct is much better than ones that stay around.
Restarting them yourself is likely to be more disruptive than a reboot. When you do a reboot the system is running pre-written scripts however fast it can execute them. If you run commands yourself to do all of this then it will happen at the speed you can type stuff in. The reboot process is likely to be FAR faster and won't miss anything.
Even something like an openssl update can impact more than you would remember and missing just one long running program can leave you screwed. You could try to keep track of each library and dependency and you better not make a single mistake.
It is just much easier and safer to reboot and check. Sure you can't do that in all situations but for most machines rebooting is quite fast before all services are up and running again. Why take a risk you don't need to take?
Slashdot Mirror
Critics watch far more movies than the average person does. As a result they tend to see patterns in movies far more often than average viewers do. This means that movies that basically just do a good job in execution but don't break any new ground tend to be seen very negatively by critics but very well by the average person. It is mostly a problem of saturation.
I just did a quick look of some of the movies I like that score really badly on rotten tomatoes. Some of these are even below 10% but I still enjoy them.
Ultraviolet
Resident evil movies
Netflix Iron Fist
The Boondock Saints 1 and 2
You can just do a search on google for movies that audiences love that critics hate and get a LOT of results.
Iron Fist and Boondock Saints I understand the least why critics hated them so much. Iron Fist is doing VERY well with regular people and was hated by critics. Boonddock saints are good action movies and I don't get what the problem with those are at all.
Oh one really important thing I forget is the Windows Subsystem for Linux.
With that I can run most linux binaries under windows without changes. This has made it far easier to work with linux systems since you have a real bash and ssh to work with. I can even compile my simulations under WSL and run them without problems.
WSL has made doing linux development much faster and easier and I look forward to the updated WSL in the april windows 10 update and integration with VS 2017.
Windows 10 runs faster on multicore processors. On a quad core i7 I get about 15% faster performance with MATLAB or my own HPC threaded applications on Windows 10 vs Windows 7 due to the newer scheduler. It was added in Windows 8 but refined a lot in Windows 10.
For my simulations I now get almost identical performance on the same hardware with Windows 10 and Linux and it used to be that Windows was always slower. This has made it easier to do testing on other systems. I mostly deploy on Linux but Windows is also a supported platform.
I also get a bit better battery life under Windows 10 compared to Windows 7 or 8.1.
With the HPC C++ project I am working on even with the Intel tools loaded and after using stuff like Vtune and a debugger I am getting about 200MB of usage with VS2017 and under VS2015 I end up with about 500MB of usage. I also do not see a bunch of children processes. At least from my experience VS2015 is faster and lower resource usage than Eclipse or PyCharm.
At least with C++ the resource usage seems pretty minimal.
The IDE is 32bit. The compile, debug, profile etc chain are 32bit and 64bit.
There is probably no reason for the IDE to be 64bit since it does not come even close to use enough memory to justify that. I have opened a few visual studio projects in 2017 and most of them don't use more than 200 MB. Resource usage so far is about half that of VS 2015.
For highly cpu and memory intensive applications I don't see how this would work. The vast majority of software runs at only a few percent of what a CPU is capable of mostly waits on IO. However for HPC applications the memory and processing usage are often tuned down to the cache line level and that pretty much requires low level access to the system.
Even more extreme are things like BLAS which is tuned to different cpu archs. Running an optimized BLAS vs unoptimized BLAS is usually 10x the performance or more and having a simulation go from a month to 10 months would not be acceptable.
Very large doses of some vitamins can actually make you very sick or significantly increase your cancer risk. The most common thing is people that think that LOTS of antioxidants are good for them. The problem is that if you take too many your body can no longer destroy cancer cells with free radicals generated by macrophages. If you take antioxidants too far you significantly increase your risk of cancer and your risk for bacterial infection.
There are also some fat soluble vitamins that are toxic if you take too much (like D).
You do need vitamins and minerals and as our technology advances we can figure out how much you truly need but for now it is still important to use moderation.
It would not take 3-6 months it would take 2-3 years.Some of these newer fields are combining knowledge of chemistry, manufacturing and computer science. It is just not that easy to teach someone to do it.
Efforts have been made to have different people do each part of it and mostly that has failed. The issues are complex enough that you really do need one mind to at least know enough about all of those that the problem can be fully developed and specialized tasks can then be divided out.
There are new fields the USA does not have yet. I went to Germany because of a research group here that is working on something that only a few other groups in the entire world are working in. Some US companies would love to hire the entire team so we could teach more people how to do what we do.
There are all kinds of advances in biotech, nanotech, quantum computing, genetic engineering etc and the fields are so complex and so specialized that it is easy to have a situation where the there are certain skills the USA just does not have yet. It takes a while for universities to catch up with the education required and meanwhile work still needs to be done. Some of the research I am working on will likely end up in the textbooks.
A heck of a lot of problems are easier to solve with people in the same room together. I have worked with some colleagues remotely trying to solve some incredibly difficult problems and what it finally took was everyone in the same room together for a few days to get it figured out.
Lots of paper, stuff written on the white board and discussions with far too many math equations but the problem was solved.
If you work for a company that makes a physical product you often have material engineers, electrical engineers, chemical engineers, programmers etc all working together and people able to be in the same place together really helps.
I see this as a separate issue. If you make it to a US university and graduate your diploma should come with a visa and a path to a greencard. There should be no H1-B or anything else. In my graduating class we had chemical engineers that had to return to their home countries in a field where unemployment is around 0.5%. They should have all been able to stay.
One company I knew took 2 years to fill a position and finally hired someone that was not qualified and is trying to train them to do the job but the problem is that they don't really know how to train someone to do the job. There where some people in other countries that could have done the job but they where never able to get them to the USA.
All H1-B visa requests should go into a pool and from that entries should be selected in descending order from the highest wages. Then the companies that are seriously trying to bring someone in that they really can't get would get the people they need. I have seen companies try to bring a single person in and where offering a LOT for the job but never won the lottery for the H1-B slot.
This seems like it would almost entirely address the current problems of H1-B being used to drive wages down. It is hard to drive wages down when the slots are essentially auctioned.
There are companies outside of the tech companies that do pull in highly qualified people with H1-B and don't screw their workers over.
I am sick and tired of the system being abused to lower wages and treat people like servants as so many of the tech companies do. Most of them where even involved in agreements with each other in silicon valley to drive wages down. The system needs to be fundamentally fixed and the companies abusing it needed to be fined MORE than what the H1-B system abuse saved them.
I know people in certain VERY new fields where american companies have more job openings than there are qualified people on earth for those jobs. These jobs would all pay well in excess of $150K/year and would not cost any american a job. It would actually create a lot more job for americans since each engineer/scientist hired usually results in more support jobs at the company (due to being able to make more products).
However due to how the H1-B system is abused it is really hard to get the people you need. I would prefer the H1-B visa system becomes a bidding system where it goes from highest to lowest paid until the slots are used up. This would ensure that those people we truly need and can't get locally are the ones brought in and not people that depress wages in the area.
The USA is working on these new fields but there are a very small number that the USA is behind on by probably 5-10 years. Eventually we will have more qualified people but it is going to be a while before the current positions can be filled.
I have had AV software have false positives on software I was compiling and it would delete it immediately. I even tried to mark that area as safe but to no avail. I ended up getting rid of the AV software since I could not get work done that way.
I have also run into AV software where a bad update went through and the software ended up attacking the OS and did quite a lot of damage in terms of downtime.
At the end of the day it is easier for me to avoid viruses than it is to deal with most AV software I have encountered and the one built into windows seems to do a good enough job. The email I get comes through gmail and it seems to do an excellent job of filtering away the spam, malware etc and the software I download from intel, mathworks, nvidia lawrence livermore labs etc is very unlikely to be infected.
From what I could see in a few different tests windows defender is about 97% and there are a few scanners that go to 99.9% but the higher the detection rate the more likely it also is to suffer from false positives and impact the system negatively while running.
I had bitdefender installed on my machine about a year ago and I was writing c++ HPC software. Everything was compiled with the Intel compiler and mkl with profile guided optimizations. Bitdefender started detecting my binaries as virus infected and deleting them. This happened a few times and I disabled it for a month and later turned it back on with newer virus definitions and the same issue kept happening. It even detected some of the binaries I had on a shared drive and deleted them also.
The false positive rate on some of these scanners is just too high.
I will just stay with windows defender since it has not interfered with any of my debugging or profiling and has never deleted the software I am compiling.
For the home version it can turn on and check for updates during the day when you are at work and then go back to sleep.
If you have the professional version windows will ask if you want to reboot and you can delay it and then keep delaying it for as long as you want. If I am in the middle of something I will normally tell it to delay for 4 hours and the system already does not check for updates at all during the normal working hours I set.
All of these issues I have run into are people using the home version for work. You can do this but it does come with drawbacks as a result.
Microsoft has been working quite hard to make windows a good development platform for linux. Between WSL and the changes to Visual Studio it has gotten pretty easy to do writing, compiling and debugging of linux software from windows.
For me this is really important since linux has never run well on this laptop. I have optimus which means I have a dedicated gpu + integrated gpu and with windows it seamlessly switches between them and everything works. Under linux there are commands to make one or the other run but it is not remotely seamless and it is really buggy. I have also run into problems with ubuntu and fedora where an update will sometimes break x entirely where the default output gets set to the device that is not activated and then having to deal with debugging that.
I also write C++ simulation software and I have found no better IDE that VisualStudio so far. With eclipse under linux once I upgraded to an SSD I sometimes had issues to compile multiple times to compile without errors about files not being found. If I compiled from the command line that never happened. Debugging is MUCH worse in eclipse vs visual studio. The worse thing though is profiling. I have no idea what happened to it on linux since I have done linux development for almost 20 years now and we used to have some of the best profilers out there but no it seems most of them just do a horrible job. Trying to profile a program that uses shared libraries in linux mostly ends up with no, poor or inconsistent results even when the program behavior is highly consistent. I ended up trying the proprietary vtune from intel and that worked great on linux and windows.
In the end it is easier to do development on windows where all the desktop type stuff works and get the software running completely correctly and debugged and then deploy it to linux servers, clusters, supercomputers etc for actual running. At this point I pretty much use windows for desktop work and linux for all the server work and the WSL system has made life much simpler.
Overall I have been screwed over FAR more commonly from a small business than a large one. What I have found is that large businesses don't screw you over on small contracts since it is not worth the time to do it. It gets them more negative press and for no reason at all. I have also found that small businesses tend to be the most abusive and deceptive in what they want to do.
I once had a contract to build an electronic voting system for a professional ethics body and they wanted the ability to tamper with the election, They where PISSED when they found out the system notified everyone voting when they tried to temper with the system. I have just never seen a large business do that.
The USA also uses limited lifetime landmines so they don't stick around after a conflict to keep killing people. I may not like landmines but I understand why they are used and having ones that self destruct is much better than ones that stay around.
Restarting them yourself is likely to be more disruptive than a reboot. When you do a reboot the system is running pre-written scripts however fast it can execute them. If you run commands yourself to do all of this then it will happen at the speed you can type stuff in. The reboot process is likely to be FAR faster and won't miss anything.
Even something like an openssl update can impact more than you would remember and missing just one long running program can leave you screwed. You could try to keep track of each library and dependency and you better not make a single mistake.
It is just much easier and safer to reboot and check. Sure you can't do that in all situations but for most machines rebooting is quite fast before all services are up and running again. Why take a risk you don't need to take?