I think they have. They want to look for a certain word on very low power. How much more battery can Google save by putting 'Ok Google' into silicon?
On the "AI" part of things how much has TensorFlow changed recently? GPUs were a good stepping stone (like they were for BitCoin) but the next step in speeding up some of the basic functions is to move from GPUs to something less general.
What is the BitC^H^H^H^H AI performance difference between a 40U rack of CPUs, 40U rack of GPUs and a 40U rack of ASICs?
ASICs have always had their use (literally) but seem to have exploded into mainstream with Bitcoin. Now it seems everyone is working on their own "AI" chip, which is fancy wording for "We put these most commonly used functions in silicon". Intel is now putting FPGAs into their Xeon chips so that customers can start speeding up their workflows.
We've kind of tapped out x86 performance lately. My 6 year old laptop is still fairly competitive. I have phone 5 generations old and it's "good enough". Are companies going to now turn to ASICs to get the competitive edge?
So you don't know how to change a setting and you're complaining. *golfclap*
I spent a week trying to diagnose *why* my machine was so chatty before narrowing it down to PA. Then turning it off took no time. But having a default setting like that makes no sense.
rewrote it from scratch.
No one is claiming that it didn't need re-written. Just like I'm not claiming the old init method couldn't be improved. I take issue with what it was rewritten into. Which is the flaming pile of PulseAudio and SystemD.
launchd has been out for 12 years now. Ported to FreeBSD and running as NextBSD's init. I have yet to hear of issues like those caused by systemd. But it was also designed to do one, launch programs.
Is there a reason OSX and Windows don't spam my network with audio data? I forget which distro/version but PulseAudio had multicast on by default.
every distribution
And FreeBSD didn't yet in 2017 my audio works. It shows up as a simple device in/dev. It behaves like any other device. Doesn't spam my network with audio chatter.
As a long time internet vagabond I tried Imzy but just couldn't do it.
The software itself wasn't bad. I could see it gaining traction for a lot of stuff that doesn't quite fit Reddit or forum discussion structure. The 'Choose a profile for this community" as well as "Post Anonymously" functions were great. I'm glad to see that some other website tried the AnonymousCoward idea.
The problem was it was the mirror universe of the Voat community where after two "Don't do that. That language shouldn't be used here" messages from mods I decided Fuck That Shit I didn't want to go online and feel like I was walking on eggshells around people that couldn't handle 'outside'.
One particular argument was that they took issue with the word "Coward" when I brought up how Slashdot used "Anonymous Coward". They didn't like the 'connotation' that it bore and calling someone a "coward" for wanting to post anonymously was answered with some logic I didn't quite follow.
I've said it before but Slashdot's founders seemingly put some forethought into how to design a forum. It's not perfect but it works. "Indeed it has been said that democracy is the worst form of Government except for all those other forms that have been tried from time to time.".
Non-nesting forums only work until you hit a critical mass. Trying to have any discussion on Facebook was futile, even within private groups of educated people because of the non-nesting commenting. Then they added it but only made it 1 comment deep. Once Fark comment threads hit a certain number of people commenting it fell apart. However in domain specific areas 'old school forums' still are best. You can find a niche of a niche of a niche forum out there to discuss why your Singer XTNEH2398 sewing machine has this weird issue and there's a half chance that it'll get seen by someone that knows how to fix it. There are multiple car specific forums out there that are infinitely better than Reddit or just a generic car site.
For large sites I take issue with Reddit's "everyone gets to vote", because it leads to bandwagoning. At least Slashdot's bandwagoning is limited to -1:+5. So while stuff can swing either way it's pointless to continue to pile on more moderation. The random, distributed nature of the moderation also seems to put a low pass filter on the moderation.
For those that think it's now Overrun with racists and what not I invite you to spend a week on Voat, 4Chan or Stormfront and come back to Slashdot.
Seems like applying patches for you too all installed software using a package management system would help tremendously.
It isn't going to help on some of the machines that got hit.
If the X-RAY and MRI machines were running Linux there's a chance it would be an old 2.4 or 2.6 kernel given the time frames needed for medical certification.
I can't speak to Medical specifically but for Aerospace and On Highway vehicles the entire software stack gets locked down for certification and unless you feel like paying for certification again, it doesn't change.
I left a place years ago because the guys were really creepy like that, and I'm a dude.
I had *one* internship like that where the boss overshared everything about his 'lifestyle' life. I then went to work in corporate America. And while some times it is a bit dry I couldn't imagine that sort of discussion ever coming up. We still find plenty of non-work stuff to talk about. I talked a co-worker into trying Linux.
But I'm not the type of guy to sit around and 'bro' out about what I do behind closed doors, let alone "Hey, you know what sounds like a good idea? Consuming drugs on the job".
It sounds like these guys had a good idea in college and were never forced to grow up. There are plenty of times and places to discuss what ever you want. While trying to get work done isn't one of them.
Same. Gave me my most valuable lesson in programming
There were two groups of people, some just typed in the equations and edited the program during the test. Others build full programs. By time I got to the test it was muscle memory because I had to do dozens of test cases by hand to make sure my program was right.
Even printed out every 'step' so that I could show my work on the test.
Has someone paid for their ISO-26262 certification? That's what holding us up. I'm trying to convince my boss that $(FOSS+Pay for the cert) $(Windriver RTOS)
they usually run at a minimum of 200MHz with a few megabytes for software because they run full-blown operating systems.
No. No. Just No.
The top of the line MPC57xx is only ranges from 32 MHz to over 300 MHz. Most of the ones that are currently in production are more than likely the MPC56xx or MPC55xx line. All of which are much more reliable than the 68ks. The highest end/safest ones run lock step cores with a 3rd core that compares the output to make sure that they're both calculating the same values.
For OS' it's running a RTOS of some sort, not a 'full blown OS'. There are a few different vendors: GreenHills, WindRiver, ETAS, etc.
For compilers it's either WindRiver's Diab or Green Hills. To my knowledge GCC doesn't work on the MPC5xxx line. I've been trying to talk my boss into sponsoring a grad student to get LLVM ported so we can at least do some prototyping without paying for a license.
And not if you're going to be using the eTPU, which requires a separate compiler.
With most of the control algorithms written in Simulink and the HAL done in C or C++.
What we need is standardized and open source ECUs that handle all the basic systems needed for the car to function.
I'll be the first in line.
So to recap:
The dev boards start at ~$500+.
Theres' no opensource compiler for the chips.
There's no opensource RTOS for the chips.
A single small team *may* be able to make ECM for vehicles ~10+ years old but unless you have a lot of money to donate to a cause, a fully opensource everything for 2017 vehicles isn't going to happen.
There is no reason we couldn't have 1000+ video lessons for any given topic; each slightly different. Periodic 3-5 question quizzes would be able to tell how well students are picking up the material, and machine learning could help identify which lessons work better for each student based on billions of other student interactions and learning results.
Exactly this. Start cutting down the ones that don't work. Eventually you'll probably have half a dozen lectures that target a particular learning style.
I *know* they're always listening. If the NSA wants to hang out with me and my wife for the most part, come along. Back when I first started saying stuff like that it was tinfoil material.
The difference is that I go out of my way to not be heard or seen when I don't want to be heard or seen. Fitting in with the norm will raise less red flags than being completely off grid.
A quick profile of the process later found a stupid design decision that ate 30% the time. (On a small scale you never noticed but on the entire build it just ate time.)
And now I'm sort of miracle worker for just looking into a problem no one wanted to look into or just assumed "that's the way it is".
Slashdot Mirror
it must first figure out what it wants to do.
I think they have. They want to look for a certain word on very low power. How much more battery can Google save by putting 'Ok Google' into silicon?
On the "AI" part of things how much has TensorFlow changed recently? GPUs were a good stepping stone (like they were for BitCoin) but the next step in speeding up some of the basic functions is to move from GPUs to something less general.
What is the BitC^H^H^H^H AI performance difference between a 40U rack of CPUs, 40U rack of GPUs and a 40U rack of ASICs?
ASICs have always had their use (literally) but seem to have exploded into mainstream with Bitcoin. Now it seems everyone is working on their own "AI" chip, which is fancy wording for "We put these most commonly used functions in silicon". Intel is now putting FPGAs into their Xeon chips so that customers can start speeding up their workflows.
We've kind of tapped out x86 performance lately. My 6 year old laptop is still fairly competitive. I have phone 5 generations old and it's "good enough". Are companies going to now turn to ASICs to get the competitive edge?
Funny how you don't hear about closed source very well integrated and tightly controlled processes
launchd is opensource. Which is how it made it into FreeBSD.
Not a single statement you made is true.
You can turn off PulseAudio in make config.
Latest download on LibreOffice's website: 5.3.3
Latest FreeBSD package available: 5.3.3
Latest available stable chrome(ium)
58.0.3029.110
Latest available in FreshPorts: 58.0.3029.110
.
So you don't know how to change a setting and you're complaining. *golfclap*
I spent a week trying to diagnose *why* my machine was so chatty before narrowing it down to PA. Then turning it off took no time. But having a default setting like that makes no sense.
rewrote it from scratch.
No one is claiming that it didn't need re-written. Just like I'm not claiming the old init method couldn't be improved. I take issue with what it was rewritten into. Which is the flaming pile of PulseAudio and SystemD.
launchd has been out for 12 years now. Ported to FreeBSD and running as NextBSD's init. I have yet to hear of issues like those caused by systemd. But it was also designed to do one, launch programs.
Is there a reason OSX and Windows don't spam my network with audio data? I forget which distro/version but PulseAudio had multicast on by default.
every distribution
And FreeBSD didn't yet in 2017 my audio works. It shows up as a simple device in /dev. It behaves like any other device. Doesn't spam my network with audio chatter.
Kik is positioning themselves for cam girls (and boys) to get paid.
and OS instead of children: R! /path/to/remove/.*
https://github.com/systemd/sys...
Pottering's Response:
I am not sure I'd consider this much of a problem. Yeah, it's a UNIX pitfall, but "rm -rf /foo/.*" will work the exact same way, no?
Unrelated, I also found sound worked much easier in FreeBSD than it did in Linux with pulseaudio. I wonder who designed that trash.
Is Source Depot better than Clear Case?
As a long time internet vagabond I tried Imzy but just couldn't do it.
The software itself wasn't bad. I could see it gaining traction for a lot of stuff that doesn't quite fit Reddit or forum discussion structure. The 'Choose a profile for this community" as well as "Post Anonymously" functions were great. I'm glad to see that some other website tried the AnonymousCoward idea.
The problem was it was the mirror universe of the Voat community where after two "Don't do that. That language shouldn't be used here" messages from mods I decided Fuck That Shit I didn't want to go online and feel like I was walking on eggshells around people that couldn't handle 'outside'.
One particular argument was that they took issue with the word "Coward" when I brought up how Slashdot used "Anonymous Coward". They didn't like the 'connotation' that it bore and calling someone a "coward" for wanting to post anonymously was answered with some logic I didn't quite follow.
I've said it before but Slashdot's founders seemingly put some forethought into how to design a forum. It's not perfect but it works. "Indeed it has been said that democracy is the worst form of Government except for all those other forms that have been tried from time to time.".
Non-nesting forums only work until you hit a critical mass. Trying to have any discussion on Facebook was futile, even within private groups of educated people because of the non-nesting commenting. Then they added it but only made it 1 comment deep. Once Fark comment threads hit a certain number of people commenting it fell apart. However in domain specific areas 'old school forums' still are best. You can find a niche of a niche of a niche forum out there to discuss why your Singer XTNEH2398 sewing machine has this weird issue and there's a half chance that it'll get seen by someone that knows how to fix it. There are multiple car specific forums out there that are infinitely better than Reddit or just a generic car site.
For large sites I take issue with Reddit's "everyone gets to vote", because it leads to bandwagoning. At least Slashdot's bandwagoning is limited to -1:+5. So while stuff can swing either way it's pointless to continue to pile on more moderation. The random, distributed nature of the moderation also seems to put a low pass filter on the moderation.
For those that think it's now Overrun with racists and what not I invite you to spend a week on Voat, 4Chan or Stormfront and come back to Slashdot.
While there have been not fun times in my career, there have been quite a few things to offset that.
How is making a 8000 HP engine run on software *you wrote* not fun?
Seems like applying patches for you too all installed software using a package management system would help tremendously.
It isn't going to help on some of the machines that got hit.
If the X-RAY and MRI machines were running Linux there's a chance it would be an old 2.4 or 2.6 kernel given the time frames needed for medical certification.
I can't speak to Medical specifically but for Aerospace and On Highway vehicles the entire software stack gets locked down for certification and unless you feel like paying for certification again, it doesn't change.
I left a place years ago because the guys were really creepy like that, and I'm a dude.
I had *one* internship like that where the boss overshared everything about his 'lifestyle' life. I then went to work in corporate America. And while some times it is a bit dry I couldn't imagine that sort of discussion ever coming up. We still find plenty of non-work stuff to talk about. I talked a co-worker into trying Linux.
But I'm not the type of guy to sit around and 'bro' out about what I do behind closed doors, let alone "Hey, you know what sounds like a good idea? Consuming drugs on the job".
It sounds like these guys had a good idea in college and were never forced to grow up. There are plenty of times and places to discuss what ever you want. While trying to get work done isn't one of them.
The number of engineering problems you can solve with a calculator on a test is a much larger set than those you can solve on a test without one.
Our tests even had Maple portions of the the tests so that you could do even more difficult problems.
Same. Gave me my most valuable lesson in programming
There were two groups of people, some just typed in the equations and edited the program during the test. Others build full programs. By time I got to the test it was muscle memory because I had to do dozens of test cases by hand to make sure my program was right.
Even printed out every 'step' so that I could show my work on the test.
Last I checked the GCC didn't support the PPC VLE extensions, which makes it a non-starter for our use. Did that change?
And damn, It looks like NXP has finally done something: https://community.nxp.com/thre...
RTEMS
Has someone paid for their ISO-26262 certification? That's what holding us up. I'm trying to convince my boss that $(FOSS+Pay for the cert) $(Windriver RTOS)
minimum of 200MHz
32 MHz seems a bit lower than your 'minimum'.
they usually run at a minimum of 200MHz with a few megabytes for software because they run full-blown operating systems.
No. No. Just No.
The top of the line MPC57xx is only ranges from 32 MHz to over 300 MHz. Most of the ones that are currently in production are more than likely the MPC56xx or MPC55xx line. All of which are much more reliable than the 68ks. The highest end/safest ones run lock step cores with a 3rd core that compares the output to make sure that they're both calculating the same values.
For OS' it's running a RTOS of some sort, not a 'full blown OS'. There are a few different vendors: GreenHills, WindRiver, ETAS, etc.
For compilers it's either WindRiver's Diab or Green Hills. To my knowledge GCC doesn't work on the MPC5xxx line. I've been trying to talk my boss into sponsoring a grad student to get LLVM ported so we can at least do some prototyping without paying for a license.
And not if you're going to be using the eTPU, which requires a separate compiler.
With most of the control algorithms written in Simulink and the HAL done in C or C++.
What we need is standardized and open source ECUs that handle all the basic systems needed for the car to function.
I'll be the first in line.
So to recap:
A single small team *may* be able to make ECM for vehicles ~10+ years old but unless you have a lot of money to donate to a cause, a fully opensource everything for 2017 vehicles isn't going to happen.
There is no reason we couldn't have 1000+ video lessons for any given topic; each slightly different. Periodic 3-5 question quizzes would be able to tell how well students are picking up the material, and machine learning could help identify which lessons work better for each student based on billions of other student interactions and learning results.
Exactly this. Start cutting down the ones that don't work. Eventually you'll probably have half a dozen lectures that target a particular learning style.
you charge your device through the USB-C port
Where did I claim that? I have a business laptop with a Xeon and 64 GB of RAM. This is a Surface.
But USB-C should let you charge
Which is all nice and good. I have a 150W adapter. It's a different device and market than the Surface.
My new laptop at work (ZBook 15 G3) has USB-C. It's everything USB should have been since the beginning.
Reversible, Just Works(Tm). It'll drive 2 4k external TVs.
Laptop itself has Ethernet, VGA, 3xUSB3.0 and 2xUSB-C ports. Holds 64GB of RAM, 2xM.2 NVMe drives and 1x 2.5" drive.
The dock could still use some work. You shouldn't have to issue a white paper on how to hook up monitors (Which is still wrong, the HDMI port drives 4k just fine.).
If I *need* to do some GPU work I can plug in an external GPU. Or gigabit ethernet or any other PCIe device.
Microsoft screwed up on this one. They're releasing old hardware. I bet they could have easily charged a surface on over USB Power Delivery. It's taken us a while but USB-C is pretty damn good as far as a physical connection. And Thunderbolt 3 is equally as good of a protocol.
For most people if the 'desktop is dead' it's because USB-C/TB killed it. I just want to plug my laptop into cluster of CPUs when I'm at my desk.
It just fits with Slashdot's history of technology predictions.
iPod: The MP3 player nobody asked for.
iPhone: The phone nobody asked for.
Amazon AWS: The server service nobody asked for.
Amazon Echo: The device nobody asked for.
Tesla: The electric car nobody asked for.
I'm in my 30s and do the same thing.
I *know* they're always listening. If the NSA wants to hang out with me and my wife for the most part, come along. Back when I first started saying stuff like that it was tinfoil material.
The difference is that I go out of my way to not be heard or seen when I don't want to be heard or seen. Fitting in with the norm will raise less red flags than being completely off grid.
"Why does a build take 2 hours"?
"Because"
A quick profile of the process later found a stupid design decision that ate 30% the time. (On a small scale you never noticed but on the entire build it just ate time.)
And now I'm sort of miracle worker for just looking into a problem no one wanted to look into or just assumed "that's the way it is".
Schedule the meeting yourself.
I put a reoccurring monthly meeting on my boss' calendar. Sometimes it gets moved, sometimes it gets canceled but it's in the calendar for a reason.