|
yeah it's not galvanic but data and clock are AC coupled also you can do PCIe over fiber
|
# ? Mar 11, 2016 18:18 |
|
|
# ? May 16, 2024 17:07 |
|
movax posted:also you can do PCIe over fiber
|
# ? Mar 11, 2016 18:56 |
|
i am personally interested in learning how to do a pcie but it is almost certainly outside the scope of this project
|
# ? Mar 11, 2016 19:05 |
|
Bloody posted:i am personally interested in learning how to do a pcie but it is almost certainly outside the scope of this project on my "stupid waste of time project list" is still sometime to make an isa card that does something silly
|
# ? Mar 12, 2016 04:10 |
|
that's all thunderbolt is
|
# ? Mar 12, 2016 09:53 |
|
hobbesmaster posted:that's all thunderbolt is Yeah but now you are graphics card can be a mile away
|
# ? Mar 12, 2016 19:12 |
|
Captain Foo posted:Yeah but now you are graphics card can be a mile away worked for the lisp machine vendors (that way the giant cabinet with all the fans was far far away) I seriously skipped buying a "new" Symbolics XL1200 when I had the chance in '98 because it draws 1500W and has the fans you'd expect from that, now I kick myself for it
|
# ? Mar 13, 2016 05:46 |
|
question for fpga dudes have a 14-bit parallel video interface into 7-series fabric that will come in at most 640x512, 30fps. data pipeline is this interface, axi to arm core, and then stream over ethernet (zynq-7000) 1) simple parallel interface that AXI DMAs data into the DDR from the ARM controller, no additional frame buffer (or use the DDR that the CPU is using and steal a few megabytes) 1a) I bet it'll be annoying from a sw pov to steal that chunk of address space from linux 2) simple parallel interface + a SRAM or SDRAM frame buffer in FPGA land (32Mbit maybe, so roughly 4 frames), and then DMA into DDR for the CPU to do stuff any canonical implementations or app notes to steal from?
|
# ? Mar 18, 2016 00:48 |
|
re 1a, that came up at work and apparently a viable solution (for an embedded system with known HW) is to pass a parameter to the Linux kernel via your boot loader that makes it use only X bytes of physical memory instead of the default which is to use all of it. you can then do w/e you like with the extra mem above the region Linux occupies we're planning on using this with x86 Linux but I can't think of a reason why it wouldn't also work on zynq if you don't do this and you need a rly large dma buffer you'll have to implement scatter gather dma because the Linux kernel will not let you allocate huge amounts of contiguous physical ram
|
# ? Mar 19, 2016 00:13 |
|
BobHoward posted:re 1a, that came up at work and apparently a viable solution (for an embedded system with known HW) is to pass a parameter to the Linux kernel via your boot loader that makes it use only X bytes of physical memory instead of the default which is to use all of it. you can then do w/e you like with the extra mem above the region Linux occupies hmm yeah, I recall that kernel arg. raw video bandwidth is 150 mbit per second, 14-bits per pixel -- wonder how rough that'll be fighting the DDR controller for priority vs. the cpu; latency doesn't matter so putting an additional 32Mb framebuffer seems reasonable -- got to see how much fabric a MIG instance takes up if not using SRAM
|
# ? Mar 19, 2016 07:22 |
|
What's up everyone? Sorry if this is the wrong place, but I saw you guys were talking about PLCs a few pages back. I'm trying to get a job working with PLCs. I have quite a few questions about the field. If anyone could tell me even anything about it I would be extremely grateful! I've been going to a community college to get a certificate as an Automated Manufacturing Technician. I don't know what that even necessarily means, but I was introduced to PLCs in the process (in addition to a bunch of classes for remedial high school students/..army vets, sorry. Love yall.), but PLCs definitely caught my interest. I like programming, but I don't like "programming" I have a BA (in Philosophy, lol) and I'm confident I'm the only person in maybe the entire college that can actually make the Rockwell RSLogix student training programs (not mine, but example https://www.youtube.com/watch?v=WZc4Jn5NyXg). I plan to provide a 3-5 minute video of all my programs. It's really not that hard, but I swear I have never been around a bigger bunch of Can't Do's than these chuckle heads. Anyway, I have a few questions! What I'm asking is, what can I do with this? What am I expecting to find in the work force? What is the ceiling? Is there anything dangerous? What is the danger? (I'm starting a family) Can I get anything from having a 4 year degree? My teacher said you can get $35/hr knowing some basic stuff. Can anyone confirm/deny? I really, really don't know even a little about any of this. I'm extremely grateful for any information. Thanks!
|
# ? Apr 8, 2016 08:13 |
|
idk that much about the scada/ics game but if you can learn that and maybe do it for a few years you should get into ICS security where you can make some serious bank if you play your cards right
|
# ? Apr 8, 2016 08:52 |
|
is the place you're getting the certificate a for-profit institution? are you using loans to pay for it?
|
# ? Apr 8, 2016 16:16 |
|
forgive me if this was posted already, but here is a guy literally janitoring individual bits with an oscilloscope: https://www.youtube.com/watch?v=9TouLD9EOnQ
|
# ? May 3, 2016 07:52 |
|
2nd year of computer engineering school down, realizing i like working at bit level, just feels more real, bought an oscilloscope, a rigol ds1054z because it seemed best bang for buck and with the 100MHz hack it should be all i need for a large amount of things for a while having to write my own usb protocol and enumeration or aprs stuff has felt really rewarding because these things no longer just seem like magic, but that i'm actually starting to understand how this poo poo all really works, and not in a hand-wavy abstracted way next semester going to take classes on embedded systems and processor and memory architecture one question i do have, is where do i look for jobs in this kind of stuff? i know i'll need to leave maine because there isn't poo poo here for anything like that except a TI and Fairchild in Portland, i think there might be one more but it got bought out recently
|
# ? May 3, 2016 15:30 |
|
kind of depends on what specifically you want to do but alllll over the place
|
# ? May 3, 2016 15:33 |
|
dream job would be something in cybernetics of rockets, but i guess i lean more towards embedded systems that use multiple sensors or ways to interact with environment? example i worked a lot of this semester on a weather station payload for a student built rocket and i enjoyed that. i think i'm looking more high level than designing processors but not interested in software development or higher, i guess embedded systems development? building a computer system for a specific purpose?
|
# ? May 3, 2016 15:42 |
|
embedded systems is the hotness in many many many places basically look at any piece of hardware and say "hey i wonder where they're headquarted" then if you like that place apply for jobs there, they have embedded people
|
# ? May 3, 2016 15:50 |
|
and also every defense contractor
|
# ? May 3, 2016 15:51 |
|
defense contractor sounds like i would enjoy it, but i don't know if i want to go back to that world after being out of the service for 6 years now i know we have an alumni that works for raytheon now for a number of years and a good chunk of the CE graduates i've talked to are ending up working at bath ironworks on navy vessels EDIT: also which do you feel are more important, a high gpa or personal projects? i have a 3.0 gpa right now but i'd rather put time into doing more projects than killing myself trying to get 4.0, what's your opinion? Apocadall fucked around with this message at 16:04 on May 3, 2016 |
# ? May 3, 2016 15:56 |
|
biw is pretty good.
|
# ? May 3, 2016 16:03 |
|
Apocadall posted:2nd year of computer engineering school down, realizing i like working at bit level, just feels more real, bought an oscilloscope, a rigol ds1054z because it seemed best bang for buck and with the 100MHz hack it should be all i need for a large amount of things for a while i do firmware for an avionics company in a town supported almost completely by regional government administration and no real "industry" to speak of. i had no idea this place existed before i saw the posting. i'm sure the there are tons of small places that are completely off the radar because they make some doodad that no one cares about locally. you might have to move if you have loftier goals, but there's probably some kind of demand for grads from your program in your area. just makes sense if you're running a tech business to do it where talent already is.
|
# ? May 3, 2016 16:44 |
|
Apocadall posted:defense contractor sounds like i would enjoy it, but i don't know if i want to go back to that world after being out of the service for 6 years now don't do it, working for defense contractors will suck out your soul and enjoyment of life ok that might be excessive and you can certainly have a fun and fulfilling time but theres lots of hazards. often very bureaucratic organizations, horrible mismanagement due to politics (splitting a big project across 400 congressional districts and 50 subcontractors is in my experience a recipe for absolutely gently caress all getting done, including me getting paid to do nothing, which while kind of a sweet gig in some ways made me want to claw my eyes out). they'll often want to hire you as a contractor (so they can terminate you without any legal fuss), be wary of election (aka layoff) years, so on and so forth also depending on the contractor you might find yourself feeling like you're stuck in a deadend job where the technology is not that exciting and cool to work on and you're not building anything that'll look good on a resume, especially for the commercial world
|
# ? May 3, 2016 18:01 |
|
i was let go from a startup on Friday, blasted out the first round of resumes yesterday, now i'm looking at my old fpga side projects during the wait to hear back reading through the RISC-V paper: http://www.eecs.berkeley.edu/~krste/papers/EECS-2016-1.pdf (which you should do, if only for the choice words about other ISA's in chapter 2) and it's awesome that they've made this whole design open and free to the world. maybe i'll implement debug hooks like x86's DR0-DR7 so i figured i'd try to get one running, buuuuut it requires Synopsys VCS. they do acknowledge it's not open and make a token request for folks to start moving it over to DPI though.
|
# ? May 3, 2016 18:17 |
|
ya risc-v is extremely ftw but i asked rjmcall about it in a couple of other threads and he said it's an academic wanking exercise because function call prolog/epilog is comically bloated or something like that also sorry about your job
|
# ? May 3, 2016 18:21 |
|
ur job, hope u find a new one soon
|
# ? May 3, 2016 20:00 |
|
Mr Dog posted:ya risc-v is extremely ftw but i asked rjmcall about it in a couple of other threads and he said it's an academic wanking exercise because function call prolog/epilog is comically bloated or something like that compared to what? aren't you saving a bunch of registers in most architectures (except for SPARC as described in the paper jawn linked)
|
# ? May 3, 2016 20:17 |
|
I mean literally you have to do a store instruction for every callee-save register in the prolog and a corresponding load instruction in the epilog. ARM at least has the STM and LDM instructions. The architecture manual suggests that the compiler call out to prolog and epilog routines to do this lol Practical implementations are probably going to exclusively use the compressed ISA so wtf is the point of the point of the regular one.
|
# ? May 3, 2016 20:29 |
|
Mr Dog posted:ya risc-v is extremely ftw but i asked rjmcall about it in a couple of other threads and he said it's an academic wanking exercise because function call prolog/epilog is comically bloated or something like that I don't really understand why designing an open ISA will ever be anything more than an academic wanking exercise. People have become accustomed to high levels of complexity and sophistication in computer chips and so designing and manufacturing a computer chip that people actually would be interested in that would use this open ISA can only be done at great cost. What's the point of opening the ISA when everything else about making a computer chip is extremely costly and proprietary? I don't get it. To me it's like being a Rolls-Royce dealer and offering free lifetime car washes to prospective buyers as a way to get them in the door. Or maybe it is like being an international standards committee whose goal is to try to get the fins on all of the nuclear missiles in various countries' arsenals to be all the same shape as a way to allow interchangeability in parts. It's kind of trivial in the face of all of the other economic barriers to being able to work on real computer chip technologies. Computer chip designing is not like software, where at least in principle anybody who has free time and owns a laptop could download the Linux source code and read it and start contributing. silence_kit fucked around with this message at 16:21 on May 4, 2016 |
# ? May 4, 2016 05:56 |
|
I am taking an undergrad logic design class and all this bit janitoring is fun as hell. Puttin things together and state machining and simulating and all that kind of stuff is really getting me stoked for future classes.
|
# ? May 8, 2016 02:01 |
|
finally have the risc-v toolchain built after some confusion on my part that left garbage all over my brew directory most tests dying around cycle 5000 on a ROM assertion, but one's running a really long time. nothing in the output file indication progress, i'll have to dig more into the toolchain to make the debug environment more familiar e:^^ yes, yes it is
|
# ? May 8, 2016 02:22 |
|
actually it's insanely bad
|
# ? May 8, 2016 02:45 |
|
Bloody posted:actually it's insanely bad the toolchains aren't complaining, but I'm not seeing reasonable results either. i just want a vcd file to open up, yet again the docs are assuming i have vcs just lying around
|
# ? May 8, 2016 21:25 |
|
like fpga toolchain people know anything about version control systems
|
# ? May 8, 2016 22:20 |
|
im not even close to the fpga point yet, this is building their emulator/simulator collaterals oh now i get the joke and hate u
|
# ? May 8, 2016 22:26 |
|
Mr Dog posted:like fpga toolchain people know anything about version control systems a joke yes but in order to get modelsim to work with cvs for a group project in school i actually had to add a bunch of stuff to the ignore file and write tcl that just blew away the existing project and readded everything because otherwise it would just merge conflict on everything when you tried to commit ise was not appreciably better at this to the point that I stole and tweaked some open source dude's makefile and just did everything outside of the gui i don't know what Actual Companies developing Actual Designs in teams of more than 4 people would do other than drink heavily and hate everyone they work with
|
# ? May 8, 2016 22:36 |
|
Storysmith posted:a joke yes but in order to get modelsim to work with cvs for a group project in school i actually had to add a bunch of stuff to the ignore file and write tcl that just blew away the existing project and readded everything because otherwise it would just merge conflict on everything when you tried to commit pretty sure it's massive stacks of arcane Tcl scripts (and drinking and hate of course)
|
# ? May 8, 2016 22:57 |
|
Storysmith posted:ise was not appreciably better at this to the point that I stole and tweaked some open source dude's makefile and just did everything outside of the gui it's this even when the team is 1 person because gently caress working without revision control, it's worse than the pain of making a scripted vivado (the successor to ise) build system (actually thats not so bad to set up if you use zero xilinx ip, but if you do use xilinx ip cores the drinking starts)
|
# ? May 8, 2016 23:40 |
|
where i can't imagine doing digital design without a GUI is when i have a fsdb of a failure and I'm working through it double clicking a signal and being taken to the assigning statement on the relevant edge needs to be that trivial to do things with any speed even with all the little headaches of "oh, i found bits [48:44] turned '0 at the start of simulation!" derailing the whole thing
|
# ? May 8, 2016 23:46 |
|
|
# ? May 16, 2024 17:07 |
|
how bad is the stuff on OpenCores?
|
# ? May 8, 2016 23:48 |