|
2banks1swap.avi posted:Then what exactly is used now for chip design? Verilog is standard among commercial houses and VHDL among defense & space. AMD and Intel will mostly do their own thing involving massively complicated homebuilt c models that translate via a massively complicated process into massively complicated libraries of things mostly built by hand 20 years ago. AMD, until recently, hadn't actually hired anyone in a LONG time. It showed when I was there interviewing. It seemed very antagonistic and I got the feeling that they didn't want new people around. Other EE's I've talked to who were in that round have said similar things. I'm sure they've gotten better by now though.
|
# ? May 11, 2011 21:37 |
|
|
# ? Jun 3, 2024 19:05 |
|
2banks1swap.avi posted:Then what exactly is used now for chip design? VHDL is a Hardware Description Language. VLSI is Very Large Scale Integration. I'm kinda confused how you could mix those two up. If you're looking for "VLSI" work the term you want is Mask Designer. Have you used Magic or a similar tool in coursework? I've honestly got no clue what sorts of requirements there are for an entry-level job there. The basic job is taking a circuit (that somebody else designed) and turning it into a layout describing exactly what types of metals will go where on the chip. Murgos posted:AMD and Intel will mostly do their own thing involving massively complicated homebuilt c models that translate via a massively complicated process into massively complicated libraries of things mostly built by hand 20 years ago.
|
# ? May 11, 2011 22:23 |
|
JawnV6 posted:Just taking a SWAG here or did somebody tell you this and you actually believed them? From conversations I had at both Intel and AMD during the interviewing process a few years ago I was under the impression that most design was done with System C and custom made cell libraries. Of course they are both large companies with many different projects, groups and etc... I.e. YMMV.
|
# ? May 12, 2011 02:39 |
|
JawnV6 posted:VHDL is a Hardware Description Language. VLSI is Very Large Scale Integration. I'm kinda confused how you could mix those two up. I'm a sophomore and actually just had my first real Physics w/ Calc class today. I'm by no means anywhere near actual engineering unless you mean "myself to school." I'll bug you in about a year when I'm done with core stuff and actually get there. For now, I'm just looking ahead.
|
# ? May 12, 2011 04:10 |
|
2banks1swap.avi posted:Then what exactly is used now for chip design? I had an interview with Intel for a Microprocessor design position last month - at least the team that I was interviewing with (Xeon Server processor design team) uses system verilog for design.
|
# ? May 12, 2011 13:54 |
|
I'm a long way off from doing anything of the sort anyway, but thanks. Good motivator to keep studying, though.
|
# ? May 12, 2011 15:54 |
|
Murgos posted:From conversations I had at both Intel and AMD during the interviewing process a few years ago I was under the impression that most design was done with System C and custom made cell libraries. Murgos posted:Of course they are both large companies with many different projects, groups and etc... I.e. YMMV. 2banks1swap.avi posted:I'm a sophomore and actually just had my first real Physics w/ Calc class today. I'm by no means anywhere near actual engineering unless you mean "myself to school." The basic coursework will be one or two Digital Design classes, preferably one with a lab component that lets you play with a FPGA. Then a basic Computer Architecture class using this book, and it's always that book. This covers a basic 5-stage pipeline. From their you want to either go deeper into architecture (e.g.) or go broader into parallel systems or something. Speaking of jobs, let me break it down this way. The Architect figures out what's possible and writes a spec for what a block should do. The Designers read that spec and write RTL code (Verilog/VHDL/etc.) and may implement the circuit that RTL describes. Now we have a circuit or RTL and there are two options: Automatically generated layout or custom-made. It's not worth getting into the distinction here, but there would be a team of people supporting either path. I wouldn't go gung-ho on one of those options until you're much more knowledgeable about it. I interviewed for one position and was recommended for a different team and took the offer. Best decision I ever made.
|
# ? May 12, 2011 19:05 |
|
Parallelization appeals to me, if only because as a ten year old I used to have pipedreams of "Some day I'll buy a computer that runs at 1ghz!" while my stepfather laughed about the impossibility. Pentiums had just come out, and I spent a lot of time playing Descent as a matter of fact. I also remember such daydreaming as "Man, I bet a 256-bit cpu would be better than a 32 bit that was 8 times as fast," even though it does't turn out that way. Now I know that you can only parallelize some tasks, and then only so far, but meh. Someone has to figure out what to do with all of these drat 10-core chips coming out. I've also had a long interest in the low-level engineering side behind networking, but I have basically no clue at all how to get into that, be it making the hardware or the protocols that operate over it. Am I just too early along to start asking meaningful questions? I'd be quite happy designing switches which give people transmission speeds as fast as their hard drives can match while yielding ping times just barely longer than the geodesic distance over the curvature of the earth between the host and their computer sometime in the future, but from where I sit now that's just more "...doped silicon sure is pretty cool!"
|
# ? May 12, 2011 21:00 |
|
2banks1swap.avi posted:Parallelization appeals to me... Nope, if you know how to write code and have an Nvidia graphics card in your laptop/desktop you can probably start messing around with CUDA. http://www.nvidia.com/object/cuda_home_new.html They have a full semester of lectures from multiple universities here also: http://developer.nvidia.com/cuda-training This is the software side of parallel processing though, not chip design. And you usually have at a minimum around 24-48 cores (I think). Also this seems to be coursework that is very lacking at universities currently. My masters program offers nothing in parallel processing outside of DSP courses in EE. Plinkey fucked around with this message at 21:13 on May 12, 2011 |
# ? May 12, 2011 21:11 |
|
JawnV6 posted:I mean this is at least a reasonable guess. "Homebuilt C models translating to 20 year old cell libraries" is a joke. 20 years ago was 800-600nm (and they were referred to as .80/.60 micron), you realize that 32nm is being shipped in volume? So, every gate gets re-architected every time a there is a process size change? The last time I did anything similar to that (which was a long time ago) it was all done as generic lambda ratios.
|
# ? May 12, 2011 21:37 |
|
So, not 2 weeks out of school I managed to land a job. To all those still hunting, good luck! Don't be too worried about marks -- mine aren't great, and I managed just fine. It was more about my project experience than my marks. To those in the thread who have answered questions and helped out, a big thanks!
|
# ? May 12, 2011 21:37 |
|
Plinkey posted:Nope, if you know how to write code and have an Nvidia graphics card in your laptop/desktop you can probably start messing around with CUDA.
|
# ? May 12, 2011 21:44 |
|
2banks1swap.avi posted:ATI here Then you do pure OpenCL, as long as you have something from the last two generations its fine. Well, relatively, AMD is definitely behind nVidia on the entire stability and not yanking the carpet out from under you thing... 2banks1swap.avi posted:I also remember such daydreaming as "Man, I bet a 256-bit cpu would be better than a 32 bit that was 8 times as fast," even though it does't turn out that way. Well, sandy bridge does have 256 bit registers... and you do get to do 8 operations on 32 bit data at a time... (this is so you can deplete your cache that much faster!)
|
# ? May 12, 2011 22:26 |
|
Right. Now how the hell do you get a compiler to really take advantage of that in most cases? What cases would that help in anyway?
|
# ? May 13, 2011 01:10 |
|
2banks1swap.avi posted:Right. Any time you want to do an operation on every element of a large array. You don't need a compiler to take advantage of it automatically, SSE primitives aren't hard to use and if you don't feel like doing that there are packages like Intel's IPP that have done all that work for you.
|
# ? May 13, 2011 03:31 |
|
2banks1swap.avi posted:Parallelization appeals to me, My last big parallelization effort was on a graph-parsing algorithm. The algorithm was described in a paper and I had to write the single-threaded version, then make it scale to more cores. I was able to split it up so each thread was working on a bucket that didn't involve the others, but because the buckets were oddly-sized I'd have one thread working after everyone else had finished and fell a little short once I had 32/64 cores. 2banks1swap.avi posted:Am I just too early along to start asking meaningful questions? Murgos posted:So, every gate gets re-architected every time a there is a process size change? The last time I did anything similar to that (which was a long time ago) it was all done as generic lambda ratios.
|
# ? May 13, 2011 18:56 |
I'll be interviewing next week for a control systems/instrumentation engineer job. For the past few years I've been in the electromagnetics field and am a bit unsure of myself wrt controls other than what I've learned from a class or two in college. What does a control/instrumentation engineer do in the day-to-day? I'm assuming that this is somewhat of a naive question but anything gained will be helpful.
|
|
# ? May 13, 2011 21:47 |
|
JawnV6 posted:The main thrust of a parallel algorithms course isn't going to be raw performance. It's more focused on the question "How does this algorithm scale as more cores are added?" e.g. when you have 32 cores, are you at 32x single-threaded performance or ~10x? It's easy to take advantage of a couple more compute resources, but when you start scaling up to 64 cores any tiny bits of communication flitting back and forth between threads can swamp your communication channel and grind progress to a halt and re-structuring of the problem (or underlying system) is often required. Now I want to bring up an old question I've had but had nobody to offer it to. What stops you, given a good pre-processor and "ideal" communication between the cores, from taking a single thread and chopping it up? I suppose if the ability to read 'ahead' into the thread and what it sets out to do means you could at the very least do some impressive prediction and thus work out every possibility before it comes time to spit out whatever it needs given whatever way it ends up going, at the very least. My innocence also guesses that this would probably be easier in a VM. Is this true?
|
# ? May 14, 2011 05:29 |
|
2banks1swap.avi posted:Now I want to bring up an old question I've had but had nobody to offer it to. Many tasks are serial - each step depends on information from the step before it. There's no way to parallelize such a task.
|
# ? May 14, 2011 07:53 |
|
Anime Dad posted:I'll be interviewing next week for a control systems/instrumentation engineer job. For the past few years I've been in the electromagnetics field and am a bit unsure of myself wrt controls other than what I've learned from a class or two in college. What does a control/instrumentation engineer do in the day-to-day? I'm assuming that this is somewhat of a naive question but anything gained will be helpful. Obviously it depends on where you're working, but probably things like programming plc's (and whatever else that might automate your facility) - I think most manufacturing facilities can be pretty much centrally controlled nowadays, I think they call this scada? Anyway, you'd certainly be analyzing and possibly editing technical drawings (particularly P&ID's), and you would very likely deal with a database that has all the controls for your plant. You might also have to troubleshoot things, although that tends to be the responsibility of electricians. My total career time on I&C projects has been less than a month so if anyone else would like to add anything feel free!
|
# ? May 14, 2011 09:57 |
|
For those already in the industry, how are tattoo's viewed? I'm currently working towards my ChemE degree, and contemplating a 3/4 sleeve on one arm, and a half sleeve on the other. No other visible tattoos. I'm assuming once you're in the industry, long sleeve dress shirts/etc are the norm, but could be wrong. Maybe once you hit the level, no one cares either.
|
# ? May 16, 2011 08:25 |
|
Flyboy925 posted:For those already in the industry, how are tattoo's viewed? I'm currently working towards my ChemE degree, and contemplating a 3/4 sleeve on one arm, and a half sleeve on the other. No other visible tattoos. I'm assuming once you're in the industry, long sleeve dress shirts/etc are the norm, but could be wrong. Maybe once you hit the level, no one cares either. Interested in this too seeing as I have a giant back-piece that pokes out of t-shirts around my neck. Will be getting more tattoos eventually, but most likely after I get a job.
|
# ? May 16, 2011 14:35 |
|
Flyboy925 posted:For those already in the industry, how are tattoo's viewed? I'm currently working towards my ChemE degree, and contemplating a 3/4 sleeve on one arm, and a half sleeve on the other. No other visible tattoos. I'm assuming once you're in the industry, long sleeve dress shirts/etc are the norm, but could be wrong. Maybe once you hit the level, no one cares either. Wear long sleeves and you're fine. We have a few people that have visible tattoos (mostly pretty small) and I know some other engineers that work for the government who have visible tattoos (hand/neck/sleeves)...it's not a big deal. Piercings are usually also fine. I have both ears pierced and the only time it's ever an issue is when I'm working on a plane because of FOD. But all jewelry is at that point.
|
# ? May 16, 2011 15:06 |
|
.
Thoguh fucked around with this message at 14:54 on Aug 10, 2023 |
# ? May 16, 2011 16:50 |
|
Flyboy925 posted:For those already in the industry, how are tattoo's viewed? I'm currently working towards my ChemE degree, and contemplating a 3/4 sleeve on one arm, and a half sleeve on the other. No other visible tattoos. I'm assuming once you're in the industry, long sleeve dress shirts/etc are the norm, but could be wrong. Maybe once you hit the level, no one cares either. I imagine it depends on the job. I suspect it might cause problems as far as being viewed as professional is concerned, but as long as you took steps to ensure it wasn't visible it wouldn't matter. Probably the only major hassle would be having to wear long sleeves on hot days, and it might be a real pain if you end up doing field tests. Most engineers I see wear jeans and polo shirts, although a lot of people can get away with t-shirts in really relaxed work environments.
|
# ? May 16, 2011 18:24 |
|
Flyboy925 posted:For those already in the industry, how are tattoo's viewed?
|
# ? May 17, 2011 16:26 |
|
MourningGlory posted:This brings up something I've been wondering about. In general, how conservative are engineers and the companies they work for? That's a really broad question, but to narrow it down a bit, I'll probably end up working on either the east or west coast, and I prefer small companies. I can't speak on any real experience, but every engineering professor I've taken who has experience in their field are all big conservatives. To the point of almost being racist/sexist a few times in class. I'd imagine engineering is also susceptible to conservative corporate culture. But I also live in Texas, so.
|
# ? May 17, 2011 16:36 |
|
Well I just got my letter that my application for PE license has been accepted contingent on passing the exam! Anyone here care to elaborate on their PE experiences including study strategy, time devoted, and references? I'm taking the EE electrical and electronics portion in October. There are several books like this that are expensive but seem well-regarded. Hed fucked around with this message at 17:46 on May 17, 2011 |
# ? May 17, 2011 17:20 |
|
Hed posted:Well I just got my letter that my application for PE license has been accepted contingent on passing the exam! Here's what I recommend: Buy yourself an approved calculator and a bunch of practice tests. Take a practice test and use that to see what areas you're weak in. Then study in those areas, and take another practice test. Repeat.
|
# ? May 17, 2011 22:19 |
|
Cross posting this: How bad does it look to stay at your first job for only 1-2 years? (Electrical Engineering) For better or worse i'm temporarily limited to my local area for jobs which due to the crappy area limits my salary to 6/7k lower than what i would be making if i was able to relocate. However i'd be up for (and probably wanting to) move after about a year. I've got a internship which i have a guaranteed job so if i try to move into another job after a year i can easily say "yeah i was working there before i graduated but it wasn't for me and blahblahblah". i.e. intership / entry level [1-2 year]-> applying for new jobs however i also have the option of local jobs which pay a little more with better benefits, but then my resume looks more like internship -> entry level [1-2 year] -> applying for new jobs
|
# ? May 18, 2011 04:41 |
|
Pizer posted:Cross posting this:
|
# ? May 18, 2011 10:49 |
|
Pizer posted:How bad does it look to stay at your first job for only 1-2 years? (Electrical Engineering) It's pretty normal, have you moved up/around/been promoted at all in your company after the first year or two? I had a boss tell me shortly after I got hired that "If you're in the same place and doing the same job for more than 4 years you're doing something wrong." He was referring to inside my company, but I think it still holds if you work at a smaller place with little room to move up or try new things.
|
# ? May 18, 2011 11:57 |
|
MourningGlory posted:This brings up something I've been wondering about. In general, how conservative are engineers and the companies they work for? That's a really broad question, but to narrow it down a bit, I'll probably end up working on either the east or west coast, and I prefer small companies. I work for a big pharma company in NJ, and the engineers in the R&D group is pretty liberal, the operations group is more conservative. I think it's just the same mix as the local population, and varies by company and department.
|
# ? May 18, 2011 22:25 |
|
What's the current outlook or unemployment rate for the ole Civil guys? The official outlook by the BLS and other sources is very favorable, but AFAIK the current unemployment rate for those guys is horrible. I've heard of the guys with 4.0s and internships working for free to have experience on their resumes! My heart of hearts wants to be one but the drat job market keeps scaring me into programming! What would be a good way to learn how to code while following a Civil core? I've heard about how Engineers who are good coders have it made making new solutions for engineers to use and developing in-house programs for design, so I might as well make that a goal for my education and self study. The obvious option is a CS Minor, but I'm not sure how useful it would be when being interviewed. Nevertheless, if the first thing I feel upon seeing some bridges, overpasses, water works and rail lines is how much I want to make them, I think the true choice is obvious. Here's hoping to a good recovery by or before 2014!
|
# ? May 19, 2011 03:45 |
|
Go where you have a passion. For real. @VT the freshman engineering classes offer basic programming foundations. C and other classes are available for those interested afterwards (and are required for some, but not all disciplines)
|
# ? May 19, 2011 04:06 |
|
Dead Pressed posted:Go where you have a passion. For real. Seriously. Do what you love, it makes life so much better.
|
# ? May 19, 2011 04:58 |
|
Don't do what you think the job market wants you to do. My dad has quite the opposite story of what you're saying (yes this is the one with the massive monitors I posted a while back). He had a science fair project with functional OCR back in like 1967 and IBM saw it and basically offered him a work study and then job once he finished college since the postal service was looking for something like this for mail sorting. BUT he thought EE was a fad (major now) and so went in to civil. He likes it and is good at it but there's always that wondering... Don't be like that. What you're suggesting sounds good (civil who knows how to code). CS minor won't do crap for you as a bullet point, but if you can point to the projects you've worked on and how your computer programming allowed you to model faster / better / reduce the tedium? That would be rad in an interview.
|
# ? May 19, 2011 05:09 |
|
I have my first "real-world" internship interview tomorrow at a company that makes job storage equipment (I'm a MechE sophomore). During the phone interview they kinda said that they're not really sure what the internship would be, exactly - they haven't had interns in a while, the VP of Design Engineering has only been there for like 6 weeks (he was one of the people who screened me), and basically if they hire me we'll figure out what they want from me, what I want to learn, get a team together and... go from there, I guess. So... any advice for a first real engineering internship interview? They also said there'd be a 1-hr plant tour/info session first, then a half-hour interview.
|
# ? May 20, 2011 05:32 |
|
Pianist On Strike posted:During the phone interview they kinda said that they're not really sure what the internship would be, exactly - they haven't had interns in a while
|
# ? May 20, 2011 12:14 |
|
|
# ? Jun 3, 2024 19:05 |
|
.
Thoguh fucked around with this message at 14:54 on Aug 10, 2023 |
# ? May 20, 2011 16:01 |