IC designing Software

dl324

Joined Mar 30, 2015
16,844
Windows 10 and TBD foundry.
Don't know of anyone doing any real designs on Windows.
What do you mean by design aspect?
Schematic entry, transistor layout, design style (cell place and route, full custom, symbolic), performance verification, reliability verification, manufacturability, design rule verification, connectivity verification, etc.
 

Thread Starter

AleksandarK1

Joined Aug 20, 2017
14
Don't know of anyone doing any real designs on Windows.
Schematic entry, transistor layout, design style (cell place and route, full custom, symbolic), performance verification, reliability verification, manufacturability?
Than transistor layout, design style and performance verification.
 

dl324

Joined Mar 30, 2015
16,844
Pick a foundry first. The collateral they provide will usually restrict choice.

You have some flexibility for transistor layout, but the foundry will dictate design rules and layer assignments. Not aware of any options on Windows 10, but it's been decades since I considered a non-Unix platform for design work.
 

Papabravo

Joined Feb 24, 2006
21,159
As a self described noobie you might want to consider learning to design with an FPGA. This is a Field Programmable Gate Array. At least two companies have design software that runs on Windows, they are Xilinx and Altera. You can explore different methods of design entry including schematic capture, state machine, Verilog, and VHDL. This introduction will be useful as you make the transition to full custom design tools. You should learn to crawl and walk before you try to run a marathon.
 

WBahn

Joined Mar 31, 2012
29,976
Don't know of anyone doing any real designs on Windows.

Schematic entry, transistor layout, design style (cell place and route, full custom, symbolic), performance verification, reliability verification, manufacturability, design rule verification, connectivity verification, etc.
ALL of my IC designs have been done on Windows machines -- all the way back to Win3.1. Up until we started getting into the 0.25 um scale that included all of the LVS/DRC verifications as well as layout and simulation. Now we (i.e., the company I used to work for) uses Linux boxes for those because they have no viable option but to use the decks supplied by the fab (we used to write our own decks based on the design rules, but that's just not viable any more). The schematic design, simulation, and layout are all still done on Windows machines.
 

WBahn

Joined Mar 31, 2012
29,976
Than transistor layout, design style and performance verification.
The tool we used was ICED by IC Editors. Not a whole lot of features, but it is rock-solid code and was written primarily by a few guys that were primarily interested in the computer graphics part of it. As a result, it provides blazingly fast rendering. It was also a very low cost item with licenses in the $3k range. They eventually got tired of it (after a couple decades of a quite successful business) and so released it (I think as open source, but not sure).

It really IS solid code -- we used to say that their motto was, "We shall release no code before it's perfect." That meant that new feature requests took forever to get into the release, but when they did they worked as advertised. Part of this approach is just the personalities of the designers, but part of it also reflects it's heritage in the Win3.1 days in which minor software bugs routinely crashed the computer. I think in 14 years of using ICED professionally, I experienced two crashes that were caused by the software (and both were in the Win3.1 days) and I think the program became unresponsive one time (and I think that was in the NT3 days).

We used to have customers that would come out to our place that were using the big Cadence suites at a quarter million dollars a seat and they would just be stunned at how fast and responsive our little Windows machines running a $3k editor was.

One of the downsides to our approach was that our schematic entry, simulator, layout editor, standard place and route, circuit extractor, DRC and LVS tools were all separate, written by at one point five different companies, and didn't talk to each other. So there was a lot of manual effort on the part of the engineers to make the workflow actually work. But since our entire suite of tools cost about $7k and since we could get by with about three seats worth of licences for a dozen engineers since it was rare that more than three people needed the same tool at the same time, it made for an extremely cost-effective approach. A big part of my time for a few years was developing programs to translate and tweak the files from one tool to make them easier to work with another tool and writing DOS batch files to manage the environment variables and run the various tools in a quasi user-friendly way.
 

dl324

Joined Mar 30, 2015
16,844
A big part of my time for a few years was developing programs to translate and tweak the files from one tool to make them easier to work with another tool
I worked at a company that spent $100 million plus annually on CAD tools and we still had to do this because no one company had an end-to-end solution. Some tools were developed internally because no commercial capability existed, or they weren't good enough.

One project we had a place and route tool from a company that couldn't comprehend our design rules, which were more complex than anything they had previously seen. I spent half a man year working on a program to generate an interface model that would make it work. Unfortunately for me, that was unplanned work that I had to do or the project would have been experienced a significant delay because that tool was being depended on to do a significant portion of the layout and we didn't have enough layout designers to take up the slack (doing full custom layout) without blowing up the schedule.

I saved dozens of peoples' hides and they didn't even say thank you.
 

WBahn

Joined Mar 31, 2012
29,976
We has similar unplanned stuff. Perhaps the biggest was when antenna rules started coming into play. By then we were using Calibre DRC tools on a Linux box with a $25k price tag, and it couldn't do them and they were talking about several months before they would be able to. So I sat down and spent a couple days figuring out how to trick our decade-old Win3.1 era $3k circuit extractor tool into coming with with circuits that represented the sensitive gate areas and the antenna areas at each level of fabrication and then wrote a program to walk through an identify violating nodes and then generate a script that could be used to highlight the gates that needed protection and a report indicating the metal layer at which the rules were violated. Spent a week or so writing the code for that. After I got it working and validated, I spent a couple of days trying to speed it up because it was godawful slow (but not out-of-line compared to the DRC and LVS tools, I just felt I could do better). By making a more complicated data structure that overlaid a binary search tree on top of a linked list, I got a factor of 500 speed up that, at first, I though had to have been an error because the program reported completion of a test file that had taken 1.5 hours to run in under ten seconds. But when I went to look at the results to start figuring out what was wrong, it was perfectly correct. I then ran it on the actual file that had previously taken over 12 hours to run and it completed it in just about a minute and with identical results. It worked very nicely and we continued using it while Calibre went through several fits and starts with their checker (which was admittedly more complex and covered more things) and we even used my tool to find and report errors in their tool to them. Only after their tools agreed with mine consistently did we switch over.

I definitely got a number of thank yous for that. Can't say whether I got any money for it, but I suspect it was a factor in my pay increase that year.
 

WBahn

Joined Mar 31, 2012
29,976
I worked at a company that spent $100 million plus annually on CAD tools and we still had to do this because no one company had an end-to-end solution. Some tools were developed internally because no commercial capability existed, or they weren't good enough.
From talking to some of our customers (some of whom spent that kind of money on tools for their own IC design teams) I believe it. Prior to going deep-submicron, our annual outlays for all of our tool licenses was in the order of $5k, plus we usually purchased the equivalent of a new seat for each of the tools every couple years, so still under $10k/yr. Yet we did designs in half a year or less that had first-silicon success for some of the big guys that had had teams working for three and four years and still had nothing but failures after fourth silicon.
 

dl324

Joined Mar 30, 2015
16,844
I definitely got a number of thank yous for that. Can't say whether I got any money for it, but I suspect it was a factor in my pay increase that year.
Personally, I preferred division recognition awards and the cash that went with them.

I did get a promotion after one project because of my work to find a process defect that was causing a catastrophic yield loss. I adapted it to work for another design on another project; didn't even get a thank you from them. Must have saved the company millions now that I think about it...
 

WBahn

Joined Mar 31, 2012
29,976
Personally, I preferred division recognition awards and the cash that went with them.
As a small company we didn't have anything like that. It was rare to get any kind of an individual bonus -- that was usually driven by a specific term in a specific project contract and that, in turn, was usually driven by the customer's desires and they seldom think in terms of individual incentives. But the company's goal was to contribute 25% of your annual gross pay into your SEP retirement account (no employee contribution required) if the company made enough profit. It worked out so that the lowest it was was 18% and they hit the 25% about half the time with several ~22% in there (that probably worked out to about the average). So people definitely knew that if the company was profitable, you would benefit from it. This was on top of the company paying 100% of family medical/dental/vision premiums and making maximum contributions to each employee's HSA account. Also, as an LLC with 51% of the stock being owned by about half the employees (and the rest being owned by the president), there wasn't much opportunity for the company to screw employees by playing games with the numbers -- since everyone except the office manager was an engineer, we ALL knew damn well how to do math!
 

BobTPH

Joined Jun 5, 2013
8,808
As a small company we didn't have anything like that. It was rare to get any kind of an individual bonus -- that was usually driven by a specific term in a specific project contract and that, in turn, was usually driven by the customer's desires and they seldom think in terms of individual incentives. But the company's goal was to contribute 25% of your annual gross pay into your SEP retirement account (no employee contribution required) if the company made enough profit. It worked out so that the lowest it was was 18% and they hit the 25% about half the time with several ~22% in there (that probably worked out to about the average). So people definitely knew that if the company was profitable, you would benefit from it. This was on top of the company paying 100% of family medical/dental/vision premiums and making maximum contributions to each employee's HSA account. Also, as an LLC with 51% of the stock being owned by about half the employees (and the rest being owned by the president), there wasn't much opportunity for the company to screw employees by playing games with the numbers -- since everyone except the office manager was an engineer, we ALL knew damn well how to do math!
You communist. Don't you know you are supposed to work for peanuts while the ones who supplied only the capital make all the money!

Bob
 

WBahn

Joined Mar 31, 2012
29,976
You communist. Don't you know you are supposed to work for peanuts while the ones who supplied only the capital make all the money!

Bob
I had absolutely NO problem with the money that the president, who supplied all the capital, made.

We got bought by another company that when down in flames during the .com bust and, naturally, took us with it. Our president made an offer to the bankruptcy court to take on the obligation for paying unpaid employee compensation for the Colorado employees in exchange for the equipment located in Colorado. Since the value of the equipment was a fraction of what was owed to the employees, the court agreed to it. The equipment allowed him to restart the company from the ashes. Not only did no employee lose a dime of compensation owed, they received it on time. It took nearly ten years for the company to repay the president what was owed him, without paying a penny in interest. In addition, he gave those of us that stayed controlling ownership shares in the new company outright (though with his 49% he only needed to get one person to side with him to have his way, but as far as I know everything that has ever been voted on has been unanimous, though sometimes quite a bit of discussion went on before the vote).

He didn't have to do any of that. I know that there were several companies aggressively courting him to bring him on at several times his pay before the crash. He probably would have made more in just a couple years than he did in the first decade after the restart -- and with a LOT less stress. All he would have had to do is turn his back on his employees and none of us would held anything against him because we would never have known that doing what he did was even an option (and, in fact, we didn't find out that he had taken out a loan against his house in order to pay us until several years after the fact).

It reminded me of my dad's experience with the small company he worked for in Denver during the 1965 flood. Their shop was located within easy walking distance of the S. Platte river and had over 8 ft of water in the one story building and they were shut down for a couple of months. But not one employee missed a single paycheck as the owner used his own money to pay them, not knowing if the company could recover or not.
 

BobTPH

Joined Jun 5, 2013
8,808
WBahn,

I fear that my sarcastic comment was taken wrong. I was referring to traditional companies where the employees and even the CEO do not have significant ownership in the company. I think it is great that you did well with a startup. I have had the same experience. I quit the day after we were acquired by Motorola and I received a check for my stock options. The capitalist I referred to is one who does nothing for the company other than providing the money, your president was obviously not that.

Bob
 
Top