Hacker News new | comments | ask | show | jobs | submit login
Ask HN: Are there any other CPU designers here?
21 points by hnu0847 8 months ago | hide | past | web | favorite | 9 comments
Are there any other CPU designers here? If so, I’d be curious to hear your thoughts on the industry. I’ve been working as a CPU designer for several years and as I’ve watched the growth of the tech sector during this time, I can’t help but wonder if I should switch to software engineering. My reasons are:

1. Limited choice of employers and cities – The semiconductor industry has been consolidating over the last several years, and the trend seems likely to continue. Consequently, there are currently only a handful of tech giants designing ICs. Jobs seem limited to a few major cities. SWE jobs can be found in most large cities across a range of company sizes.

2. Lack of startup/solo opportunities – SWE seems to offer many opportunities to found/join a startup or work as an independent contractor. CPU design seems to have far fewer of these opportunities, likely due to the much higher capital requirements. Will the tech sector’s current interest in AI/ML lead to many more startups in CPU design, as it has done for software design?

3. Lack of community – A search on Meetup reveals many software-related groups in my city. No such groups seem to exist for CPU design, likely due to its closed-source nature. Many free resources exist to learn the basics of coding, but learning CPU design using free tools is not as straightforward. Additionally, there seems to be little incentive to design and fabricate an open-source CPU when cheap ones can already be purchased, whereas there are many motivations for creating open-source software.

4. Compensation – I would certainly not describe CPU designers’ compensation as poor, but I’ve never read/heard about anyone outside of executives receiving annual stock grants that exceed their base salary, whereas such things seem common for SWE’s at the largest tech companies.

Are my points above incorrect, "grass is greener" observations, or does SWE have much better career prospects than CPU design?

1) I agree, but from my perspective it has always been a very limited profession. But IMHO it is also a very admired job.

2) I agree, startup cost in this field is extremely high, with a lot of competitors.

3) Although software has a lot of community it is also very fractured. Depending on what you specialize in you might find a couple of groups in your area or none. Of course you will find some online groups regardless.

4) Yes the grass is always greener. I am sure that there are 1000 SWE who received great bonuses, but there are a million who got zip.

In conclusion, if your dream is to start your own company, then yes, switch. If your dream is to make money, and you are a super excellent SWE, and / or an average SPU designer, then switch.

A large part of it comes down to what brings you pleasure. We in the IT industry have an advantage over most. We are paid well and love our jobs. Well some of us love the work. Many people have one and not the other. I have known people who worked in waste processing plants who worked in their profession (Biology) for very poor pay. They all said 'you need to get a degree that pays well'. I have had as many friends in IT who loved the pay but absolutely hated the work. Those that hated the work were rarely any good.

Working at a job you hate because of the money is refered to as the 'Golden hand cuffs'

All your points are valid, but I think that the RISC-V workshops are creating a bit more of a community than there has been up to now. And those involved in bleeding edge CPU designs have been meeting at Hot Chips once a year. https://riscv.org/category/workshops/proceedings/

My startup is doing work in this area (more SPMD than traditional single thread CPU). We work with FPGAs to get the incredible costs of ASIC manufacture down so we can get our prototypes out there. For the specific workloads we target we can be competitive even with this technology handicap - which says a lot about the low hanging fruit that is out there.

In my opinion the industry has really stagnated, with nobody really stepping back and looking at the whole picture.

SWE certainly has more mindshare and development. But the CPU industry is seriously greenfield at this point if you are willing to try something different than stuffing more of the same onto a smaller die.

There are a lot of startups popping up these days that are designing IC's for low powered, high speed, neural net processing. Lots of different approaches they are taking. However, many of them are still in stealth, but try scouring chip designers on linkedIn, especially in the bay area and see where they are working.

Most of them have some kind of partnership with a large company that has a fab, that will allow them to produce prototypes and get some validation going.

I used to work in cpu design and I had the exact same thoughts as you. I think it boils down to the maturity of the technology and the associated high marginal cost in getting any improvements.

I switched to SWE 2 years ago and am definitely happier with my career prospects.

What steps did you have to take to make the switch? How involved with software were you while (and before) working as a CPU designer?

Lots of hiring in FPGAs right now, especially in vehicle sector. Bitmain is rocking custom ASICs. Micron processors in memory are still maturing.

I'd look into applying at a national lab in their team that looks into new/custom chips. The supremacy of TSMC has evened out the market and even Intel is no longer a sure bet long term.

I just found out about Bitmain's Sophon card, but I don't think you can buy it yet.

I'd like to explore chip design but I think that requires an electrical engineering degree which I don't have.

I wonder about the same points for a CPU Architect too.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact