5 ms·
I wasn't aware of the connection between volume and technological progress in semiconductor manufacturing: “When you have volume orders, then you can do yield
by WithinReason 5y ago
I wasn't aware of the connection between volume and technological progress in semiconductor manufacturing:
“When you have volume orders, then you can do yield experiments, you can improve your yield, and yield is everything because that’s how you cover your costs,”
"Foundries that can juggle multiple clients and technologies can swiftly advance on their competitors. For example, TSMC’s scale allowed it to master extreme ultraviolet lithography faster than anyone else, which reduced the number of steps it took to make advanced chips and boosted the throughput of its fabs."
- amelius 5y agoIt's called Wright's law. Contrary to Moore's law, it's not slowing down.
- Nokinside 5y agoI tought Wright law (experience curve effect ) is related to repetitive manual labor.
- bsedlm 5y agoisn't chip making is extremeley repetitive? (though not manual, or maybe we could say manual but through extermely complex tools?)
- Nokinside 5y agoMachines don't learn from experience and repetition.
- krastanov 5y agoThe operators that calibrate them and pick parameters, do. And some machines have internal long-term calibration and feedback (no need to call it Machine Learning), so in a way they learn on their own too.
- zsmi 5y agoObviously you're correct about that, but those productivity gains from those sources are unlikely to be accurately predicted by Wright's Law, which is what this thread is about.
- simonh 5y agoIf you scale up production by just building more of the same machines that's right, but over time with new process nodes and increasing fab sizes there are plenty of opportunities to apply new skills to improve automation and scale effects.
- vlovich123 5y agoI thought it was a more general principle and applies almost any task. For example, the software written today is done at a larger volume and more cheaply than at any other point in history and can do more things than ever before (heck printing “hello world” or adding 2 numbers was a feat on the original machines). My understanding is that it’s a general phenomenon. The more we do something, the faster we get at it (either as a species or individuals), the more tools we accumulate to help us smoothie out any bumps, tackle larger projects more efficiently, etc.
- Nokinside 5y agoThat's general principle called increasing returns to scale and decreasing marginal cost.
- Nokinside 5y agoSemiconductor manufacturing R&D and pathfining includes many expensive dead ends. Massive volume is required to justify the gigantic upfront cost to develop a new process every 2-4 years (overlapping with previous and next process node). Economist would say that volume gives you: 1) Increasing marginal returns after massive fixed capital investments. 2) increasing returns to scale from variable investments when volumes increase.
- SavantIdiot 5y agoEvery CPU project I worked on at Intel from 1997-2010 was cancelled. That's 13 years of one senior engineering salary that never turned a profit, let alone the thousand other people on the related teams. One could argue the failed learnings helped guardrail successful products, but my internal perspective says, Nah. Intel throws/threw a staggering amount of money at failed projects. Wondering if Gelsinger will shake out the numerous boondoggles that litter that company.
- balls187 5y agoSeveral air-frames met similar fates at various defense contractors.
- bhouston 5y agoThis is fascinating. Where were they cancelled? I know about Larrabee and Itanium of course. But the others? Would they have succeeded in the market if pushed? Or was Intel just going in all directions a bit randomly without strong purpose?
- SavantIdiot 5y agoItanium shipped, and I have a soft spot in my heart for it, VLIW was brilliant. The hate is administered by knownothings, it was a CPU for engineers. Larrabee is Xeon Phi. Aren't they shipping today? These never saw the light of day, only one taped out. I don't know why, I was too busy engineering to care: - a Pentium3 + on-die, integrated graphics - a 7.2 GHz pentium 4 (yes, 7.2 GHz. in 2002) - the successor to the i740 graphics which was cannibalized for the i815 chipset - another Pentium 4 variant which I can't recall what it was, but we spun up an entire arch team to work on it - variants of Itanium (McKinley, Madhall) > Or was Intel just going in all directions a bit randomly without strong purpose? DING DING DING.
- HarHarVeryFunny 5y agoIsn't part of it also due to the cost of new fabs ($1B+)... you need the scale to generate enough profit to plough back in to keep up.