>that's several generations behind on chip fabrication. They are still trying to ramp up 10 nm while TSMC is unveiling 4 nm any day now.

It's worth noting that Intel 10nm can't be directly compared with TSMC 4nm. AFAIK Intel 10nm approximately equals TSMC 7nm. TSMC 4nm sounds like an iterative refinement on 5nm, so intel is only really one to two generations behind, not "several".

Worth noting that TSMC is relentlessly incremental. While Intel wanted to make a bigger, slower jump from 14 to 10, TSMC did an N7, an N7+ (not to be confused with N7P), an N6, N5, and N5P, and later this year, N4 is on the roadmap.

So counting "generations" is a little abstract anyways. You should probably look at the PPA numbers and the kind of scaling boosters they've managed to ship, if you want to compare fairly.

> AFAIK Intel 10nm approximately equals TSMC 7nm.

I don't think so, TSMC 7nm is ahead in many ways. For example, Intel 10nm power efficiency is much worse.

https://blog.cloudflare.com/the-epyc-journey-continues-to-mi...

Intel 10nm is maybe more comparable with Samsung 8nm.

TSMC 7nm = Intel 10nm is a crazy old trope that dates back to before everyone realised that Intel couldn't deliver 10nm. I don't know why people keep repeating it.

Any node after they stopped measuring gate dimensions is a trope.

We need a better metric, a marketing term doesn’t work.

>TSMC 7nm is ahead in many ways. For example, Intel 10nm power efficiency is much worse.

Is this based on the lithography process, or because of the uarch?

The 7nm = 10nm was from https://en.wikichip.org/wiki/File:7nm_densities.svg, which looked at the density. Based comparison tables of lithography processes, it doesn't even look like power is something that they compare (eg. https://en.wikichip.org/wiki/7_nm_lithography_process)

In your second link there's Power @ iso-speed comparisons within TSMC's nodes and TSMC has 3 revisions of their N7 with 3 different power usages. Presumably, Intel can do similar things within their generations too.

The industry should really use a different metric like "transistor count / area" because in 5 years TSMC will need to either go negative NM or 0.99 to 0.001 nm to keep the marketing slogans going.

That would only make sense if all transistors were the same size.

They're not. Not even close. Especially, in combinational logic they're all over the place.

When verilog is being turned into gates there are many arguments the engineer can give the tool and they will give very different results. It's kinda like picking which optimizations get turned on during compile, like picking speed or binary size, although not exactly like that.

The industry tried using bit/area, because an important advantage of small transistors is more memory/area, but that has basically all the same complications as well so it doesn't really work like one wants.

So just counting transistors can be totally unfair when two designs are made for totally different applications.

In the end picking a node is a complicated, multi-dimensional, decision. Often, and more often than the public thinks, the smallest one is the wrong decision. Note: Google's TPU is at least a generation behind the bleeding edge, if not two, and not because Google can't afford it if they want to.

So, basically expect the marking names to continue for process node names for the foreseeable future.

For the more EE minded, see this deck: http://www.cse.psu.edu/~kxc104/class/cmpen411/14f/lec/C411L1...

If you don't want to go through the whole thing note slide 17 especially. Those are all the same boolean function but they have very different areas, speeds, and power. Which one the tool picks will be (mostly) controlled by the designer.

To the general reader, since maybe I have your attention, and maybe you like computer science and algorithms, let me direct your attention to:

DAG-Aware AIG Rewriting: A Fresh Look at Combinational Logic Synthesis http://people.eecs.berkeley.edu/~alanmi/publications/2006/da...

Which is the inspiration for this tool: https://github.com/berkeley-abc/abc

Which how this flow does synthesis: https://openroad.readthedocs.io/en/latest/

Ever heard of picometer?

Hasn't Intel put 10A on the roadmap? I guess 10A means 10 Atoms wide.

1 - 2 generations in the silicon world is 5 - 10 years

No? TSMC has a new "full node" (whatever that means now) every 2 years. And Intel 4 should be released the same year as TSMC N4.

Several is inclusive of two.

Only to a tiny minority of people: https://github.com/zonination/perceptions

The figure in your linked page looks like it is based on data from a survey in which participant were asked which number they thought of when they heard the word 'several'. Even if everyone taking the survey was aware that the word 'several' is inclusive of two, I'm not sure we would expect a different result. So, the results can't be used as evidence that anybody taking the survey did not think several is inclusive of two.

IMHO, VMware did the best it could under Gelsinger's leadership. They evolved their cash cow (vcenter) better than other on-prem vendors.

- NSX (via Nicira)

- VMware in the cloud through all hyperscalers

- Tanzu (via Heptio)

Also made security as a core successful pillar for their offering.

They did not have the DNA to build a hyperscaler. They tried vcloud, but it was a failure.

I've sufered their virtual cloud solution professionally, I concur. And not only they failed, while they focused on providing Software Defined Data Center solutions to enterprises, the enterprises were already trying to figure out how to actually kill the infrastructure cost. It wasn't too much of a software feature problem than it was a hardware cohesion problem, which VMware weren't even very good at providing support for to make their software actually be somewhat on par with say AWS which itself isn't particularly shiny in term of performance.

We were getting sub ms latency between applications on the same DC site, across all layers up to application. With VM were we were well over 2ms and at the virtualised network level. With marginal improvements on that front observed after 2 years of back and forth support projects.

>we need large US-based fabricator if China gets more aggressive on Taiwan

So take that concept further of what happens if/when China makes that move. What happens if TSMC's fabs are sabotaged/destroyed during this scenario? This would make today's chip shortage seem like just a blip. Would that be the thing that makes Intel relevant again? If this current shortage actually motivates US based chip manufacturing equal to TSMC, the loss of TSMC would obviously not be as severe. How much would China benefit from being the sole customer of TSMC at that point? Time to crowd source a new scifi/techfi plot

Now that you mention it, it's a little odd that VMware isn't used by any of the major cloud providers while they also haven't created their own Cloud Provider. It seems like they are very few hires away from a core competency in competing with at least GCP or Azure.

Makes me wonder why, is it because they see a big future yet in on-prem and can't really embrace cloud computing until that future looks less certain?

Because cloud is commoditized VMs and VMware doesn’t have an edge in managing cattle at scale.

Their bread and butter is in managing pets and cloud providers expressly don’t want to encourage people to bring their pets into the cloud because people get mad when a few get killed.

OpenStack with the KVM backend is better than VMware at multi-tenancy so there really isn’t much value for VMware to add even over open source state of the art IaaS.

As someone who works for a non-major datacenter. We use VMware and have evaluated OpenStack. Lots of moving parts for OpenStack, administrative burden. So we don’t use OpenStack.

OpenStack, VMware and all the rest should go through a process that audits and certifies them like what is done to transport systems on and off the ground while not getting in the way of step level improvements introducing innovative new nicer things.

Once you start charging enterprise prices you can't stop, but public clouds can't pay enterprise prices. It would have appeared to be financial suicide for VMware to move downmarket into public cloud.

Enterprise costs are negotiable. I can easily imagine Google paying VMware several billion per year for a product they want. I can’t imagine Google using ESX — they want a product they fully control.

I don't believe any bigger cloud providers want to adopt closed unmodifiable products as a core. Windows/Hyper-V is similar from outside but it's in house for Azure(MS).

Even second-tier clouds that can't afford to develop their own hypervisor don't use ESX though. Literally no public clouds use it.

It's a bit less, er, cloudish and self-service than I usually think of cloud as being, but these folks do use VMware:

https://expedient.com/services/cloud/

The cloud is a capex game not a tech game. Also, the layer that VMware compete in, is fully commoditized (this is the infra as a service layer). The game is moving to higher layer (e.g. ML/ DB).

I think that for on-prem they faced a treat from containers, which they are embracing by betting the company on Kubernetes.

Vmware spent over a billion on their own cloud provider which failed utterly and the remnants were sold to Amazon.

What is blocking this industry in the US? If it's staff, surely some kind of special visa program combined with tax breaks could be a pretty generous offer to TSMC to setup shop onshore.

The Taiwanese industry people want to go to China because that is where the market will go. So much so that Taiwan had to block Chinese recruitment centers and if I'm not mistaken restricting travel on some of those key people.

TSMC is one of Taiwan's biggest strategic assets, they will never manufacture the most cutting-edge products offshore.

Did he run that into the ground? Or the tech just became obsolete for modern environment?

That is very much like running it into the ground. They could have had a strong platform to pivot existing customers to cloud solutions or made on premises better. Instead a book retailer pivoted to that.