Micron to Sell 3D XPoint Fab to Texas Instruments for $900M | Hacker News Reader

Micron to Sell 3D XPoint Fab to Texas Instruments for $900M

tomshardware.com

98 pointsrbanffy35 comments

I've got several friends working at this fab and it's a shitshow over there. Unsafe working environment with low pay. Many experienced workers are jumping to the new Tyson foods plant which pays much better, or switching careers entirely.

You'd think a fab, like most high tech, would be a "good job".

But close to where I lived in Cupertino there was a superfund site due to the all the insane chemicals that leached into the ground from a long gone clean room.

Disclaimer: I was an intern at Micron while 3D XPoint was in development. I am entirely unfamiliar with the Lehi facility.

I can't say what I saw of fab production in Boise made it a "bad job", but like anything, there is a range of experiences. I've worked in a variety of manufacturing environments, and while clean rooms are certainly different than the others, I'd probably prefer it if I were a production employee. Personally I'd prefer it to cutting protein, but different strokes for different folks.

Overall it looks like Micron is taking a hit, given that they bought Intel's 49% stake for 1.5 billion less than 3 years ago. https://siliconangle.com/2019/01/15/micron-buy-intels-stake-...

Semiconductors inherently have a plethora of harsh chemicals involved. The country as a whole did a terrible job with waste management until we {society} got serious about not being absolutely awful to the environment. There are many waste sites which were never really documented, the environment wasn't considered when selecting them, etc. If that was the state of affairs at the time, it's not surprising that there are superfund sites at former clean rooms.

It does become difficult applying current standards to past actions. The condoned practice for certain categories of chemicals was to bury them. When those best practices are changed, you can't reasonably just dig it all up and deal with it. Digging it up is sometimes more hazardous to the environment than leaving it be. Some manufacturers have teams of contractors on site dealing with legacy waste issues. They're conducting and reviewing soil and water samples, reviewing drawings with known dump sites, water monitoring, etc before allowing any dirt to be moved.

To clarify, the workers moving to Tyson aren't working on the food production. They're doing similar work that they did at Micron: maintaining and repairing chemical systems.

Is this a situation where technical and colloquial language mean different things, like when a company counts every molecule involved in a process is called a chemical? Like if you used water to clean the meat then it would be considered another chemical used?

If the answer is no, how do the skills necessary to build the two types of plants overlap? If I was asked an hour ago if I thought the same skills necessary to maintain chemical systems in a chip fab we’re the same kind of skills necessary to maintain a meat processing plant, then I would have confidently said no.

These are the guys hooking up valves, fixing pipes, monitoring how chemicals are loaded from the trucks into the plant systems... doesn't much matter that's flowing through the pipes- the workers can read the data sheets and procedures and work on anything.

Again, these are blue collar jobs. High school education at best. Mostly a manual labor job. They're not designing the plant, just operating and repairing it according to plans designed by someone else.

EDIT: Here's a list of chemicals used in meat processing: https://www.osha.gov/meatpacking/hazards-solutions. Ammonia for refrigeration. Chlorine, hydrogen peroxide and peracetic acid for disinfectant. The chemicals are delivered to the plant in concentrated forms and diluted for use.

A lot of what go into both are the same: Water, heat, filtration, purification, etc.

In fact, a lot of chemical engineering (some would say the lion's share) has to do with heat and water management. It's a huge factor whether you're operating a clean room or an oil refinery.

I toured the Lehi “fab” in high school in 2002 or 2003. It’s still the largest emptiest building I have ever been in. I think something like 5% or less of the square footage was used. Supposedly it used more concrete than I-15.

When Micron expanded their fab in Boise, they didn't fill it up right away. They built it much larger than they needed at the moment because building fabs is incredibly expensive, tooling fabs is incredibly expensive, but building one marginally bigger isn't that much more expensive in terms of added construction and maintenance costs.

I’m well aware, although afaik they struggled to populate most the space for a long time. My uncle is a pipe fitter and Fab7/11 work has been practically seasonal since the 80s.

> My uncle is a pipe fitter and Fab7/11 work has been practically seasonal since the 80s

Is it really seasonal? Like tied to their Capex announcements or something?

Not quite, but it’s on and off many times over the years. The complex is more than a mile long so there’s always work in there. There used to be a small Philips fab too nearby, last several years more work has been available in refineries and power plants (natural gas conversion)

I've been saying for years now (probably gonna get down voted for this) - everyone wants to buy American and to make more products in America - with the amount of chemicals involved in electronics manufacturing required, who the hell wants these things made here?

Most companies are not safe or environment cautious - how many tails of leaks or problems with chemical processing plants have there been? No one wants that stuff in our ground water...

Don't get me wrong; it's not acceptable anywhere / other parts of the world - just looking at the whole "let's make everything here!..." but why?

Economic security?

I hear stories from them all the time. Casual discussion of the times they're gotten sprayed in the face or splashed with various chemicals. Of people being injured by falls. Of the chemicals that will kill you 10 minutes after skin contact.

I'm really glad my friends are jumping ship before they get seriously hurt. A few commute by motorcycle and we joked that the ride in was the least dangerous part of their day.

Fab workers are jumping ship to a meat processing plant? Is working at a fab an unskilled job or are the working conditions/job prospects really that bad?

Most of the guys aren't Micron employees doing tech work in clean rooms. They're contractors doing dangerous blue collar work with chemicals, pipes and factory equipment.

3D XPoint is amazing stuff. I'm really bummed out it hasn't taken the SSD market by storm. Even highly technical people fail to understand the critical importance of fast random access for a multitude of problems.

I think at least for now Optane is being marketed to IOPS heavy servers and workstations at a premium. I too hope it makes its way out into general storage applications, but also in more creative ways - like potentially replacing DRAM - if it gets fast enough.

[edit] I will say I'd think Micron would be in a better position to get it there than TI, though, given their position in the DRAM and NAND markets.

Note that Micron is giving up on 3D XPoint entirely, and TI is only buying the fab and most or all of the current tooling inside it, but not the IP. So TI can't pick up this torch. If 3D XPoint has any significant future, it's solely with Intel and their Optane branded products.

It's interesting that TI would want all of that tooling: part of 3DXPoint's problem is that it has some very nonstandard fab steps to build the central structure. (This makes it expensive, and attempts at cost reduction seem from the outside to have failed.) So why would TI want that tooling (if indeed they are getting and keeping it all)?

It makes one wonder about TI's FRAM product line, which if I'm remembering correctly involves similar materials as 3DXPoint. But since they do not get the IP, they're not going to manufacture 3DXPoint. And FRAM is sufficiently different that it is very unlikely to replace NAND. In particular, FRAM reads are destructive and wear down the cell! That may work out with a large, lower-density cell being accessed relatively infrequently by a lower-end embedded part, but will certainly fail for small, high-density cells accessed heavily. FRAM is weird stuff!

I don't know much about the exact tools needed for 3DXPoint, but the value of semiconductor manufacturing capacity generally in the U.S. has dramatically increased in the past year. This is now seen as a strategic resource and the government is willing to subsidize it. The value of 3DXPoint production though is not really strategic to the U.S. since hardly anything uses it that couldn't also use regular flash memory.

If a fab can show that they have the ability to at least partially fill the role of a TSMC in case of a disruption, there are large profits to be had. This may be the context and motivation for the sale.

With NVMe drives, load times for personal use applications are largely CPU-bound. For example, benchmarking game load times comparing a SATA SSD to an NVMe drive over four times as fast resulted in a less than ten percent shorter load time. Going to a tier above that just doesn't make sense for 98 percent of home users.

A lot of this is due to legacy software decisions.

To deal with old slow drives, people use expensive compression algorithms which take the CPU longer to decompress than reading the uncompressed data itself.

Also, if we were to directly copy texture data from a fast SSD into video memory, instead of SSD --> RAM --> VRAM, the loading times would see a huge improvement as it bypasses the extra copy.

> Also, if we were to directly copy texture data from a fast SSD into video memory, instead of SSD --> RAM --> VRAM, the loading times would see a huge improvement as it bypasses the extra copy.

The extra trip through CPU memory is not a significant bottleneck. DRAM bandwidth is still much greater than a single SSD's bandwidth. And there's no PCIe traffic saved, because even with P2P DMA from the SSD to the CPU, data needs to get into the CPU package where the PCIe root complex is to be routed to the GPU. There's a marginal latency improvement, and you can save a bit of the CPU's RAM capacity by not needing to use as much of it for IO buffering.

You've probably misunderstood the mechanisms behind the claims that new NVMe-based video game consoles enable much faster loading times, and the somewhat-related Microsoft DirectStorage technology (which is still rather ill-defined). P2P DMA is a minor optimization. Offloading decompression work from the CPU to the GPU is a big help, if the GPU has the right hardware to accelerate that decompression.

But by far the biggest improvement comes simply from increasing the baseline hardware capabilities that developers are targeting. When game developers can count on running off an SSD, it becomes worthwhile to overhaul the software's approach to IO so that the game issues requests with the high queue depths necessary for extracting full performance from flash-based SSDs, and the game doesn't have to pre-load multiple minutes worth of content into RAM because it can count on being able to perform multiple asset fetches per frame.

NVIDIA calls this GPUDirect Storage. You might find their blog post to be useful: https://developer.nvidia.com/blog/gpudirect-storage/

https://devblogs.microsoft.com/directx/directstorage-is-comi...

There is some debate over the actual price, discussed here: https://semianalysis.substack.com/p/micron-sells-3d-x-point-...

3D XPoint SSDs should be amazing for databases I think. I am curious to the point that I would try to write product targeting this specific tech. Unfortunately 3D XPoint's commercial status is in a very iffy state

In PCIe gen 3 it is already possible to saturate the PCI lanes with NAND. Even in gen 4 you get 6 GB/s. You gain some in latency but you really need to invest in the software to eke out that latency gain.

3DXpoint really shines in persistent memory but that requires investing quite a bit to change software to make use of it and the problem is that of a single vendor and high prices.

There are users for it but I havent seen anything to convince me its really viable.

Pmem serials should be more suitable for databases. Intel should put more efforts on the market strategy.

As long as the hardware and accompanying libraries support transactional memory then sure. Otherwise it is more easy to just use the vast existing transactional storage libs. The say that Pmem supports block access so maybe it is win win in this area

Connection between XPoint and the performance characteristics of Optane products was never clear.

For example, why would XPoint devices have near same throughput to conventional SSDs but half the latency? Why smaller densities (the nm node?)?

I'm pretty sure a decent chunk of this is related to the block granularity. NAND flash is a block-oriented structure; NOR flash and 3DXPoint are byte-oriented. There's a pretty different internal storage and ECC path for those two approaches. In particular, due to how it is accessed, 3DXPoint does not (as far as I know) run background erases and block GCs. This is a major help to its latency.

> 3DXPoint does not (as far as I know) run background erases and block GCs.

Seems like they do need wear-leveling. https://twitter.com/stuntpants/status/1410679934515773441

Reply on news.ycombinator.com