Micron Ships 2400 PCIe Gen4 Client SSD Based on 176-layer 3D QLC NAND Flash

[AusWolf]

Function over fashion. F^&% fashion.

Fewer cables - easier maintenance.

 

[lexluthermiester]

Fewer cables - easier maintenance.

I don't care and I'm hardly alone on that point. I grew up with rats nests of cables in PC cases. Floppy, MFM, IDE, SCSI and others were all over the place. So modern PC's with a few power and SATA cables are delightful by comparison and they're easily managed.

 

[mashie]

With QLC and 176 layers I would expect drivers larger than 2TB.

 

[trsttte]

Nothing. The solution is not data density per cell.

Makers need to tell the industry pundits to release the SATA4-72gbps spec and release SATA drives with lots of TLC chips for higher capacity drives. M.2 is a great interface for ultra compact PC platforms like ultra compact desktops, laptops and ultrabooks. Normal desktops need normal drives. Practicality and functionality need to prevail. We don't need motherboards with 3 or 4 NVMe slots. What we need is standard size(2.5" & 3.5") drives of decently large capacity and a bus the supports the speed they can provide.

Yes and no. Density is also important, we can't just throw more nand chips at the problem, there's a point where the tech needs to get better as well (as an example look at HDD that are stuck in the 20tb ballpark and with barely any speed or price improvement). QLC has it's uses but with current pricing it makes very little sense for main drives. We don't have good performance numbers on this new micron tech (only the usual fluff while SLC is still available) but if it improves on previous QLC models it will broaden the uses for it more.

Now we don't really need SATA4 for anything, we have U.2 for pcie 2.5'' ssds that don't need to follow the m.2 space constraints, we just need it to get to the consumer space but for that to happen large capacity SSDs need to get cheaper. Most consumers only need a terabyte or two and m.2 can easily fill that space. For people with bigger needs they'll still prefer HDDs because the price is still much lower, and for ssds to get cheaper you can't just be throwing more nand at drives

 

[lexluthermiester]

Density is also important, we can't just throw more nand chips at the problem

Sure we can. Instead of 2, 3 or 4 NAND chips on a 2280 form factor, we use 4, 6, 8 or a few more to get higher capacity SSD's in a 2.5" 7mm form factor.

QLC has it's uses

Sure, as incedental storage in external/usb drives, not as boo/OS drives.

We don't have good performance numbers on this new micron tech

Hint, it's not any different. And again, QLC durability in P/E cycles is pathetically low.

Now we don't really need SATA4 for anything

That's an opinion. SATA4 would be a solid way forward in terms of compatibility and continued expandability.

ssds to get cheaper you can't just be throwing more nand at drives

Cost is not as important as durability, performance and capacity. I would rather spend $200 on a 2TB MLC drive than $100 2TB QLC drive. All day, any day.

 

[trsttte]

Sure we can. Instead of 2, 3 or 4 NAND chips on a 2280 form factor, we use 4, 6, 8 or a few more to get higher capacity SSD's in a 2.5" 7mm form factor.

Sure, as incedental storage in external/usb drives, not as boo/OS drives.

Hint, it's not any different. And again, QLC durability in P/E cycles is pathetically low.

That's an opinion. SATA4 would be a solid way forward in terms of compatibility and continued expandability.


Cost is not as important as durability, performance and capacity. I would rather spend $200 on a 2TB MLC drive than $100 2TB QLC drive. All day, any day.

Just using more nand chips is expensive. You can buy large capacity u.2 2.5 with lots of nand packages today but they're expensive and for that reason reserved for enterprise (which makes them even more expensive).

I don't know if QLC is any better or not than previous versions but any progress is good progress. I'd also prefer 2tb 200$ MLC to 2tb 100$ QLC but everyone has different priorities and, looking at what's available, the MLC Samsung 970 Pro costs ~190€ for 1tb while a 2tb QLC drive like the Crucial P2 goes for about ~160€. It's a tough comparison, the P2 is kind of garbage but MLC is way too expensive, different uses will have different requirements, the P2 can be a good game drive for example but for OS and regular use is a tought sell for me.

I don't have any info on SATA 4, might be great and all, but we already have U.2 ready and on the market. SATA has nice backwards compatibility but will that remain true on a jump from 6 to 72gbit/s?

 

[kapone32]

I agree and disagree. SATA is great for data storage, although I prefer having only M.2 drives in my desktop PC for better cable management.

It's too bad SATA suffers the same issue. NAND flash is overpriced past 2TB.

With QLC and 176 layers I would expect drivers larger than 2TB.

Yes but the price is the mitigating factor.

 

[lexluthermiester]

Just using more nand chips is expensive.

Not as expensive as you might think. And I'm ok with a more costly, but much more durable high capacity(4TB, 6TB or 8TB) drive.

I don't know if QLC is any better or not than previous versions but any progress is good progress.

That's the problem, the progress made is so minor as to be insignificant. So instead of a QLC cell having a P/E durability of 800 it's been improved to 850. It's still pathetic and unacceptable for OS drive usage.

I don't have any info on SATA 4

I was being theoretical. It would be trivial to bump the SATA3 speeds up dramatically without changing much to the SATA bus spec. It's beyond stupid that they haven't done this yet.

but we already have U.2 ready and on the market.

Not everyone wants to use a new connection system that isn't readily backward compatible with previous SATA devices.

SATA has nice backwards compatibility but will that remain true on a jump from 6 to 72gbit/s?

Can't see any reason why not.

 

[AusWolf]

Just using more nand chips is expensive. You can buy large capacity u.2 2.5 with lots of nand packages today but they're expensive and for that reason reserved for enterprise (which makes them even more expensive).

I don't know if QLC is any better or not than previous versions but any progress is good progress. I'd also prefer 2tb 200$ MLC to 2tb 100$ QLC but everyone has different priorities and, looking at what's available, the MLC Samsung 970 Pro costs ~190€ for 1tb while a 2tb QLC drive like the Crucial P2 goes for about ~160€. It's a tough comparison, the P2 is kind of garbage but MLC is way too expensive, different uses will have different requirements, the P2 can be a good game drive for example but for OS and regular use is a tought sell for me.

I don't have any info on SATA 4, might be great and all, but we already have U.2 ready and on the market. SATA has nice backwards compatibility but will that remain true on a jump from 6 to 72gbit/s?

I'm actually using a 1 TB Crucial P2 as my OS drive. Its sequential read/write speeds are sub-par, and its endurance rating is terrible, but for £60, I can't complain. You don't feel an OS drive's linear speed anyway, only its latency, which is good in case of every SSD (even SATA ones).

Honestly, I've had extremely fast PCI-e Gen 4 OS drives, I've had Gen 3, I've also had SATA. The real-life difference you feel is zero. You only feel something when you work with a spinning drive. Not to mention that I boot my OS maybe once a day. I couldn't care less if it takes 10 seconds instead of 8.

I don't care and I'm hardly alone on that point. I grew up with rats nests of cables in PC cases. Floppy, MFM, IDE, SCSI and others were all over the place. So modern PC's with a few power and SATA cables are delightful by comparison and they're easily managed.

I have some fond memories of that time myself, but computer maintenance is not one of them.

 

[lexluthermiester]

but computer maintenance is not one of them.

That can be said of any PC build. Do a hard-tube custom loop and you will know a whole new brand of maintenance pain. Even flex tubing is unpleasant. If a PC is built properly, headaches can be minimized.