News Kioxia's Developing PCIe 5.0-Ready SSDs With New E3.S Form Factor

[Admin]

Kioxia demoed new EDSFF E3.S SSDs that promise higher densities and performance for enterprise and datacenters.

Kioxia's Developing PCIe 5.0-Ready SSDs With New E3.S Form Factor : Read more

 

[Kamen Rider Blade]

WTF Kioxia?

Normal 2.5" HDD/SSD that follow specs:
- 2.5" = LxWxH (mm) = 100.00 x 69.85 x (5, 7, 9.5, 12.5, 15, 19)

Your specs:
LxWxH (mm) = 104.9 x 76 x 7.5mm

Why on earth did you feel the need to break standard 2.5" drive specs?

 

[USAFRet]

Why on earth did you feel the need to break standard 2.5" drive specs?

"The EDSFF E3.S and E3.L form factors enable a higher power budget in comparison to the conventional 2.5-inch form factor. "

 

[InvalidError]

Why on earth did you feel the need to break standard 2.5" drive specs?

This is for servers where drives plug directly into a backplane for convenient hot-swaps. There is no point in maintaining mechanical compatibility with 2.5" HDDs/SSDs when the PCIe x4/x8 backplane connector will be completely different from SATA/SAS, all it does is possibly introduce confusion when a drive that looks like it should physically fit doesn't.

 

[hannibal]

These would be nice Also in consumer computers. Get rid of normal harddrives and sata connector and get something like these instead!
But yeah... I could nor afford these
$10000+?

 

[USAFRet]

These would be nice Also in consumer computers. Get rid of normal harddrives and sata connector and get something like these instead!
But yeah... I could nor afford these
$10000+?

We already have M.2 drives. No cables.
Not everything that exists is meant for every market or use.

 

[hannibal]

I have m2 pci 4.0 in my rig, so I,know and there is two Empty slots in there. But there is Also room for 16 3.5” hard drives and I would prefer filling those with big ssd drives 😉
And sata is just too slow, so new connector for these big next gen pci 5.0 ssd would be more than fitting!

 

[2Be_or_Not2Be]

Kioxia marketing at work - so far, nothing they demonstrated is going to exceed PCIe 4.0 x 8. To claim they are going to "deliver PCIe 5.0 performance and beyond" is greatly forward-looking on their part. 🙄

 

[InvalidError]

Not everything that exists is meant for every market or use.

If someone really wanted to toss one of these in a PC or (E)ATX server, I'm sure someone will make a PCIe slot adapter for it just like M.2 adapters were made to toss 1-4 of those on motherboards without M.2 slots.

 

[USAFRet]

If someone really wanted to toss one of these in a PC or (E)ATX server, I'm sure someone will make a PCIe slot adapter for it just like M.2 adapters were made to toss 1-4 of those on motherboards without M.2 slots.

Yep.
And then someone here will inquire how to do a RAID 0 with 3 of them, to go into their Z68/i7-2600 system...🔫

 

[bit_user]

We already have M.2 drives. No cables.
Not everything that exists is meant for every market or use.

I would argue that M.2 is not really meant for desktop use. They burn too much mobo realestate and are annoying to cool. Physical accessibility can also be an issue, sometimes requiring removal of the graphics card or even the whole motherboard!

I think the standard was just focused on laptops and maybe servers.

 

[InvalidError]

I would argue that M.2 is not really meant for desktop use. They burn too much mobo realestate and are annoying to cool.

I think M.2 is fine, plenty of empty space between x1/x4 slots begging for something to put there. Access may be annoying but most normal people won't change the SSD more than once or twice over the board's useful life so that isn't so much of an inconvenience.

That said, it would be nice to have "SATA4-Express" to bump 2.5/3.5" storage to 12-16Gbps, maybe even upgrade the connector to x2 signaling for 24-32Gbps.

 

[Kamen Rider Blade]

I think M.2 is fine, plenty of empty space between x1/x4 slots begging for something to put there. Access may be annoying but most normal people won't change the SSD more than once or twice over the board's useful life so that isn't so much of an inconvenience.

That said, it would be nice to have "SATA4-Express" to bump 2.5/3.5" storage to 12-16Gbps, maybe even upgrade the connector to x2 signaling for 24-32Gbps.

I concur, we need at least SAS 12 Gbps to be ported over to SATA4.0 with built in NVMe capability and push HDD makers to go Multi-Actuator. Maybe throw in some 16/32 GB Optane to help as a buffer.

Stop being lazy WD, you heard me, if Seagate can go "Multi-Actuator" and deliver higher linear throughput, you should be able to as well.

And bring back 15k RPM 2.5" HDD's with "Multi-Actuator", get some real Linear R/W through-put.

Stop holding back HDD technologies. Stop being quitters.

You don't want low end QLC/PLC SSD's to eat your market, start delivering at least some linear through-put.

We don't expect much Random IOP improvement, but you can stand to improve Linear R/W through-put and max out that interface =D

 

[InvalidError]

And bring back 15k RPM 2.5" HDD's with "Multi-Actuator", get some real Linear R/W through-put.

Nah, leave 10k and 15k HDDs dead and buried. They have too low density, too low performance, too high power and too high prices to compete with today's TB-scale SSDs.

 

[USAFRet]

I concur, we need at least SAS 12 Gbps to be ported over to SATA4.0 with built in NVMe capability and push HDD makers to go Multi-Actuator. Maybe throw in some 16/32 GB Optane to help as a buffer.

Stop being lazy WD, you heard me, if Seagate can go "Multi-Actuator" and deliver higher linear throughput, you should be able to as well.

And bring back 15k RPM 2.5" HDD's with "Multi-Actuator", get some real Linear R/W through-put.

And still be slower than a rational solid state drive, across all use cases.

 

[InvalidError]

I concur, we need at least SAS 12 Gbps to be ported over to SATA4.0 with built in NVMe capability

On second thought, USB4/TB3 is coming to the mainstream and the most natural thing to do would be to simply have a stripped-down USB4 spec (TB3/PCIe-only mode) to provide 20-40Gbps headers for internal off-board devices.

 

[Kamen Rider Blade]

Nah, leave 10k and 15k HDDs dead and buried. They have too low density, too low performance, too high power and too high prices to compete with today's TB-scale SSDs.

I want those for me, the end user who archives alot of stuff.

I want fast HDD's that are cheap.

I know Enterprise won't buy into it given the energy cost nature.

But me as Joe User, I need cheap storage that is reasonably fast.

 

[InvalidError]

But me as Joe User, I need cheap storage that is reasonably fast.

At a glance, the typical going rate for 10000RPM HDDs is about $150/TB and they appear to max out at 2.4TB. If your focus really is on cost and performance, then you are actually better off getting 1-4TB SSDs at $90-130/TB than 10kRPM HDDs.

High capacity HDDs are relatively slow because reading and writing smaller bits requires greater positional accuracy that can only be achieved at lower speeds. If you want affordable high capacity HDDs, you'll have to put up with 100-200MB/s sequential transfer speeds. (Or even all the way down to ~10MB/s when writing to SMR drives.)

 

[Kamen Rider Blade]

At a glance, the typical going rate for 10000RPM HDDs is about $150/TB and they appear to max out at 2.4TB. If your focus really is on cost and performance, then you are actually better off getting 1-4TB SSDs at $90-130/TB than 10kRPM HDDs.

High capacity HDDs are relatively slow because reading and writing smaller bits requires greater positional accuracy that can only be achieved at lower speeds. If you want affordable high capacity HDDs, you'll have to put up with 100-200MB/s sequential transfer speeds. (Or even all the way down to ~10MB/s when writing to SMR drives.)

I'm sure their HDD engineers can figure their way out of 15k RPM + High Density + Multi-Actuator.

There's nothing their engineers can't figure out.

 

[USAFRet]

I'm sure their HDD engineers can figure their way out of 15k RPM + High Density + Multi-Actuator.

There's nothing their engineers can't figure out.

Multi arm drive:

MACH.2 Multi-Actuator Hard Drive | Seagate US

The world’s first multi-actuator hard drive with two independent actuators that transfer data concurrently.

www.seagate.com

 

[InvalidError]

There's nothing their engineers can't figure out.

If they could figure it out, they would have done it 5+ years ago to avoid going obsolete with the rise of SSDs. Mechanical HDD speed and density are mutually exclusive, that's why HDDs above 7200RPM are effectively extinct now that SSDs completely outclass magnetic storage on all performance metrics.

The only thing HDDs are still good for is rock-bottom cost per GB for online/near-line storage.

 

[hannibal]

So true! My good old Velocity raptor is still fast (for a hard drive) but so small capasity and the price was so high. Ssd is better option for those and hd can be big and slow for long time storage usage!
Highend HD Are dying preed... high capasity hd still have their merits.

 

[bit_user]

I concur, we need at least SAS 12 Gbps to be ported over to SATA4.0 with built in NVMe capability and push HDD makers to go Multi-Actuator.

https://www.tomshardware.com/news/s...oubles-hdd-performance-microsoft-jumps-aboard

Maybe throw in some 16/32 GB Optane to help as a buffer.

Doesn't make sense, for most current HDD use cases. HDDs are about cheap volume storage. A persistent cache wouldn't tend to deliver much benefit, because if you're not booting of a HDD, then your access patterns are probably a lot more varied over time. Also, for streaming media or browsing photos and videos, HDDs are plenty fast enough.

And bring back 15k RPM 2.5" HDD's with "Multi-Actuator", get some real Linear R/W through-put.

Linear is already improving as a natural consequence of areal density improvements. The only real problem is that it's not keeping pace with capacity increases (which increase as a square of that rate). The practical consequence being that RAID rebuilds are starting to take multiple days, which is bad for a number of reasons.

Stop holding back HDD technologies. Stop being quitters.

They're constrained on price, and their play is fundamentally one of TB/$.

You don't want low end QLC/PLC SSD's to eat your market,

It's already happening. Fewer and fewer HDDs are going into PCs. In the consumer market, the predominant use is quickly becoming NAS.

 

[bit_user]

And still be slower than a rational solid state drive, across all use cases.

But they have better cold storage longevity. So, if you need high-performance cold storage, then they could still make sense.

Maybe Optane could beat them at that game, but its $/GB is still way too high.

 

[InvalidError]

In the consumer market, the predominant use is quickly becoming NAS.

And for the most part, in the 4TB and up sizes, not the 600GB of 15000RPM or 2.4TB of 10000RPM HDDs. Also, most people don't want their NAS to do double-duty as space heaters.

I don't have a NAS but my PC is almost HDD-free: when my 1TB WD Blue died, I replaced it by a 4TB Barracuda and stuffed it in an external HDD box. Only my 1TB WD black left as an active HDD in my PC, the others are offline storage. Probably going to replace it with a 1TB NVMe when I eventually upgrade my PC.