Tuesday, February 9th 2016

Plextor Unveils the M6S Plus Series SSD

Plextor announced the M6S Plus line of performance consumer SSDs, succeeding the company's current M6S lineup. The drives implement Toshiba A19 toggle NAND flash memory built on the 15 nm process, and are driven by Marvell 88SS9188 controllers. Available in 128 GB, 256 GB, and 512 GB variants, the drives serve up sequential transfer rates of up to 520 MB/s reads on all three variants, and up to 300 MB/s, 420 MB/s, and 440 MB/s writes, respectively. Random access speeds range up to 90,000 IOPS reads, with up to 80,000 IOPS writes. Plextor's proprietary PlexTurbo data-caching and PlexCompressor data-compression tech carries over from the M6S. Built in the 7 mm thick, 2.5-inch form-factor, the drives feature SATA 6 Gb/s interface.
Add your own comment

9 Comments on Plextor Unveils the M6S Plus Series SSD

#1
hojnikb
>
The drives implement Toshiba A19 toggle NAND flash memory built on the 15 nm process


Last time i checked, A19 means second generation 19nm flash, so this statement makes no sense.
Posted on Reply
#2
bug
Basically, progress has stopped at SF-2000 levels. All we see since then is manufacturers moving to lower nodes in an attempt to lower prices, but having to battle data retention in the process.
Reads have been pegged at 500-550MB/s and writes at 400-450MB/s for a few years now. That's as much as SATA3 can sustain. NVMe can do better, but you only need that if you frequently do large, sequential data transfers (which most consumers do not). The more critical 4k random reads are still at under 100MB/s where they where three years ago, except that budget drives can do far worse than 100MB/s (usually around the 50MB/s mark).

So any SSD launched can simply be summarized as capacity, price, random 4k read performance.
Posted on Reply
#3
Parn
This is just like when consume HDDs hit the 100 - 120MB/sec sequential transfer rate barrier about 10 years ago. Now the focus of any new SATA SSDs will be capacity and $ per GB.

This is alright with me. I'd like to see the prices of 512GB and 1TB SSDs falling to reasonable levels (preferablly equal to the prices of HDDs with double capacity, e.g. 1TB SSD = 2TB HDD). When that happens I can go fully non-mechanical with my desktops. The file servers can stay on monstrous size HDDs with RAID controllers for now.

For people who do need faster SSDs, there are PCIe x4 versions available to them.
Posted on Reply
#4
RejZoR
I miss SandForce durability. In some test SandForce drives were the only ones to reach 2 PB of writes (yeah, 2 petabytes!).

I wonder how PlexCompressor works. From what I've read on Plextor site, it's a space saving feature, not write durability saving. In fact, compressing data probably involves rewriting of data on NAND cells, actually making writes worse.

I wonder why no controller uses compression like SandForce did, but not for speed, but for durability instead (so that they write data at the same speed regardless if it's compressible or not). Marvell controllers don't have this out of the box afaik. And I don't remember Samsung having it either.
Posted on Reply
#5
hojnikb
RejZoR said:
I miss SandForce durability. In some test SandForce drives were the only ones to reach 2 PB of writes (yeah, 2 petabytes!).

I wonder how PlexCompressor works. From what I've read on Plextor site, it's a space saving feature, not write durability saving. In fact, compressing data probably involves rewriting of data on NAND cells, actually making writes worse.

I wonder why no controller uses compression like SandForce did, but not for speed, but for durability instead (so that they write data at the same speed regardless if it's compressible or not). Marvell controllers don't have this out of the box afaik. And I don't remember Samsung having it either.
Only certain sandforce drives. Most of them were hardcapped to 3000*capacity (so around 700TB for 240G drives)

Besides, compression is only partly good. Its not like all data is compressible. Its better to do more efficient garbage collection, wearlevelling and ecc then to solely rely on compression to reduce nand writes.


As for the actual compression; afaik Phison does it on 4K writes on some of their controllers.
Posted on Reply
#6
RejZoR
I know compression is not the sole reason for such durability, but it helps. Still, they had some impressive magic sauce to achieve that. I think it was Kingston HyperX with SandForce. Not sure about the capacity.
Posted on Reply
#7
hojnikb
RejZoR said:
I know compression is not the sole reason for such durability, but it helps. Still, they had some impressive magic sauce to achieve that. I think it was Kingston HyperX with SandForce. Not sure about the capacity.
You must be thinking techreports endurance test.
It was hyperx 3k.



""The life attribute takes compression into account, so it's clear this HyperX survived on more than just SandForce mojo. The low number of reallocated sectors suggests that the NAND deserves much of the credit. Like all semiconductors, flash memory chips produced by the same process—and even cut from the same wafer—can have slightly different characteristics. Just like some CPUs are particularly comfortable at higher clock speeds and voltages, some NAND is especially resistant to write-induced wear.
The second HyperX got lucky, in other words.""
Posted on Reply
#8
bug
RejZoR said:
I miss SandForce durability. In some test SandForce drives were the only ones to reach 2 PB of writes (yeah, 2 petabytes!).

I wonder how PlexCompressor works. From what I've read on Plextor site, it's a space saving feature, not write durability saving. In fact, compressing data probably involves rewriting of data on NAND cells, actually making writes worse.

I wonder why no controller uses compression like SandForce did, but not for speed, but for durability instead (so that they write data at the same speed regardless if it's compressible or not). Marvell controllers don't have this out of the box afaik. And I don't remember Samsung having it either.
That durability was most likely because of the older manufacturing process. When SSDs were released in the consumer market, they pushed 10k p/e cycles. Then it was 5k, then 3k and now ordinary TLC gets 1k. And in the meanwhile this number has become increasingly difficult to find.
The only bright spot in this picture is when Samsung went V-NAND, they were able to actually increase the physical size of the memory cell, thus their TLC is as durable as planar MLC. But that's still ~2k p/e cycles (estimated @Anandtech).
Posted on Reply
#9
hojnikb
2k p/e is a very very conservative estimation.
Posted on Reply