MSI Shows Off Spatium M570 PRO and M570 PCIe Gen 5 NVMe SSDs

[btarunr]

MSI in its 2023 International CES booth showed us their upcoming M.2 PCIe Gen 5 + NVMe 2.0 SSDs, the Spatium M570 PRO and the Spatium M570. The M570 PRO unsurprisingly is the faster of the two, and rocks a sophisticated new cooling solution by MSI. A vapor-chamber plate makes direct contact with the controller and other hot components on the drive, conveying heat to a heatsink with a very fine aluminium fin-stack, whcih is capped off by a die-cast aluminium top-plate. This design, MSI claims, optimizes convectional airflow without interference from nearby airflow sources (such as CPU coolers).

The Spatium M570 PRO offers sequential transfer rates of up to 12 GB/s reads, with up to 10 GB/s writes, and is available in capacities of 1 TB, 2 TB, and 4 TB. The MSI Spatium M570 (non-PRO) rocks a simpler ridge aluminium monoblock heatsink, and comes in the same capacity options, but with lower maximum read performance. You get up to 10 GB/s sequential reads, with up to 10 GB/s sequential writes with this one.



View at TechPowerUp Main Site

 

[DeathtoGnomes]

so how long to market for these?

 

[dj-electric]

so how long to market for these?

I can tell you preproduction samples are already all around. My guess is that not a lot of time at all, possibly spring for some models

 

[TheLostSwede]

so how long to market for these?

Q1 maybe, depends a bit on firmware related stuff and availability of the right type of NAND.

 

[AnarchoPrimitiv]

Forget sequential speeds...when are random, low queue depth r/w's going to get a marked increase in performance? Is anyone intimately familiar with NAND flash? Is it an inherent limitation of the NAND flash that's the reason why random reads and writes haven't seen drastic performance increases in the evolution from PCIe 3.0 to 4.0 to 5.0?

 

[Wirko]

Forget sequential speeds...when are random, low queue depth r/w's going to get a marked increase in performance? Is anyone intimately familiar with NAND flash? Is it an inherent limitation of the NAND flash that's the reason why random reads and writes haven't seen drastic performance increases in the evolution from PCIe 3.0 to 4.0 to 5.0?

I don't know all the details but yes, it's an inherent limitation of the NAND technology. OS, application and game developers will have to adapt and use queueing more. Random writing isn't actually so slow, it's several times faster than reading in many SSDs, but it is so because data goes to SLC cache first, probably in a sequential manner.

Toshiba was trying to make "something better", I don't know what came of it and whether it exists in the TLC world too:

Toshiba Launches XL-FLASH 3D SLC NAND

www.anandtech.com

 

[Chris_Ramseyer]

Forget sequential speeds...when are random, low queue depth r/w's going to get a marked increase in performance? Is anyone intimately familiar with NAND flash? Is it an inherent limitation of the NAND flash that's the reason why random reads and writes haven't seen drastic performance increases in the evolution from PCIe 3.0 to 4.0 to 5.0?

If you look at low queue depth random read performance from Gen3 to Gen5 we doubled the IOPS (cut latency in half). The new drive already surpasses 100MBps in Q1T1 CrystalDiskMark. That will increase again when we progress from 1600MT/s NAND to 2400MT/s over the life of E26.

With DirectStorage coming later this month, there will be a real need for throughput and more importantly, sustained constant throughput, to allow your games to run at high image quality settings.

so how long to market for these?

Very, very, very soon.

 

[Minus Infinity]

Just quote the random performance please, no one cares about the sequential marketing BS.

 

[Chris_Ramseyer]

Just quote the random performance please, no one cares about the sequential marketing BS.

1.5m iops

 

[TumbleGeorge]

Forget sequential speeds...when are random, low queue depth r/w's going to get a marked increase in performance? Is anyone intimately familiar with NAND flash? Is it an inherent limitation of the NAND flash that's the reason why random reads and writes haven't seen drastic performance increases in the evolution from PCIe 3.0 to 4.0 to 5.0?

This is because security BS. Windows check any file for signature. This eat lot of time when read many thousands small files.

 

[ToxicTaZ]

I'm wanting these new SPATIUM M570 PRO on my new Motherboard....MEG Z790 ACE MAX

14900KS 6.5GHz (Raptor Lake Refresh)

Cheers.... Hopefully by Black Friday 2023