Gen5 NVMe have active cooling

[BFG10K]

Consumer PCIe Gen 5 Gaming SSDs Arrive For Sale In Japan At A Smoking 10GB/s

The first PCIe 5.0 SSD popped up in Japan with a familiar face and blistering performance.

hothardware.com

A fan on an SSD? This is really getting quite ridiculous. These things are turning into GPUs.

 

[biostud]

Next gen Phison E28 @ CES25

Phison unveils next-generation high-end PCIe 5.0 SSD platform: PS5028-E28

SSDs using Phison's PS5028-E28 controller potentially could arrive in 2026

www.tomshardware.com

 

[biostud]

Samsung is launching their next gen PCIe 5.0 drives

Samsung Unleashes Its Next-Gen 9100 PRO Gen5 SSDs, Up To 14,800 MB/s Speeds With Up To 8 TB Capacities

Samsung has officially unveiled its next-gen 9100 PRO Gen5 SSDs, featuring impressive speeds and up to 8 TB capacities.

wccftech.com

 

[CakeMonster]

Finally the controllers are decent enough to match what we should have had at the initial Gen5 launch, no silly massive heatsinks or extreme throttling.

I can't wait to see the random performance in benchmarks. A bit of a downer that the 8TB model is delayed for so long, but I'd be surprised if they launched at acceptable prices anyway...

 

[biostud]

Predator GM9000 Gen5 2TB SSD Review – SMI SM2508 Controller Creates SSD Perfection - The SSD Review

It's pretty rare that I might use the word perfection when it comes to SSDs. There is always somewh

www.thessdreview.com

First retail SM2508 drive, seems good.

 

[CakeMonster]

No 8TB model though... well maybe its coming later.

Endurance for the GM9000 starts at 800TBW (terabytes written) for the 1GB version and doubles for every capacity jump with a total of 3200TBW for the 4TB SSD.

That's 33% better endurance than the standard on the high end Gen4 SSD's (2400TBW for 4TB), I wonder how they got there.

 

[00Logic]

That's pretty informative stuff. Do any of those changes lead to better drive performance as well? And not just less heat?

This is all IIRC as I am now on Linux.

My reading on the subject 4K sectors showed no, to very little difference in performance.

I assume this is because the ctlr chip (actually a 2 to 4 core ARM processor nowadays) is very quick at the translating of 512e to 4Kn.
So the advantage is just in the fact that the the ctlr produces less heat if it does not have to do this step.

When it comes to aligning or matching the FS cluster size to that 4K sector size; I have seen improvements in Q1T1 R4K.
As 68% of windows I/O is in fact Q1T1 R4K with less than 1% of I/O being the large sequential numbers advertisers like to ave around like burning flags, there is a noticeable improvement in OS snappiness here.

https://www.thessdreview.com/ssd-guides/beginners-guide/the-ssd-manufacturers-bluff/

https://gamersnexus.net/guides/1577-what-file-sizes-do-games-load-ssd-4k-random-relevant

NB that when it comes to NAND based storage (SSDs and flash drives etc) the NAND has a Page Size and an Erase Block Size.
Here aligning partitions to Erase Block boundaries and FS cluster sizes to Page sizes can make a large 30%+ difference to speed, as ell as increasing lifespan considerably.

This is especially true of USB flash drives and SD Cards which have very small, weak ctlr chips.

You may also be interested in 2 of Uwe Sieber's apps:
(He's the guy you wrote the DOS and/or in 95 USB drivers IIRC)

1st one is USB-WriteCache V0.2:
WindBlows does NOT enable caching for any FS besides NTFS.
It only pretends to, or LIES. (Probably to make NTFS look shinier than it actually is)
This tool allows you to actually enable RAM (write) caching for exFAT etc.
In my experience exFAT, with caching is faster than NTFS for small directory trees.

The same registry entries in the right place can be used to enable RAM caching for drives that "don't support caching" according to WindBlows.

2nd is MaximumTransferLength:
This adjusts the USB 'packet' size from the stock 64kb, all the way up to 2MB.
Here matching that packet to the average file size can make a big (30%) difference.
BUT also NB the size to the Erase block size of the NAND. (normally 512KB)
That's the std MTL for UASP enabled USB3.