Seagate FireCuda 520 PCIe Gen4 NVMe SSD Linux Performance

[phoronix]

Phoronix: Seagate FireCuda 520 PCIe Gen4 NVMe SSD Linux Performance

For those that have been considering the Seagate FireCuda 520 as a PCI Express 4.0 NVMe solid-state drive, here are some benchmarks under Ubuntu Linux with this ZP500GM3A002 drive.

http://www.phoronix.com/vr.php?view=29365

[DanL]

Read performance seems slow for Phison E16. Maybe a future firmware update will help?

Oh, and are there any Phison E18-based SSD's out in the wild yet?

[Chewi]

Very pleased with my 510 although I don't have anything in the same league to compare it against. There was no point holding out for the 520 as my motherboard only has PCI Express 3.0.

[Adarion]

I still don't fully trust all that TLC and "higher" stuff. MLC is my chice, if I was rich it would even be SLC.
I'm still surprised about the performance, there is no real ingredient by Seagate here, they just combine things, but it seems they had a good hand at this.

[Melcar]

Been wanting to see some Linux benchmarks for these drives. Anyone with a Seagate SSD has tried their SeaTools? I see it works on Linux and that would be the biggest reason for me to consider their drives in my next upgrade run.

[HarlemSquirrel]

I would be curious to know how similar SSDs perform on PCIe 3.0 vs 4.0. Is it worth going from an x470 board to x570?

[oiaohm]

Phoronix: Seagate FireCuda 520 PCIe Gen4 NVMe SSD Linux Performance

For those that have been considering the Seagate FireCuda 520 as a PCI Express 4.0 NVMe solid-state drive, here are some benchmarks under Ubuntu Linux with this ZP500GM3A002 drive.

http://www.phoronix.com/vr.php?view=29365

I am sorry Michael Larabel we are getting more and more to the point that the fault that the XFS developer detailed in 2012 at LCA ( https://youtu.be/FegjLbCnoBw?t=1242 )with ext4 with excess IO and not well ordered IO will be messing with your results. Lot of your M.2 items are using multi bit per cell flash 3-4 per cell for long term storage and 1 or 2 bit per cell for fast writes. Yes excess IO not well ordered IO can completely flip your results because the SSD performance can be stalling as it attempt to move data from 1 or 2 bits to 3-4 bits per cell for long term storage that are only happening because the writes to drive were not well ordered.

I know it means more testing having to run the tests with ext4 and xfs this would kind of give best and worse cases.

In some ways we need a more general benchmark of raw device IO and raw device IO patterns than a file system and file systems bench-marked on their IO usage pattern.

[wizard69]

I would be curious to know how similar SSDs perform on PCIe 3.0 vs 4.0. Is it worth going from an x470 board to x570?

Well not now when we sill be getting new hardware in a few months.

Due to an old PC failure early this year I did build an X570 based machine at the beginning of the year. I went with a Samsung drive due to the lack of low thermals PCI-Express 4 drives at the time. Even then the performances is pretty shocking. Reboot times are so fast I hardly got upset over the teething pains for the newish 5500 GPU card.

[starshipeleven]

I was somewhat disappointed it did not detect my seagate spinning rust drives.

It's a SSD-only utility