ADATA Launches XPG Gammix S50: A PCIe 4.0 x4 SSD
by Anton Shilov on July 31, 2019 11:00 AM EST- Posted in
- SSDs
- Storage
- ADATA
- XPG
- Phison
- 3D TLC
- PCIe 4.0
- PS5016-E16
- XPG Gammix S50

ADATA has introduced its first PCIe 4.0 x4 NVMe SSD, which is aimed at the latest AMD Ryzen-based PCs using the AMD X570 platform. The ADATA XPG Gammix S50 will be the company’s new flagship drive for users looking for the maximum performance possible today.
Set to be available in 1 TB and 2 TB capacities, the XPG Gammix S50 drive is based on Phison’s PS5016-E16 controller (the only PCIe 4.0 SSD controller available today) paired with 3D TLC NAND memory. The drive comes in an M.2-2280 form-factor and features a PCIe 4.0 x4 interface (an, of course, the unit is backwards compatible with PCIe 3.0 systems). What is a bit surprising is that unlike other SSDs powered by the same controller, ADATA’s XPG Gammix S50 is not equipped with a large heat spreader, but comes with a relatively regular sized one.
As far as performance is concerned, ADATA says that the XPG Gammix S50 offers up to 5000 MB/s sequential read speeds and up to 4400 MB/s sequential write speeds when SLC caching is used (data based on CDM benchmark). Meanwhile, the SSD is rated for up to 750K random read/write IOPS.
When it comes to endurance and reliability levels, the XPG Spectrix S50 drives feature up to 1800 or 3600 TB written over a five-year warranty period, depending on the drive's capacity.
ADATA XPG Gammix S50 Specifications | ||
Capacity | 1 TB | 2TB |
Model Number | AGAMMIXS50-1TT-C | AGAMMIXS50-2TT-C |
Controller | Phison PS5016-E16 | |
NAND Flash | 3D TLC NAND | |
Form-Factor, Interface | M.2-2280, PCIe 4.0 x4, NVMe 1.3 | |
Sequential Read | 5000 MB/s | |
Sequential Write | 4400 MB/s | |
Random Read IOPS | 750K IOPS | |
Random Write IOPS | 750K IOPS | |
Pseudo-SLC Caching | Supported | |
DRAM Buffer | Yes, capacity unknown | |
TCG Opal Encryption | No | |
Power Management | 0.8W Active (Typical) 0.05W Slumber (Typical) DevSleep |
|
Warranty | 5 years | |
MTBF | 1,700,000 hours | |
TBW | 1800 TB | 3600 TB |
Additional Information | Link | |
MSRP | $258 | $484 |
ADATA’s XPG Gammix S50 will be available in the near future. Prices will vary by region.
Related Reading:
- Phison’s PS5016-E16 & PS5019-E19: The First PCIe 4.0 Client SSD Controllers
- Phison PS5016-E16 PCIe 4.0 SSDs: Random Read/Write Performance Disclosed
- TeamGroup Cardea Zero PCIe 4.0 x4 SSD: Up to 1 TB
- More 2TB PCIe 4.0 SSD Options: Patriot with Custom Firmware coming Q4
- Galax Goes White: A PCIe 4.0 2TB SSD with a White PCB
Source: ADATA
13 Comments
View All Comments
npz - Wednesday, July 31, 2019 - link
When I'm working with video files on my older laptop which only has 850 EVO SSD, I actually wish I had NVME for those files. For operations not cpu bound, like stitching video without transcoding, editing or converting containers, it can saturate full bandwidth ~540 MB/s+ and I simply take a break if I'm working with many files or very large files.The problem comes with files a few GB. It's too short to actually take a break or switch to something else, yet long enough that I'm just waiting in front of the screen for like 8-10 secs. With nvme I can cut it down to 1-2s for each file or operation. Reply
tygrus - Thursday, August 1, 2019 - link
Yes, we ask too much and we are seeing diminishing returns for the extra performance in real world usage. A few edge cases remain for some to make use of the high performance. Others just need to optimise their software better.The first few non-SATA drives took a while before they got over R/W 1500/1000 MB/s. Current limitations: thermal limit of controller; limited number of chips & channels per 2280/22110 stick; storage chip internal and external data rates/speeds; PCIe to CPU to RAM and back again limitations. Everything needs another few generations and they will be closer and closer to the practical limit of 7200MB/s. VM host servers will love it, most IT users will be non-the-wiser. Reply
msroadkill612 - Thursday, August 1, 2019 - link
I agree nvme speeds are pointless. Others dont go far enough. We should all underclock our ram by a factor of 5x too - our machines will seem to boot almost as fast - so there is your proof. Reply