Solidigm’s datacenter SSD lineup consists of fashions focusing on totally different efficiency, endurance, and value tradeoffs. Last 12 months, the corporate had launched the D5-P5336 QLC drive as a low-cost high-capacity drive for read-heavy workloads, whereas additionally making ready the SLC-based D7-P5810 for terribly write-intensive workloads requiring excessive endurance. The D7-P5520 / D7-P5620 Gen Four drives with Solidigm’s personal 144L 3D TLC have been the high-performance choices for generic workloads over the past couple of years.
Solidigm is saying the supply of the successor to the D7-P5x20 immediately – the brand new D7-PS1010 and D7-PS1030. Both of those NVMe drives use SK hynix’s 176L 3D TLC NAND and include a PCIe 5.Zero interface. The third digit within the mannequin quantity matches the DWPD score, with the D7-PS1010 focusing on blended workloads with a 1 DWPD score, and the D7-PS1030 focusing on write-intensive use-cases with a Three DWPD score.
Compared to the earlier technology D7-P5x20, the D7-PS10x0 sequence brings concerning the following upgrades:
- Move from PCIe 4.Zero x4 to PCIe 5.Zero x4
- Move from 144L floating gate 3D TLC (Solidigm) to 176L cost lure 3D TLC (SK hynix)
- 25% longer mean-time between failures (MTBF) at 2.5M hours
- 10x greater uncorrectable bit-error price (UBER) at 1E-18
- 1.8x to 2.8x enchancment in excessive queue-depth random entry IOPS
- 2.0x to 2.2x enchancment in excessive queue-depth sequential entry throughput
The specs of the 2 new SSD households are summarized within the desk beneath.
| Solidigm D7-PS1000 Series Enterprise SSDs | |||
| D7-PS1030 | D7-P1010 | ||
| Form Factor | U.2 2.5″ 15mm E3.S 7.5mm |
||
| Interface | PCIe 5.Zero NVMe 2.0 | ||
| Capacities | 1.6TB 3.2TB 6.4TB 12.8TB |
1.92TB 3.68TB 7.68TB 15.36TB |
|
| NAND | SK hynix 176L 3D TLC (Charge Trap Architecture) | ||
| Sequential Read (128 KB @ QD 128) | 14500 MB/s | ||
| Sequential Write (128 KB @ QD 128) | 4100 MB/s (1.6 TB / 1.92 TB) 8200 MB/s (3.2 TB / 3.84 TB) 9300 MB/s (6.Four TB / 7.68 TB / 12.eight TB / 15.36 TB) |
||
| Random Read (Four KB @ QD 512) | 2.35 M (1.6 TB / 1.92 TB) 3.1 M (3.2 TB / 3.84 TB) 2.eight M (6.Four TB / 7.68 TB) 2.75 M (12.eight TB / 15.36 TB) |
||
| Random Write (Four kB) | 0.35 M (1.6 TB) 0.716 M (3.2 TB) 0.eight M (6.Four TB / 12.eight TB) |
0.15 M (1.92 TB) 0.315 M (3.84 TB) 0.Four M (7.68 TB) 0.38 M (15.36 TB) |
|
| Power | Sustained Write | 13 W (1.6 TB / 1.92 TB) 18 W (3.2 TB / 3.84 TB) 23 W (6.Four TB / 7.68 TB / 12.eight TB / 15.36 TB) |
|
| Sustained Read | 17 W (1.6 TB / 1.92 TB) 19 W (3.2 TB / 3.84 TB) 22 W (6.Four TB / 7.68 TB) 23 W (12.eight TB / 15.36 TB) |
||
| Peak | 18 W (1.6 TB / 1.92 TB) 22 W (3.2 TB / 3.84 TB) 29 W (6.Four TB / 7.68 TB) 30 W (12.eight TB / 15.36 TB) |
||
| Idle | 5 W | ||
| Write Endurance | Three DWPD | 1 DWPD | |
| Warranty | 5 years | ||
Based on Solidigm’s personal inside testing, the D7-PS1010 compares very favorably in opposition to the Gen 5 datacenter SSDs already out there from Samsung and Kioxia. However, the not too long ago launched Micron 9550 sequence might current a greater problem to Solidigm’s claims.
Gen 5 SSDs are well-suited for the storage-intensive duties in AI workloads. Every new product must tie itself to the AI buzzword at the moment, however we should always excuse SSD producers for doing the identical – in any case coaching and inference wants to maneuver massive quantities of knowledge backwards and forwards between the processing engine and underlying reminiscence. Solidigm expects the D7-PS10x0 to be a great match as direct-attached storage inside to GPU servers or as all-flash tier supporting a HDD-only object tier within the cloud. For on-premises GPU servers, the flash / HDD tiered storage will be changed by an all-QLC object tier.
Solidigm claims higher vitality effectivity in comparison with the opponents’ Gen…
