It's no secret that Silicon Motion is preparing its SM2508 PCIe 5.0 SSD controller to power the best SSDs. The company shared more information on SM2508 at the Memory Trend Seminar 2023 event in Shenzhen, China.
A previous roadmap showed that the SM2508 would be made on a 12nm manufacturing process. Silicon Motion has officially confirmed that the upcoming PCIe 5.0 SSD controller leverages the TSMC's 6nm EUV manufacturing process. The SM2508 is on a more advanced process node than the competition, such as the PS5026-E26 and the Innogrit IG5666 PCIe 5.0 SSD controllers on the 12nm process node. The 6nm process node offers a 45% power reduction compared to the 12nm process node, and up to 20% less logic cell area than the 7nm process node.
The SM2508 features an Arm-based processor with four Cortex-R8 cores up to 1.25 GHz and one Cortex-M0 core. The configuration provides a balance between performance and power consumption. It contrasts with the Phison E26, which has a dual-CPU design with Cortex-R5 cores in conjunction with Phison's CoXProcessor 2.0 accelerators. Nonetheless, Silicon Motion's material mentions the usage of an intelligent PMC (power management controller) on the SM2508 to assess and manage the power so that the chore isn't placed on the Arm chip.
Like most PCIe 5.0 SSD controllers, the SM2508 features an eight-channel design supporting DRAM speeds up to 3,600 MT/s — leaping over the Phison E26, which tops at 2,400 MT/s. It supports both TLC and QLC NAND flash chips. According to Silicon Motion, the SM2508 delivers sequential read and write speeds up to 14.5 GB/s and 14 GB/s, respectively. It seems that the company optimized the sequential read performance on the SM2508 since it was previously reported to only hit 14 GB/s. Random read-and-write performance is rated for up to 2.5 million IOPS. In comparison, the Phison E26 has peaked at 14 GB/s with Micron's highly sought-after B58R 2,400 MT/s NAND.
While the SM2508 is ultra-fast on paper, the PCIe 5.0 SSD controller's greatest attraction is its super low power consumption — Silicon Motion claims that the SM2508 draws less than 3.5W. The issue, however, is that the manufacturer doesn't specify whether that value corresponds to the SM2508's idle, average, or peak power draw. For comparison, the Phison E26 draws around 5W, meaning the SM2508 seemingly consumes up to 30% less power. It's plausible that SM2508-based SSDs may not require the same grade of active cooling as Phison E26-powered drives. However, we'll need to test the former when the SSDs become available because the SSD controller's power consumption is only one aspect since you still have to factor in the power draw from the 3D NAND and DRAM.
The SM2508 sounds excellent, but we should wait for reviews before concluding because real-world performance doesn't always equal what's on paper. The SM2508 is scheduled to come out before the end of this year, so we won't have to wait long to see if Silicon Motion's SSD controller can disrupt the PCIe 5.0 market, which predominately utilizes the Phison E26. Unfortunately, laptop users will have to wait longer as Silicon Motion doesn't foresee PCIe 5.0 SSDs making their way to laptops until late 2024.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Zhiye Liu is a news editor and memory reviewer at Tom’s Hardware. Although he loves everything that’s hardware, he has a soft spot for CPUs, GPUs, and RAM.
-
Order 66 Pretty soon, water cooling our SSDs will be a necessity if we don't get more efficient controllers like this one.Reply -
cyrusfox
As this is 30% more efficient than the other PCIE gen 5 controller on the market, this actually means we will see much better power/performance.Order 66 said:Pretty soon, water cooling our SSDs will be a necessity if we don't get more efficient controllers like this one.
Also highest power draw is only 15W. a passive heat sink is more than sufficient, WC is Overkill -
USAFRet "30% less" can be misleading.Reply
30% less than 50 watts is significant.
30% less than 5 watts (as seems to be the case here)....not so much.
5W -> 3.5W does not really matter to your PSU or cooling situation. -
williamcll
Pretty important for laptops when every watt matters.USAFRet said:"30% less" can be misleading.
30% less than 50 watts is significant.
30% less than 5 watts (as seems to be the case here)....not so much.
5W -> 3.5W does not really matter to your PSU or cooling situation. -
bit_user
In an era when fast PCIe 5.0 SSDs need bulky heatsinks and active cooling to avoid throtting, 30% is definitely significant. Even if it's a max of 1.5W savings, that's still significant even for a drive using 10 W. It could mean the difference between throttling or not. It could mean using a cheaper, passive heatsink (or an even smaller active one).USAFRet said:"30% less" can be misleading.
30% less than 50 watts is significant.
30% less than 5 watts (as seems to be the case here)....not so much.
5W -> 3.5W does not really matter to your PSU or cooling situation.
Heat also shortens the life of NAND. So, there's that benefit, as well. -
Order 66
That is true right now, but if future (PCIE gen 6+) drives get faster (which they will) they will probably require more power until a more efficient controller comes out for those drives, meaning that WC may be necessary in the future. My point is that the trend has been somewhat worrying, we went from not needing any cooling (or very little) for PCIe 3 and 4 drives to now needing active cooling for some PCIe 5 drives.cyrusfox said:As this is 30% more efficient than the other PCIE gen 5 controller on the market, this actually means we will see much better power/performance.
Also highest power draw is only 15W. a passive heat sink is more than sufficient, WC is Overkill -
bit_user
I don't foresee PCIe 6.0 coming to consumer platforms any time soon. Although it doesn't increase the clock frequency, it does require a higher signal-to-noise ratio, which increases board costs.Order 66 said:That is true right now, but if future (PCIE gen 6+) drives get faster (which they will) they will probably require more power until a more efficient controller comes out for those drives,
IMO, we're likely to see CXL first. If & when that happens, maybe client devices will eventually get PCIe 6.0 / CXL 3.0, which share the same PHY standard. That's a long ways off - I think the next generation of server CPUs don't even have those technologies, yet.
Yes, it's not a good trend. However, SSD performance has been increasing a lot faster than power consumption. If there's a pause on performance improvements, then efficiency can (mostly) catch up.Order 66 said:My point is that the trend has been somewhat worrying, we went from not needing any cooling (or very little) for PCIe 3 and 4 drives to now needing active cooling for some PCIe 5 drives. -
Order 66
The PC industry is (generally) obsessed with things going faster, I highly doubt (at least for now) that SSD manufactures would pause on gen 5 performance since they haven't been on the market for even 6 months yet.bit_user said:If there's a pause on performance improvements, then efficiency can (mostly) catch up.