Sabrent’s range of internal and external SSDs has gained popularity among tech enthusiasts, largely due to the company’s knack for being at the forefront of introducing products equipped with the latest controllers, offering both cutting-edge technology and good value for money. A key component in their strategy is the collaboration with Phison, whose controllers drive many of Sabrent’s offerings. With the Rocket nano V2, a portable SSD offering 2 GBps-class speed and powered by Phison’s U18 native controller, Sabrent has been making a solid statement in the market for over a year now.
In this review, we take a comprehensive look at the Rocket nano V2 External SSD, focusing on aspects like performance consistency, power consumption, and thermal efficiency.
Introduction and Product Impressions
Over the past decade, the rapid evolution of flash technology—including advancements in 3D NAND and improvements in TLC reliability—has coincided with the development of faster host interfaces for external devices. This synergy has fueled the progress of bus-powered direct-attached storage, enabling slimmer devices with higher storage capacities and speeds. The Type-C standard is widely embraced in the consumer market, paving the way for compact flash-based storage options capable of hitting 2 GBps+ speeds through protocols like USB 3.2 Gen 2×2 / USB4 and Thunderbolt.
However, high-speed storage devices face significant challenges related to heat management. Traditional bridge solutions, which rely on multiple protocol conversion chips, tend to generate more heat due to these additional components. Previously, high-performance portable SSDs had no choice but to use these, progressing from SATA to NVMe bridges. The introduction of native UFD controllers by Phison and Silicon Motion, capable of achieving speeds of 10 Gbps and 20 Gbps respectively, has presented a new path forward in this segment. The Crucial X6, featuring the Phison U17, marked a significant milestone by surpassing SATA speeds without a NVMe bridge, setting a new benchmark at 800 MBps. Around this time, the Silicon Motion SM2320 made headlines with the Kingston XS2000, reaching 2 GBps speeds without a bridge.
Products leveraging Silicon Motion’s SM2320 have struck a chord with consumers, often maxing out interface speed limits for sequential operations in the 10 Gbps and 20 Gbps categories. While Phison’s U17 and U18 controllers initially boasted slightly lower peak performances, recent improvements have allowed SSDs with Phison’s native controllers to catch up, with increased adoption noted. The PNY EliteX-PRO, Sabrent Rocket Nano V2, and Corsair EX100U all use Phison’s U18 controller. Despite the underwhelming performance of the EliteX-PRO, we reached out to Sabrent for a sample of the Rocket nano V2 to evaluate if it could parallel the success of the OWC Envoy Pro Mini with the U18 controller.
Sharing the same hardware foundation—Phison’s U18 controller with SK hynix’s 128L 3D TLC NAND—the PNY EliteX-PRO, Corsair EX100U, and the Sabrent Rocket nano V2 feature notable differences in firmware, thermal management, and chassis design.
The Rocket nano V2 measures a compact 72 mm x 32 mm x 14 mm, and thanks to its aluminum body and weight of 50 grams, it feels both sturdy and premium. The addition of a silicone cover further protects the metal surface from scratches and dents, while enhancing its ruggedness.
A nifty feature is the Rocket nano V2’s S.M.A.R.T passthrough capability, visible in the CrystalDiskInfo screenshots. While TRIM isn’t explicitly advertised, we successfully verified its functionality on an NTFS volume processed by Windows. All S.M.A.R.T features, temperature readings included, performed flawlessly.
The following section provides a side-by-side comparison of specifications for the various portable SSDs reviewed.
Testbed Setup and Evaluation Methodology
For evaluating direct-attached storage devices, we used the Quartz Canyon NUC, equipped with Intel’s Xeon / ECC configuration, 2x 16GB DDR4-2667 ECC SODIMMs, and an ADATA IM2P33E8 PCIe 3.0 x4 NVMe SSD.
The Quartz Canyon NUC stands out for having two PCIe slots available for additional cards, thanks to the absence of a discrete GPU. We also inserted a spare SanDisk Extreme PRO M.2 NVMe SSD to bypass any potential DMI bottlenecks during Thunderbolt 3 device testing. Given the lack of a native USB 3.2 Gen 2×2 port, we installed a Silverstone SST-ECU06 card to enable Type-C testing.
The testbench’s components are detailed below:
2021 AnandTech DAS Testbed Configuration
- System: Intel Quartz Canyon NUC9vXQNX
- CPU: Intel Xeon E-2286M
- Memory: 32 GB (2x 16GB) ADATA DDR4-3200 ECC
- OS Drive: ADATA IM2P33E8 NVMe 1TB
- Add-on Card: SilverStone Tek SST-ECU06 USB 3.2 Gen 2×2 Host
- OS: Windows 10 Enterprise x64 (21H1)
We extend our thanks to ADATA, Intel, and SilverStone Tek for providing the components.
The evaluation of these direct-attached storage devices covers multiple workloads reflecting current trends like high bit-rate 4K video at 60fps and burgeoning 8K recordings. Game installations on portable consoles have also grown considerably, driven by high-definition textures and art. Our assessment includes a range of workloads, described further in the coming sections.
- Synthetic Benchmarks: Utilizing CrystalDiskMark and ATTO
- Real-world Access Traces: Using PCMark 10’s storage benchmark
- Custom File Transfer Tests: Employing robocopy-style tasks
- Sequential Write Stress Test
In the subsequent sections, we delve into the Rocket nano V2’s performance in these benchmarks, followed by insights on its power consumption and thermal management.
Synthetic Benchmarks – ATTO and CrystalDiskMark
ATTO and CrystalDiskMark are benchmarks used to quickly evaluate direct-attached storage device performance. They provide instantaneous performance metrics for specific tasks but may not reflect longer-term variability caused by things like thermal throttling. They also allow us to verify the presence of features affecting performance.
Sabrent claims that the Rocket nano V2 can hit 1500 MBps, corroborated by ATTO tests. While ATTO is limited to singular configurations like queue depth, it offers a snapshot of performance changes as I/O size shifts, with peak performance usually around 512 KB at a queue depth of 4.
CrystalDiskMark runs four different access scenarios, both read and write, spanning sequential and random 4K operations. Testing tools like Microsoft DiskSpd underpin its operations. Comparing certain 4K results helps identify if NCQ/UASP support is present. If ‘4K Q32T16’ numbers align closely with ‘4K Q1T1’, NCQ/UASP is likely unsupported, assuming the PC’s host port and drivers do support UASP.
The Rocket nano V2 shares similar performance profiles with the PNY EliteX-PRO in these synthetic benchmarks, due to platform similarities. For tasks involving smaller data sizes, the SM2320-based PSSDs outperform the U18-based options.
AnandTech DAS Suite – Benchmarking for Performance Consistency
Our testing for storage devices emphasizes their intended real-world uses. Details on the AnandTech DAS Suite and its composition are documented here.
In most workloads, the performance differences between models are negligible. Therefore, typical users shouldn’t notice a significant disparity in everyday usage, though power users might observe a preference for SM2320-powered PSSDs or bridge-based models for simple transfers. The key is understanding each device’s limitations, which our performance consistency evaluations seek to uncover.
Performance Consistency
Performance consistency is influenced by factors like SLC caching and thermal management. Power users, in particular, dislike seeing their transfer speeds drop to USB 2.0 territory while handling large data chunks.
Not only do we track read/write speeds during the AnandTech DAS Suite’s execution but we also monitor the drive’s temperature. Earlier reviews showed that constantly measuring NVMe SSD temperatures via USB 3.2 Gen 2 bridge chips could impair transfer rates. To circumvent this, we capture temperature readings only during idle times. Below graphs illustrate our findings.
During the AV segment of testing, the Rocket nano V2 remains relatively cool, peaking at 55 degrees Celsius, thanks to its aluminum construction and effective thermal design. This stands in stark contrast to the PNY EliteX-PRO, which struggles with thermal throttling, reaching 76 degrees Celsius during intensive disk-to-disk transfers. Subsequently, the Rocket nano V2 completes the workload at a noticeably faster pace.
PCMark 10 Storage Bench – Real-World Access Traces
Several storage benchmarks simulate artificial access traces with different read/write mix ratios, block sizes, and queue depths. Popular benchmarks like ATTO and CrystalDiskMark provide insights, yet real-world traces from actual tasks offer more practical significance. Real-world access traces measure performance impacts during scenarios like booting an OS or loading games.
PCMark 10’s storage bench leverages diverse real-world traces, detailed here.
Component Scores and Overall Scores
The Rocket nano V2 lands in the lower half across component scores, though the differences aren’t huge. Despite the synthetic benchmarks, the device ranks closely with the PNY EliteX-PRO, while SM2320-based models and bridge-driven solutions perform better in specific areas.
Miscellaneous Aspects and Final Thoughts
Overall, the Rocket nano V2’s consistency and performance remain evident across various workloads, both real-world and synthetic. Its power usage and thermal efficiency are crucial for power users concerned with long-duration tasks and energy consumption, particularly with battery-dependent devices.
Worst-Case Performance Consistency
Flash storage devices can lag significantly during exhaustive small-sized random write operations—something benchmarkers use to determine worst-case scenarios. Luckily, direct-attached storage tasks, often sequentially oriented, render these tests less generalizable. Sequential writes and thermal caps in DAS devices could lead to speed drops under constant load.
Our Sequential Writes test writes data sequentially to 90% of the drive’s space, testing performance consistency. The test captures temperatures and instantaneous write rates, revealing that the Rocket nano V2 performs notably better, sustaining near 800 MBps throughout and maintaining a final temperature of only 61C. It’s impressive when compared to the PNY EliteX-PRO, which sees write speeds tumble to 450 MBps and ends at a warmer 76C.
Power Consumption
Bus-powered storage devices must mind their power demands. While Thunderbolt ports offer up to 15W, USB 2.0 guarantees just 2.5W. Monitoring power usage reveals interesting aspects about the Rocket nano V2’s efficiency.
The Rocket nano V2 idles at a slightly elevated 0.76W, about 100 mW higher than other PSSDs with native controllers. Also of note, its sleep mode triggers only after 50+ minutes, unlike others that activate within 20 minutes. While not critical for desktops, this might affect notebook users.
Final Words
Priced around $182, similar to other 2 TB PSSDs, the Rocket nano V2 represents an attractive choice for many use cases. Although its value may not seem obvious based solely on peak transfer rates or short burst performance, it truly excels in long-term, sustained write scenarios, such as video recording. The thoughtful thermal design and protective silicone cover enhance its reliability and usability, making it a worthy option depending on user needs and preferences.
