Review – SSD Nextorage NEM-PA 8TB – The best solution for the PS5!

Today, we will be testing a top-of-the-line NVMe SSD from Nextorage, the NEM–PA model, which was kindly provided to us by Nextorage for review. Thank you! 🙂

It comes in the M.2 form factor with a 64Gbps interface, which means 4 PCIe 4.0 lanes, and uses the NVMe 1.4 protocol. Capacities range from 1TB to 8TB. The 8TB unit is typically priced between $869 and $1,099.

Specifications NEXTORAGE NEM-PA 8TB

Here are more detailed specifications about the SSD we will be testing (8TB unit):

SSD’s Software

Unfortunately, these SSDs do not have proprietary software, requiring third-party programs for their management.

Unboxing

On the front of the box, we see white packaging with an illustration of the SSD along with its capacity, as well as its respective sequential speeds and an image of a PlayStation 5, as it was designed in partnership with Sony to work on the console. On the back, there are only texts about the warranty and product information.

When you remove the SSD from the box, it comes in an anti-static plastic holder accompanied by installation tutorial material and warranty terms.

The SSD has a double-sided design, meaning that components are found on both sides of the PCB.

A positive aspect is that it already comes with a robust aluminum heatsink that helps with heat dissipation and is backward compatible with the PlayStation 5.

Its heatsink is 11 mm in height and is secured by 6 Philips screws, which join the upper part to the lower part, given that it has components on both sides.

Now, upon disassembling the SSD, we can better observe its internal components. An interesting detail we noticed was the attention to detail: the Phison E18 controller package is slightly smaller than the NANDs. To compensate for this, Sony/Nextorage placed a high-conductivity thermal pad on top of another thermal pad. This ensures better thermal conductivity since the controller is the main component and most sensitive to high temperatures.

Regarding its PCB, we can notice that on the front side it has 7 main components: the controller, the DRAM cache, 4 NAND flash modules, and the PMIC (Power Management Integrated Circuit) to power these components.

On the backside, there are 5 more components: 1 DRAM cache and 4 NAND flash modules.

Controller

The SSD controller is responsible for managing all data operations, over-provisioning, and garbage collection, among other background functions. This management ensures the SSD maintains good performance.

This SSD utilizes a high-end controller from Phison: the PS5018-E18-41, featuring a 32-bit ARM ISA (Instruction Set Architecture) with “5” Cortex® R5 cores (Penta-core) manufactured by TSMC using a 12nm process. It operates at a clock speed of 1 GHz on its main cores. This controller is commonly found in other top-tier SSD projects such as the Corsair MP600 HydroX, MP600 Pro, Galax Hall of Fame Extreme, Aorus Xtreme, and various other models including the Sabrent Rocket 4 Plus, which we have tested in the past.

In this case, we’re dealing with a triple-core controller, meaning it has 3 main cores responsible for managing the NANDs, with support for a technology called “CoXprocessors” – which is essentially another Dual-Core Cortex® R5 core with a significantly reduced frequency (usually between 200 and 300 MHz) aimed at performing simpler and predictive tasks. This helps alleviate the workload on the 3 main cores, as well as reduce power consumption and heat dissipation that could lead to thermal throttling, considering that this SSD can consume almost 9W. One of the functions of these processors is to handle repetitive code segments and firmware functions that the main cores wouldn’t need to perform, as well as manage data storage in the DRAM Cache, while the main cores are allocated for tasks such as Writing/Reading/Host.

This controller also supports up to 8 channels with a bus speed of up to 1600 MT/s, where each of these channels supports up to 4 Chip Enable commands, allowing the controller to communicate with up to 32 Dies simultaneously using the Interleaving technique.

DRAM Cache or H.M.B.

Every high-end SSD aimed at providing consistent high performance requires a buffer to store its mapping tables (Flash Translation Layer or Look-up table). This enables it to achieve better random performance and be more responsive.

We also see that this SSD comes with 2 chips as DRAM Cache from the manufacturer SK Hynix, which are used to store metadata tables. In this case, they are DDR4 type chips, models H5AN8G6NDJR-XNC, which operate at speeds of up to 3200 MT/s with CL-22 latencies.

However, due to the limitation of the memory controller of the Phison E18 itself, they are limited to up to 2666 MT/s, but with reduced latencies.

NAND Flash
Regarding its storage integrated circuits, the 8TB SSD features 8 NAND flash chips “TA8IG85AYV.” These are NANDs from the Japanese manufacturer Kioxia, formerly known as Toshiba, BiCS5 models. In this case, they are 1Tb (128GB) dies containing 112 layers of data and a total of 128 gates, resulting in an array efficiency of 87.5%. Out of the 128 layers in the SSD, 112 are allocated for storage, resulting in this efficiency.

In this SSD, each NAND Flash consists of 8 dies with a density of 1Tb, totaling 1TB per NAND, resulting in 8TB overall. They communicate with the controller using a maximum bus speed of 1200 MT/s for improved performance.

In this case, these dies, as per Kioxia’s publication at ISSCC 2019, offer performance of up to 132 MB/s using 4-planes along with a CuA (Circuit under the Array) architecture. However, as we will see next, the TLC-type dies of 512Gb (64GB) demonstrated at ISSCC 2019 are quite different from the ones used in the 8TB unit.

This happens because Kioxia decided to reduce the number of planes per die to focus on production scale and die scalability per wafer. After all, TLC dies with 4 planes would be completely unfeasible for them. Not only that, but these TLC dies, besides being only dual-plane, resulting in a significant performance loss, also lack the CuA architecture. Therefore, these dies would be able to deliver around 66 MB/s in an ideal scenario.

In this SSD, it was decided to use 1Tb dies because achieving an 8TB density with 512Gb dies would require 128 dies, which would lead to insane issues, both in terms of power consumption and thermal management. Additionally, it would exceed some M.2 standards, where the height of each NAND Flash could be higher than allowed due to the need to stack 16 dies per NAND Flash.

Therefore, Kioxia developed these 1Tb dies to try to work around these issues. In this SSD, we can observe that each NAND Flash is an 8 DP, meaning each NAND has 8 dies of 1Tb, resulting in each NAND Flash having a capacity of 1TB. To compensate for the reduced number of planes and the “nerfed” architecture, they increased the bus speed of these dies from 1066 MT/s, which was officially supported according to Kioxia, to a bus speed of 1200 MT/s.

Even so, even using 1Tb dies, we would have a total of 64 dies, which is a greater quantity than the controller supports. Therefore, it was necessary to adopt the Multi-Die-Select (M.D.S) command, which we explained in more detail in the analysis of the 4TB version that can be found below (in Portuguese).

This is necessary because the Phison E18 controller can communicate with up to 32 dies using interleaving. Therefore, in the SSD’s Flash ID, it allocates 2048Gb/C.E., which means it allocates 2 dies (each with 1024Gb or 1Tb) per C.E. to achieve this capacity.

PMIC (Power Delivery)

Just like any electronic component that performs work, SSDs also have a level of power consumption that can vary from a few milliwatts to close to 10 watts, nearing the limit of some connectors or slots. The circuit responsible for all power management is the PMIC, which stands for “Power Management IC,” a chip responsible for providing power to other components.

By default, in Phison SSDs, we see that this SSD comes with the PMIC PS6108-22, which is commonly used in projects with the Phison E18 controller.

SSD Power States

As always mentioned in power consumption analyses, in this section, we will delve deeper into the power states of this SSD.

Of the 5 power states it has, there are 3 active ones with excellent latencies and 2 idle ones with higher latencies. Another interesting point to note is that the manufacturer decided to configure the SSD with a reasonable thermal throttling temperature, ranging from 84ºC to 89ºC, unlike other SSDs we’ve tested, which reached ridiculously high temperatures of over 100ºC.

CURIOSITIES: SSD NEXTORAGE

Similarly to how integrated circuits on a RAM module can vary, the same happens with SSDs, where there are cases of component changes such as controllers and NAND flash chips.

Although there aren’t different versions of the SSD with component variations among the same capacities, the 8TB unit, up to the time of this analysis, was the only one using 1Tb dies from Kioxia instead of using Micron B47R dies as seen in the 1TB, 2TB, and 4TB units.
TEST BENCH
– OS: Windows 11 Pro 64-bit (Build: 23H2)
– CPU: Intel Core i7 13700K (5.7GHz all core) (E-cores e Hyper-threading desabled)
– RAM: 2 × 16 GB DDR4-3200MHz CL-16 Netac (c/ XMP)
– Motherboard: MSI Z790-P PRO WIFI D4 (Bios Ver.: 7E06v18)
– GPU: RTX 4060 Galax 1-Click OC (Drivers: 537.xx)
– (OS Drive): SSD Solidigm P44 Pro 2TB (Firmware: 001C)
– DUT SSD: SSD Nextorage NEM-PA 8TB (Firmware: EIFS51.3)
– Chipset Driver Intel Z790: 10.1.19376.8374.
– Windows: Indexing disabled to avoid affecting test results.
– Windows: Windows updates disabled to avoid affecting test results
– Windows: Most Windows applications disabled from running in the background.
– Boot Windows: Clean Image with only Drivers
– Test pSLC Cache: The SSD is cooled by fans to prevent thermal throttling, ensuring it doesn’t interfere with the test results.
– Windows: Antivirus disabled to minimize variation in each round.
– DUT SSDs: Used as a secondary drive, with 0% of space being utilized, and other tests conducted with 50% of space utilized to represent a realistic scenario.
– Quarch PPM QTL1999 – Power consumption test: conducted with three parameters—idle, where the drive is left as a secondary, and after a period of idle, a one-hour write test is performed, and the average power consumption is recorded

WHERE TO BUY

In the links below you can choose which capacity to buy throught amazon.

Amazon – SSD Nextorage NEM-PA 1TB – U$89.99

Amazon – SSD Nextorage NEM-PA 2TB – U$159.99

Amazon – SSD Nextorage NEM-PA 4TB – U$319.99

Amazon – SSD Nextorage NEM-PA 8TB – U$869.99

CRYSTALDISKMARK
We performed synthetic sequential and random tests with the following configurations:

Sequential: 2x 1 GiB (Blocks 1 MiB) 8 Queues 1 Thread

Random: 2x 1 GiB (Blocks 4 KiB) 1 Queue 1/2/4/8/16 Threads

In these sequential tests, we observed that the SSD performs averagely in its read speed compared to others. However, its write speed lags behind, mainly because Kioxia BiCS5 NANDs don’t typically offer such high sequential performance, which is further exacerbated by the 1Tb dies and dual planes.

In terms of latency, one area where Kioxia BiCS5 excels is in random performance and latencies. In this test, we see that its read latency was slightly higher than others, but the SSD showed excellent results in writing.

At QD4, we observe the same pattern: its read speed is lower than others, but its write speed surged ahead, taking first place, almost reaching 1400 MB/s.

Once again, now at QD1, we see a much worse result in its read speed, 69 MB/s, which can be achieved by good-quality PCIe 3.0 x4 SSDs. But in its write speed, it excelled, as of the time of this analysis. No SSD surpassed it in terms of small QD random write, not even the Solidigm P44 Pro.

ATTO Disk Benchmark QD1 and QD4

We conducted a test using ATTO to observe the SSDs’ speed across different block sizes. In this benchmark, it was configured as follows:

Blocks: from 512 Bytes to 8 MiB

File Size: 256MB

Queue Depth: 1 and 4.

The ATTO Disk Benchmark is software that performs a sequential speed test with compressed files, simulating a data transfer load as seen in Windows. Typically, we observe performance around block sizes of 128KB to 1 MiB. Now, we see that it exhibits a performance somewhat similar to the Aigo SSD, which features the new Silicon Motion controller, the SM2268XT, which showed remarkable results in our recent analyses.

Now, at QD1, it fell below the SM2268XT in its read performance but lands in the middle among other SSDs, which we can consider a technical tie. Meanwhile, in write performance, it leads in very small block sizes and then ties again with other SSDs.

3DMark – Storage Benchmark

In this benchmark, various storage-related tests are conducted, including game loading tests such as Call of Duty Black Ops 4, Overwatch, recording and streaming with OBS of a 1080p 60 FPS gameplay, game installations, and file transfers of game folders.

In this benchmark, since it uses a controller similar to the Kingston SSDs we tested, one would expect it to have similar performance. However, this is not the case due to the NANDs we mentioned, which, although they have good performance, are inferior in multiple scenarios compared to the Micron B47R found in the Kingston Fury Renegade we tested.

PCMARK 10 – FULL SYSTEM DRIVE BENCHMARK

Here, the Storage Test tool and the “Full System Drive Benchmark” were used, which perform both light and heavy tests on the SSD.

Among these traces, we can observe tests such as:

• Boot Windows 10
• Adobe After Effects: Launching the application until it’s ready to use
• Adobe Illustrator: Launching the application until it’s ready to use
• Adobe Premiere Pro: Launching the application until it’s ready to use
• Adobe Lightroom: Launching the application until it’s ready to use
• Adobe Photoshop: Launching the application until it’s ready to use
• Battlefield V: Loading time until the start menu
• Call of Duty Black Ops 4: Loading time until the start menu
• Overwatch: Loading time until the start menu
• Using Adobe After Effects
• Using Microsoft Excel
• Using Adobe Illustrator
• Using Adobe InDesign
• Using Microsoft PowerPoint
• Using Adobe Photoshop (Intensive use)
• Using Adobe Photoshop (Lighter use)
• Copying 4 ISO files, 20GB total, from a secondary disk (Write Test)
• Performing a copy of the ISO file (Read-Write Test)
• Copying the ISO file to a secondary disk (Read)
• Copying 339 JPEG files (Photos) to the tested disk (Write)
• Creating copies of these JPEG files (Read-Write)
• Copying 339 JPEG files (Photos) to another disk (Read)

In this benchmark, focused more on writing, it ends up with a worse result than other SSDs, but that’s because of the NANDs the SSD uses.

Adobe Premiere Pro 2021
Next, we used Adobe Premiere to measure the average time it takes to open a project of around 16.5GB with 4K resolution, 120Mbps bitrate, and filled with effects until it was ready for editing. It’s worth noting that the tested SSD is always used as a secondary drive without the operating system installed, as this could affect the result, causing inconsistencies.

When using Premiere to load a project of over 16GB, it performed well, with a low loading time, being equal to the Kootion we tested earlier.

WINDOWS BOOT AND GAME LOADING TIMES
Here’s a comparison between multiple SSDs and an HDD, using a clean installation of Windows 10 Build 21H1 along with the Final Fantasy XIV benchmark opening the campaign mode. The test consists of the best result after three consecutive system boots, considering the total time until reaching the desktop with the score reported by the application, so it’s slower than the boot until showing the desktop screen.

When testing game loading times, we see that in this benchmark, it presents results more consistent with SSDs of similar construction, performing 1 second faster than the KC3000.

In this program, it includes the time from boot to the loading of the latest OS drivers, which in this case, is done with a clean installation with only operating system drivers such as Network, Wireless + Bluetooth, Audio, Nvidia Drivers, PCH, among others. It showed good results that were below 20 seconds.

SLC CACHING

A good portion of SSDs on the market currently use SLC Caching technology, where a certain percentage of their storage capacity, whether it be MLC (2 bits per cell), TLC (3 bits per cell), or QLC (4 bits per cell), is used to store only 1 bit per cell. In this case, it is used as a read and write buffer, where the controller initiates the writing process, and when the buffer is depleted, it writes to the native NAND Flash (MLC/TLC/QLC).

Through IOmeter, we can get an idea of the SLC cache volume of this SSD, as the manufacturer often does not provide this information. Based on the tests we conducted, it was observed that it has a dynamic and large pSLC Cache volume of approximately 880GB. It managed to maintain an average speed of ~6583MB/s until the end of the buffer.

After recording 880GB, it starts writing to its NANDs in native TLC mode, and in this scenario, it operates at an average speed of 1031 MB/s, which is a good speed, although not as spectacular as the results from 2TB or 4TB units, which can likely achieve speeds of more than 3 times this velocity. It remains at these speeds for a good amount of time, writing over 5700GB or 5.7TB at speeds above 1GB/s.

Right after exhausting its space due to the SLC Cache, it begins the folding process because it has allocated all its capacity to work as pSLC. So, now we see the true Achilles’ heel of SSDs. However, its sustained speed was somewhat low, averaging 670 MB/s, which is quite typical for these 8TB SSDs with these Kioxia BiCS5 1Tb dies.

We also conducted a test to see how long it would take for the SSD to recover part of its buffer. Throughout our test battery, which lasts from 30 seconds to 2 hours in idle, using TRIM and garbage collection versus not using TRIM/GC, we observed that without using TRIM/GC, it couldn’t recover any of its SLC Cache. While this may seem like a negative point, it doesn’t necessarily qualify as poor SLC Cache design. It simply indicates that in this unrealistic scenario, it couldn’t utilize this opportunity.

But when tested with TRIM/GC activated, it was able to recover its full volume in just a few seconds.

FILE COPY TEST

In this test, files were copied from a RAM Disk to the SSD to see how it performs. The Windows 10 21H1 ISO file of 6.25GB (1 file) and the CSGO installation folder of 25.2GB were used.

When using the Windows 10 .ISO image, it yielded impressive results, as it was the SSD with the shortest transfer time for this small file so far. Although the difference wasn’t significant between one SSD and another.

With larger files, it still achieves good results even though it didn’t maintain its position in first place.

Temperature Test

In this part of the analysis, we will observe the SSD temperature during a stress test, where the SSD receives files continuously, to determine if there was any thermal throttling with its internal components that could lead to bottleneck or performance loss.

As seen above, this SSD has a default thermal throttling limit of 84°C to 89°C, which is a decent value. Through the sensors, we see that it exceeds 79°C, causing the controller to enter one of the Thermal Throttling states where its speeds drop from 1000 MB/s of sustained writing to approximately 600 MB/s. This represents an initial state of thermal throttling.

Indeed, the heatsink that it comes with truly brings several benefits because without it, the SSD would certainly have experienced more severe thermal throttling.

POWER CONSUMPTION AND EFFICIENCY

SSDs, like many other components in our system, have a certain level of power consumption. The most efficient ones can perform tasks quickly while consuming relatively little power, allowing them to transition back to idle power states where they tend to consume less.

In this part of the analysis, we’ll use the Quarch Programmable Power Module sent to us by Quarch Solutions (pictured above) to conduct these tests and assess how efficient the SSD is. This methodology will include three tests: the maximum power consumption of the SSD, an average power consumption in practical and casual scenarios, and the power consumption while idle.

This test suite, particularly the efficiency and idle power consumption tests, is especially important for users intending to use drives in laptops. SSDs spend the vast majority of their time in low-power states (Idle), so this data is crucial for optimizing battery life.

As we observed in its power consumption, it was to be expected that this SSD would not break any records in terms of electrical efficiency, as it has a very high maximum and average power consumption. However, its bandwidth in the benchmark was also one of the highest we’ve seen so far.

With regard to its maximum power consumption, we can see that it is quite high but still falls below other SSDs like the new Sabrent, as they have a much higher bandwidth.

During this benchmark, we see that these Phison projects with Kioxia NANDs in high-density SSDs really leave something to be desired, as the average power consumption is very high, being 35% higher than the Sabrent with Phison E18 and Micron B47R.

Lastly and most importantly, in the idle test, which is the scenario where the vast majority of SSDs find themselves in daily or routine use, it achieves a decent result for a high-density SSD like this one. High-capacity SSDs, especially those with the Phison E18 controller, tend to have quite high power consumption.

Conclusion

This SSD performed well, but it is far from being the fastest Gen4 SSD in the world. Its most advantageous feature is that it was specifically designed for use with the PlayStation 5, which is pretty cool, although it can also be used on other platforms. Despite this, it also has some drawbacks.

ADVANTAGES

• Good sequential speeds
• The best random write performance I’ve tested to date.
• Good latency results, especially in writing, surpassing all others I’ve tested.
• Great practical performance for casual scenarios and even professional environments like video editing, and excellent for the PS5.
• No variants with different components.
• Excellent internal construction, with a high-quality controller and NAND flashes.
• Immense pSLC Cache volume.
• Decent sustained write speed post-SLC Cache, although not the highest.
• Well above average durability.
• 5-year warranty.
• Reasonable idle power consumption.
• Interesting price for an 8TB SSD.

DISADVANTAGES

• Suffers slight thermal throttling under very extensive loads
• High power consumption
• Low energy efficiency due to high consumption
• SLC Cache takes time to recover (not too much in practice)
• No management software included
• No encryption