CORSAIR MP600 Mini 1TB Gen4 M.2 SSD
More and more PC game consoles are on the market, they all use M.2 2230 SSD and the capacity is relatively small, so CORSAIR launched the new MP600 MINI 1TB Gen4 M.2 SSD for the PC game console market, using M.2 2230 size performance Supports next-generation PC game consoles such as Steam Deck / ROG Ally, adopts PHISON PS5021-E21 Gen4 master controller with Micron 176 Layers 3D TLC NAND particles, provides up to 4,800MB/s Read, 4,800MB/s, 4K random read and write The highest level can reach 850K IOPS, making your PC handheld have higher capacity and higher performance.
Received the MP600 Mini 1TB Gen4 M.2 SSD sample from CORSAIR for testing. The yellow and black outer box has a physical picture of the SSD body printed on the front, and multi-language instructions, speed specifications, and regulations in different countries are written on the back, and EAN Lable is printed on it. Made In Taiwan.
The 1TB version of CORSAIR MP600 Mini is CSSD-F1000GBMP600MN, which uses M.2 2230 Form Factor, PCI Express 4.0 x 4 transmission interface, supports NVMe 1.4 transmission specification, 1TB total writing capacity reaches 600 TBW, MTBF mean time between failures reaches 1,500,000 hours , the standby power consumption is about <5mW, and the load power consumption is 4.3W.
Using PHISON E21 controller
‘
It adopts PHISON PS5021-E21 Gen 4 main controller, focuses on the mainstream market, DRAMless design, uses 12nm process, built-in 32bit ARM Cortex R5 processor, has 4 Channels, 16CE data transmission channels, supports up to 1,600MT/s 3D TLC NAND particles bring 5GB/s speed and 800K IOPS performance.
Using a single Micron ICCIG94AYA 3D TLC NAND (B47R) FortisFlash particle, the speed is 1600MT/S, with up to 176 Layers, 16 512Gbit Dies are stacked inside, and the single capacity is up to 1TB (8Tbit). The maximum writing speed of each Die is 130MB/s, 5000 P/E Cycle write life, 40000 P/E SLC Mode write life.
Designed with DRAMless
Many people will misunderstand the purpose of DRAM Cache. It is mainly used to store the FTL logic-physical mapping table, and at the same time, temporarily store the fine data to be written in DRAM, and wait for a page to be written to the NAND particles at one time to improve IOPS. Performance, and DRAM cache can be used to do garbage collection and wear leveling. Without DRAM Cache, the FTL table must be stored in NAND Flash. Although the cost is reduced, the speed will be slower. Since continuous writing will not use DRAM Cache, so The difference with or without DRAM Cache is mainly in IOPS performance and response time.
However, NVMe has added Host Memory Buffer (HMB) technology since version 1.2. In Windows 10 / 11 operating systems, PCIe protocol can be used to borrow part of the system memory as Cache for storing FTL logical-physical mapping table and fragment The data to be written is temporarily stored to improve the performance of DRAMless SSD in terms of IOPS. Usually, 64MB to 128MB of capacity is borrowed. The memory called by this part will intelligently coordinate the data exchange between the host CPU and SSD according to the read and write conditions.
In fact, everyone has a bad impression of DRAMless SSD. The main reason is that the performance of the main controller of early DRAMless SSD is too poor, and it does not support HMB acceleration technology. In fact, when DRAMless starts HMB acceleration, the performance is not as good as that of real DRAM Cache, but it will definitely not lose too much. ugly.
If you want to test the difference in performance with and without HMB, you can make changes in the Registry on Windows 10/11:
In HKEY_LOCAL_MACHINESYSTEMCurrentControlSetControlStorPort, create a new DWORD value named HMBAllocationPolicy,
A setting of 0 disables HMB.
Set to 1 to enable and allocate the minimum value allowed by the device
Set to 2 to enable and 1 to assign the maximum expected by the device
Set to 3 to enable and dynamically adjust according to SSD usage
testing platform︰
CPU︰AMD Ryzen Z1 Extreme
MB︰ROG ALLY (2023) RC71L
RAM︰LPDDR5 16GB on board (6400MT/s) Dual Channel
SSD: CORSAIR MP600 MINI
OS︰Windows 11 Professional 22H2
ATTO Disk Benchmark test:
ATTO Disk Benchmark is a disk performance testing tool launched by ATTO. This test uses a data size of 256MB, and increases the data size to 64MB in multiples of 0.5K, 1K, and 2K to test the impact of different file sizes on the speed of disk writing. , and the test results are expressed in columnar graphs. It uses sequential writing to test the best read and write capabilities of the disk, which is the theoretical value benchmark test that marks the highest performance of products in the industry.
According to the test results, the maximum reading speed of the CORSAIR MP600 Mini 1TB version is about 4.70GB/s, and the writing speed is 4.90GB/s.
CrystalDiskMark test:
CrystalDiskMark is a set of disk test programs developed by Hiyohiyo, a Japanese. It provides continuous disk read and write speed tests and 4K random multiplex read and write performance tests with different queue depths (Queue Depth).
Tested with CrystalDiskMark version 8.0.4, the sequential read and write speeds of the CORSAIR MP600 Mini 1TB version Q8T1 are 4,819.83 MB/s Read and 4,847.52 MB/s Write; the sequential read and write speeds of Q1T1 are 3095.49 MB/s Read and 4,847.57 MB/s Write; 4K Q32T1 random multitasking read and write speeds are 1,059.63 MB/s Read and 842.69 MB/s Write; 4K Q1T1 random multitasking read and write speeds are 94.61 MB/s Read and 480.44 MB/s Write.
AS SSD Benchmark test:
AS SSD Benchmark is also one of the main benchmarks for testing SSDs in the current industry. On the premise of not using system memory acceleration, it tests the read and write performance and speed of SSDs in all aspects by measuring the reading and writing of large files of 1GB.
The performance benchmark score of CORSAIR MP600 Mini 1TB version in AS SSD Benchmark is 7878, of which the read performance score is 3,027, the write performance score is 3,266, and the 4K random read and write performance with a queue depth value of 64 is 2,502.39 MB/s Read and 2,428.29 MB/s Write.
Anvil’s Storage Utilities Test:
Anvil’s Storage Benchmark is a set of professional disk performance testing software. It has a built-in IO Threaded test program, which can analyze the IO performance of multitasking read, write, and mixed read and write of a disk with a queue depth of 4K 32.
In terms of IOPS performance, 4K QD128 Read recorded 663,342.8 IOPS, 4K QD64 Write recorded 751,870.9 IOPS, 4K QD64 Read/Write mixed read and write recorded 811,618.4 IOPS, and finally scored 25,555.23 points in Anvil’s Storage Benchmark test results.
AIDA64 Disk Benchmark full disk write test:
AIDA64 Disk Benchmark provides a full-disk write test. The test uses 4MB Block Size to write. The SLC Cache strategy of the SSD Firmware is measured in a clean and empty state, which truly reflects the capacity range of the SLC Cache. After the SLC Cache is saturated, it enters the TLC direct write mode. Extent, and write capacity has reached a critical point that requires three different phases of the Merge job.
It can be seen from the test that the SLC Cache of CORSAIR MP600 Mini 1TB has an optimal range of up to 25% capacity (232.87 GB), and recorded an average write speed of about 4.4 GB/s in this range.
When the disk reaches 25% to 60% capacity (232.87 GB to 558.94 GB), the SLC Cache is saturated and enters the TLC writing stage. The speed at this stage is the pure TLC real speed. This range recorded an average of about 1.7 GB/ s Write speed.
After the written capacity exceeds 60% (558.94 GB), the SLC Cache has reached the threshold set by the Firmware, and the main controller starts to enter the Merge operation, releasing the SLC Cache data back to the TLC space, and the SSD must process data writing at the same time And Merge jobs, so the writing speed will drop significantly. It can be seen that the writing speed will drop to about 250MB/s at this stage.
When the write capacity is between 80% and 90%, part of the SLC Cache data has been written back to the TLC space, and the speed at this stage will hover between the real speed of pure TLC and the speed of Merge operations.
It needs to be emphasized that many people will misunderstand that when the SSD is written to a certain capacity, the SLC Cache will no longer work. In fact, the current SLC Cache strategy is dynamically replenished. The SSD will execute the Merge job in the background when it is idle, and the SLC Cache The data is written back to the TLC space, and then the main controller divides the SLC Cache again in proportion according to the remaining capacity of the SSD, so that the next continuous write speed will return to full capacity, but the less the remaining capacity, the capacity of the SLC Cache will also decrease. Just drop faster when disc.
PCMark 10 Storage test:
The PCMark 10 Storage test simulates the speed of reading and writing programs such as document software, art software, and games on computer storage devices, and can reflect the performance of opening programs or saving files in real use.
In the PCMark 10 Storage test, the CORSAIR MP600 Mini 1TB scored 2,834. After combining different usage scenarios, its average Bandwidth is 459.76 MB/s.
3DMark Storage test:
The 3DMark Storage test simulates computer storage devices to read games, record games, install games, save games, etc., and can reflect the game performance of SSDs in real use.
In terms of 3DMark Storage Benchmark test, CORSAIR MP600 Mini 1TB scored 2,988. After combining different usage scenarios, its average Bandwidth is 511.03 MB/s.
CORSAIR MP600 Mini 1TB Gen4 M.2 SSD
Price: HK$959
Enquiry: Felton Distribution (2273-8393)
Editor’s comment:
Now that GAME is getting bigger and bigger, the capacity of the SSD attached to Steam Deck and ROG Ally is really not enough. You can consider this CORSAIR MP600 Mini 1TB Gen4 M.2 SSD to expand the capacity, which greatly improves the usability of the device.