Large capacity, high and durable - CRCUIAL Ying Ruida MX500SSD4TB version is completely interpreted

Home > Sci-Tech

Large capacity, high and durable - CRCUIAL Ying Ruida MX500SSD4TB version is completely interpreted

2021-11-26 06:03:55 4 ℃

Last month, I evaluated 1TB MX500. I didn't expect that I didn't unexpectedly got a 4TB ultra-large version. Try to read the official website of Ying Ruida, the MX500 capacity has always been extended from 250GB to 4TB, and she has to admire this four-year series of strong vitality. Since the delay, the MX500 Series has got a total of more than 16 awards worldwide, and it can also enjoy a five-year warranty period, which is quite rare in the SATA hard drive series.

Let's review the characteristics of 1TB products, first is the main control, flash particles continue to upgrade with mainstream technology, and the MX500 is a high-end product in the SSD product, which has retained independent cache. As a result, it is even better than the reading and writing performance in some real scenes, or in other words, some of the read and write scenarios of low-end NVME will fall to SATA 6Gbps of 600MB / s. The interface is under extreme throughput, and the MX500 can be 560Mb / s in many times. Coupled with the price-priced factor, this is actually the survival of SATA SSD.

Suddenly got a 4TB SATA SSD, I was a bit awkward. Is it necessary to achieve a dream? I wanted to give myself a large-capacity mobile hard disk. At that time, the price of SSD was basically 1.2 to 1.5 yuan for 1GB, and the large capacity product is basically very small, so the storage product of 4TB or more, I can only accept HDD. price. The result is actually expected, accidental falling in use, so that I have to spend a lot of time energy to recover data. This 4TB SSD is first a good choice as a storage medium for an external hard disk.

For ordinary game users, audio and video enthusiasts, a 4TB hard disk means that there can be loaded into more than 60 "Sai Bo Depens 2077", which also means that it can be loaded into more than 80 4K original formats. Movies, such as BDMV, BDISO, REMUX, etc. Storage space is meaningless). In fact, an SSD that has a large enough capacity is quite large.

If you are a heavy user on a storage, I will give you a better purchase capacity. It seems that small capacity products are "enough", it means capacity, there is a parameter I think that ordinary users should also understand, that is TBW (Total Bytes Written) - the durability, or write life . To know, this parameter is connected with the warranty period, just like the car, such as "50,000 kilometers 3 years Quality guarantee", either reach insolvenous.

For SSD, the larger capacity, the larger the TBW, but your cost of investment in GB is actually reduced. OK, I believe friends all want to know how much durable this Yingruida 4TB MX500 --1000TBW, yes, you didn't see the wrong, that is, 1PB; 2TB parameter is 700TBW; and 1TB parameters The last time I also saw it, 360TBW, is there a 4TB SSD special in your hand?

TBW, change one saying, after all, this is the value set by the manufacturer. The SSD life is very close to the hard disk's flash particle relationship. Open the disc, we can see that the PCB board is inversely arranged two 1TB flashing particles, numbered Micron "IKE2D NY135", memory good friends should still remember that the 1TB version has two numbers "0SB2D NW952 "500GB flash particles, according to the 5th bits of FPGA numbers, can be queried to 96-layer stack 3D NAND TLC particles of Magic light.

However, the speed of visiting foreign websites is really slow, plus the Query system for Mei light is not very familiar, you need to query the accessory number first to find the specific parameters. This efficiency is much higher, soon, the FPGA number of NY135 is quickly checked, and the 176-layer TLC particles are obtained. Previous P5 Plus particle FPGA numbered NY124, we did not know the level difference of the two particles, but it was confirmed that NY's FPGA number specifically referred to as TLC of the 176 layer.

On the flash, we finally succeeded once, NY124 We have known the MTTF (the average failure time) in P5 PLUS is 2 million hours, and the MTBF of NY135 is 1.8 million. In the hour, these two indicators actually have a difference, and it is not necessarily that two kinds of particles will definitely have a difference, and we will further confirm this indicator to the manufacturer.

From the 3D NAND technology route, from 96 layers across 100+ or ​​more, the storage vendors pay a considerable investment and time leapped this limit. In conquering more flash layers have their own technology, how to make the word line (movable layer with the storage unit), how to make the control circuit do not interfere with the stack structure ...

When the magnesium is cooperated in Intel, the floating gate unit architecture is used. After independently developing the product, the updated charge capture unit architecture is immediately used, and after the technical accumulation of the 128th floor, the 176-layer flash unit is quickly launched. The technique of magnesium light can be seen. Not only that, after the last month we tested the first PCIe 4.0 SSD - P5 Plus in Ying Rua last month, the 176-story TLC quickly destroyed the strike, directly applied on a SATA interface, English Rada is still extremely thick in product design. In terms of main control, 4TB and 1TB versions use the same Huirong SM2259H (SM2258 on the previous generation), which is slightly different, which is AD P1H339.00. Huirong This master is more high in the SATA field, although only four read-write channels, each channel supports 8CE chip selection signal. It is to know that many low-end NVME's main control uses 8 channels, 4CE configuration, and the data throughput of the two is actually the same, so the main control should not be the same. The main control is the same, so the 2TB nominal connection is identical to 1TB, still 560Mb / s and 510Mb / s.

In addition, we have previously evaluated that the MX500 supports the NANDXTEND error correction code (ECC) technology, which can extend the life of the particle, improve the decoding efficiency, and provide enterprise-level functions such as end-to-end data path protection and SRAM ECC capabilities. The built-in advanced Direct-to-TLC and SLC Caching algorithm expands the application scenario from the data reliability angle.

Providing stand-alone cache is where MX500 has been very popular. The 4TB version of memory particles are encoded to "1LP77 D9SHD", which is available on the magnesium light official website. This is a 4GB DDR3-1866 DRAM, bit wide to X16. This is also the practice of MX500, providing 1GB independent cache per TB.

The internal space of the MX500 is actually more familiar. The PCB board has not accounted for half of the 2.5-inch disk. On the main control, the main control and the two flash particles have silicone, but the two particles share a piece of silicon. fat. We all know that relatively maintained, the particles are smaller, the reason why you want to make a piece of silicone lipid to these two particles, which is probably the improvement measures that are close to the main control.

After a simple dismantling, let's take a look at the performance of Ying Ruida MX500 4TB: I have said that the 4TB and 1TB sequential read and write in the same main control are the same, and the sequential read speed reaches 560Mb / s, order The write speed is 510Mb / s; the same, random read and write is also the same, random read speed up to 95K IOPs, and the random write speed is 90K IOPS.

As we have previously tested, the new hard drive first run CrystalDiskinfo first, check if its firmware version is the latest, then view its power-on number and time, check the relevant parameters and performance of its support, such as smart, apm, ncq Wait, it is actually a basic function, and it is not here. Of course, you can also view the maximum temperature in the run.

testing platform:

CPU: i5-11600

Memory: 16GB Kingston Haracter, Fury DDR 4-3200 × 2

Graphics: NVIDIA GeForce RTX 2060

Hard disk: WD_BLACK SN850

Motherboard: Gigabyte Z490 AORUS MASTER

System: Windows 10 Family Chinese Edition 20H2

In Crystaldiskmark, we can see that the order of reading and writing of the 4TB version has been almost the same as the official nominal, and the results of 1TB are not big.

In AS SSD Benchmark, the results of the two are similar to Crystaldiskmark, which is 4TB slightly more than 1TB, and the consistency is still quite good.

However, when the 4TB version runs 10GB files in the AS SSD Benchmark test, the case where the write speed in the CRYSTALDISKMARK 64GIB test has not occurred, which is still relatively stable, and the hard disk is relatively stable. The hard disk is relatively stable.

In Atto Disk Benchmark test, 4TB's maximum read and write speed reached 489MB / s and 535MB / s, slightly better than 1TB of 469MB / s and 523MB / s.

In an ANVIL`s Storage Utilities Test, the measurement order read / write rate is 504.19Mb / s and 448.93MB / s, QD16 with deep 4K random write speed is 54K IOPS, 4K Random Read Speed ​​80K IOPS, overall rating 4127.82. Such a score is quite good in SATA SSD.

Then the HD Tune PRO is readily read and written, which is a test of the caching capability of SSD.We can see that there is no reduction in the whole process of reading and writing, quite stable.The average read and write scores are 236MB / s and 258.7Mb / s, respectively, and the delay is also controlled.In File Benchmark, sequential read and write scores are 482078kb / s and 441729kb / s, respectively, and it is also normal.EF Reviews:

From a price, the 4TB version of the MX500 can not be considered approximately, but the same capacity, the PCIe 3.0 SSD with 4K + price is within the acceporay compared with the 5-digit PCIe 4.0 SSD.At least one of the users who need large capacity and stability storage.In the evaluation, the performance of the MX500 is quite good, especially with 176 layers of flash particles, and there is also a corresponding increase in performance than 1TB, especially up to 1 pb TBW, which is undoubtedly able to meet the needs of severe users.