Crucial M4 128GB SSD Hard Drive

[Trog]

The M4 uses the same base, namely a controller Marvell 88SS9174 but in a more recent revision of the BJP2 C300. In practice we had in the hands of a 256 GB version for review and another BLD2 BKK2 (such as 64 and 128 GB), performance is identical. According to Crucial, the revision of the chip is incidental and is the firmware that makes the difference. The Marvell controller comes with 256 MB of DRAM acting as cache, 128 MB against the C300, and the memory used is now the 25nm IMFT, whereas it was previously IMFT 34nm memory.

SATA3 or correctly formulated SATA 6Gb / s are implemented, of course, the same applies to the quality MLC flash chips from the production of Crucial Micron subsidiary. The price per gigabyte can calculate very easily, in which we divide the price by the amount of memory. The hard disk drive manufacturers expect the storage capacity after the decimal number system, even though the computer is working in the binary system. So we have to share the advertised Memory capacity by a factor of 1.074 to determine the real memory size and then calculate the price per gigabyte accordingly. The MTBF value gives a statistical indication of the reliability of a hard disk.

It does not represent the actual assumed life. MTBF values range from hard drives in the range of tens of thousands of hours. This does not mean that a hard disk for example, runs 100,000 hours at a stretch ensures error-free, which is dependent on many factors, such as ambient temperature, operating time, one-off operations, vibration, etc.

The value is calculated from the accumulated running time of a selected number of test samples under laboratory conditions, divided by the number of errors. When a manufacturer such as 1000 copies of a hard drive can run for a year under laboratory conditions, discerned statements. The accumulated operating time is thus 1000 x 24 x 365 hours (8.76 million hours). Fall in this period comes from eight plates, the maker of 1.095 million hours MTBF proud of.
Since a SSD has no moving parts, but others are trying very similar algorithms. With SSDs, for example, MTBF of 2,000,000 hours or more common, representing about 228 years. It can be calculated the probability that it occurs during the lifetime to failure. Estimates of the MTBF can be determined by life tests, where appropriate, with extreme stresses such as radiation, humidity and vibrations and heat. Such tests are not standardized, that is very theoretical.

Specification

• Read speeds up to 415MB / s
• Second-generation SATA 6Gb/sw / Native Command Queuing (3Gb / s backward compatible)
• Form Factor 2.5 "
• 128GB Capacity
• MLC Memory Components
• Interface SATA III
• Sustained Sequential Read Up to 415 MB / s (SATA 6Gb / s)
• Sustained Sequential Write Up to 175 MB / s (SATA 6Gb / s)
• 1,200,000 hours MTBF Features
• 4k Random Read: 40,000 IOPS
• 4k Random Write: 35,000 IOPS
• Integrated 8-channel single-chip Controller
• Self-Monitoring, Analysis, and Reporting Technology (SMART) command in September
• RAID Support
• Average Access Time <0.1 ms
• Data Reliability: Built-in EDC / ECC

[Trog]

Storage

The random access read are almost similar across the range C300, except with a very high level of control at a time that is less favorable to the 64 GB Crucial M4 are generally slower in this exercise that the C300, and if 64 and 128 GB versions have similar results (except in QD16, with which the parameter C300 64 GB is also behind the 128 GB version) 256 GB version is slower. First explanation, IMFT 25nm memories are slower than 34nm IMFT for reading a page, as one move from 65μs to 75μs. This explains the poor performance on the M4 and 64 compared to 128 GB and 128 GB C300 64.

The performance of the 256 GB version comes from the difference in architecture of the memory chips used. Of the 64 and 128 GB versions, the chips are organized with pages - the smallest readable - 4 KB, while the 256 GB version, these are 8-KB pages, therefore, to read 4 KB required during the test, the 256 GB version should actually read 8 KB. The passage in SATA 3G does have a negative impact on performance, but it is lower than in sequential read, the absolute rates are lower.

In writing, the performance is better than reading. When writing randomly on a SSD, it actually writes the data sequentially, while reorganizing its internal allocation table to match the addresses known to the OS on the storage system (the LBA) with the good memory cells. Crucial here are the M4 much faster than the C300, and the capacity is more important is the performance as long as you use multiple commands simultaneously. As sequential write, the flows are limited by the SATA interface so that 3G is far from its theoretical maximum.

Reliability

SSDs do not write the data on ferromagnetic disks, but similar to flash chips to USB sticks. Since a SSD has no moving parts, the benefits are quickly brought together in this regard there are no compulsory mechanical damage possible. A virtually noiseless operation, shock resistance, minimal heat generation. Because driving is not only a read / write head to a specific position must be that the data is immediately available, resulting in extremely fast access times. But not only in terms of linear transfer rate, have SSDs had a fundamental advantage over conventional hard drives. Mainly scattered data requests are their great strength. Joining them will be a significantly lower power requirement, which protects the environment and the pocketbook.

Performance

The flow rates in reading increased from 16.9% in the same capacity from 26.7 to 20.9% against the best in writing. If the random writes are also increasing (from 11.1 to 33%), this is not the case of random reads as one moves from 60,000 IOPS to 40,000 IOPS for the C300 on the M4. The Crucial 64 GB M4 tested combines a controller-BKK2 88SS9174 Marvell, Micron DRAM for the cache chips and Flash 8 Micron 29F64G08CFACB. These chips are engraved in 25nm die and combine two 32 Gb. Page size is 4K, and that a block of 1 MB finding a revision BKK2 here is a surprise since the 128 and 256 GB versions use a revision BL02 that seems more recent. However, according Crucial is the firmware that makes the difference between C300 and M4, not the revision of the controller is incidental.

Crucial to the M4 128 GB Micron DRAM chip is placed on the back and fleas Flash, if they are of the same references are now 16 in number. The 256 GB Crucial M4 differs from 64 and 128 GB versions by the Flash chips, which are 29F128G08CFAAB. Capacity is doubled, due to the use of two die of 64 Gb. This time the pages go to 8 KB and 2 MB block Crucial on M4, approximately 6.9% of the space is used for the Flash wear leveling and internal optimizations, which is the space resulting from the difference between counting the size of the chip Flash (advertised on a base 1 KB = 1024 bytes) and capacity of SSD (advertised on a base 1 KB = 1000 bytes).

[Trog]

Some Negative Aspects

Apart from their current SSDs competitors hang loosely from the above three criteria. It starts with the transfer rates, where it quickly provides current SSDs now 300 MB / s and more. These values are determined by consumer HDDs not even begin to reach and server hard drives need to stretch badly. Although one must also distinguish, because hard drives reach their maximum performance on the outermost regions of their magnetic disks, and this performance differs significantly to the results on the inner regions. In this performance SSDs zones are unknown.

Another issue is the writing of data, since SSDs have here a serious disadvantage, which in turn does not have to deal HDDs: HDD describe sectors, regardless of whether data was previously stored in them or not. SSDs, however, first have to initiate a delete operation if they want the data in a flash chip and overwrite it costs time. This is also the reason why the write performance of an SSD does not correspond completely with the reading performance.

Add to that the limited number of write accesses is possible on the flash memory. While the individual storage items can be on the HDD disks in hard drives erased and rewritten many times, the number of these cycles is limited to the flash chips and is subject to large fluctuations. Thus, the lifetime due to the limited number of erase and Rewrite procedures is limited.

Conclusion

Read all the C300 is worth. Slower with small files, M4 arrive at the same level with the average file then pass by with large files and extras, with speeds up to 355 Mb / s 64-GB and 128 GB version is basically the 256 slower the M4, since it shows slightly lower performance including playback of small, medium and extras. The transition to 3G impacts the speed of all transfers, but it is obviously more experience on the highest flow rates and therefore with large files.

As expected, write performance will vary greatly depending on the capacity of the SSD. The M4 64 GB beats the C300 64 GB in all areas, but the M4 128 and 256 GB are preceded by C300 128 GB and 256 regarding the writing small files. They are faster in other cases. This time the impact of the transition to SATA 3G is limited, but the strange thing M4 128 is more affected than the C300 256 while base rates offered are higher. The 256 GB M4 suffers most from the passage in SATA 3G as it offers the best performance.