ADATA GAMMIX S10 in the test

SSDs in the compact M.2 design are becoming increasingly popular in gaming PCs. Since current mainboards are usually equipped with the necessary slots, more and more users want to benefit from the higher transfer rates. For some, however, it still has to be cheap, so today we're looking at a cheap entry-level SSD from ADATA in the M.2 design: the XPG GAMMIX S10.

Intro

SSDs in the compact M.2 design are becoming increasingly popular. This is not necessarily because they take up less space and are therefore easier to install in compact notebooks and computers. Much more decisive is the fact that different logical interfaces can be used on M.2 modules. For example the well-known SATA specifications with speeds of up to 600 MB / s or the much more interesting one for PC enthusiasts PCI Express based on NVM Express. This enables speeds of over 3.200 MB / s, which is a significant leap in performance.

Impressions

Today we're looking at a model from Taiwanese storage manufacturer ADATA: the XPG GAMMIX S10. This SSD was placed in the entry-level area of ​​the ADATA portfolio last fall, and its task is to offer high read rates at a low price and thus outperform conventional SATA models in terms of performance. In this price segment, however, the use of cost-efficient TLC-NAND is inevitable, with which high write rates are usually not to be expected. Most manufacturers usually use SLC caches to compensate for this as much as possible. In this review, we'll see how ADATA addressed this issue.

SSD bookmarks:

Recent SSD Reviews:

The test candidate

ADATA's GAMMIX S10 relies on the widespread combination of inexpensive 3D NAND with TLC control and an SLC cache to accelerate write access. The TLC control of the memory cells makes it possible to store three bits per cell. The writing process takes longer and the writing rates drop. An SLC cache counteracts this and makes it possible to write a few gigabytes much faster. Of course, the cache is emptied again in the background by moving the data to the TLC area so that the write acceleration is available again later. Apart from the lower production costs, this mixture already more than adequately covers the needs of most home users and gamers: The entire amount of data on the data carrier can be read quickly, while writing processes are usually only necessary in moderate amounts. An overview of the different storage types can be found here.

ADATA calls its cache "intelligent cache", which is usually an indication that the size of the cache is dynamic. Unfortunately, exact information on its size cannot be found in the data sheets, so we will try to derive its size in the test. The Silicon Motion SM2260 controller is a well-known representative. Announced in 2015 by Silicon Motion and used in entry-level SSDs such as the Intel 2017p since 600, it is inexpensive and well-engineered.

Overview of the technical data

In the end customer market, Samsung's 960 EVO and Intel's 600p are direct competitors to our test candidate. The following table compares the technical specifications of the manufacturers again:

manufacturer's instructions ADATA XPG Gammix S10 Intel 600p 512GB Samsung 960 EVO 512 GB
controller Silicon Motion SM2260 Silicon Motion SM2260 Samsung Polaris, 8 channels
Protocol and interface NVMe 1.2 protocol over PCIe 3.0 x4 NVMe protocol over PCIe 3.0 x4
form factor single-side M.2 2280
Flash NAND Intel / Micron 32-layer 3D NAND Samsung 48-layer 3D V-NAND
NAND control TLC with SLC cache
Endurance 80 TBW (128 GB) 160 TBW (256 GB) 320 TBW (512 GB) 640 TBW (1 TB) 72 TBW (128 GB) 144 TBW (256 GB) 288 TBW (512 GB) 576 TBW (1 TB) n / a 100 TBW (250 GB) 200 TBW (500 GB) 400 TBW (1 TB)
SLC cache without specification 4 GB (128 GB) 8,5 GB (256 GB) 17,5 GB (512 GB) 32 GB (1 TB) n / amax. 13 GB (250 GB) max. 22 GB (500 GB) max. 42 GB (1 TB)
Max. Read 660 MB / s (128 GB) 1370 MB / s (256 GB) 1.750 MB / s (512 GB) 1.750 MB / s (1 TB) 770 MB / s (128 GB) 1570 MB / s (256 GB) 1.775 MB / s (512 GB) 1.800 MB / s (1 TB) n / a 3200 MB / s (250 GB) 3.200 MB / s (500 GB) 3.200 MB / s (1 TB)
Max. Write(with SLC cache) 450 MB / s (128 GB) 820 MB / s (256 GB) 860 MB / s (512 GB) 850 MB / s (1 TB) 450 MB / s (128 GB) 540 MB / s (256 GB) 560 MB / s (512 GB) 560 MB / s (1 TB) n / a 1.500 MB / s (250 GB) 1.800 MB / s (500 GB) 1.900 MB / s (1 TB)
Max. IOPS read 4K @ QD32 35k (128 GB) 70k (256 GB) 130k (512 GB) 130k (1 TB) 35k (128 GB) 71k (256 GB) 128.5k (512 GB) 155k (1 TB) n / a 330k (250 GB) 330k (500 GB) 380k (1 TB)
Max. Write IOPS4K @ QD32 95k (128 GB) 130k (256 GB) 140k (512 GB) 140k (1 TB) 95k (128 GB) 112k (256 GB) 128k (512 GB) 128k (1 TB) n / a 300k (250 GB) 330k (500 GB) 360k (1 TB)
Encryption none 256 bit AES 256 bit AES, TCG Opal
Manufacturer warranty 5 years 5 years 3 years

Impressions

The most striking feature is the glued-on heat spreader. It is relatively flat and designed with red, dynamic shapes on a black background. In principle, a heat spreader can help to dissipate waste heat from the SSD controller more easily, but it also depends on being supported by the case ventilation - especially in very compact cases.

The 32-layer 3D-NAND can with TLC control 384 Gbit per The save and comes from IM Flash Technologies, a joint venture between Intel and Micron, which also works on high-performance memory 3D XPoint files. We had Intel / Microns 3D-NAND here briefly introduced. As described, the well-known Silicon Motion SM2260 is used as the controller.

The cooler is attached to the circuit board by means of two heat-conducting adhesive pads. However, if you look under the cooler from the side, you can see that most of the controller is not covered by the adhesive pads and thus the heat dissipation from the metal controller surface is at least somewhat limited. Due to the low heat development, this should not be seen as a problem.

The adhesive pads and the cooler itself are 2 mm high in total.

Equipment

ADATA does not provide an NVMe driver for this SSD, so it is addressed with the native driver of the operating system. For all other tasks there is the ADATA toolbox. This supports almost all ADATA models and enables the display of the operating parameters and service life, the optimization of operating system settings with regard to the SSD (e.g. TRIM), firmware updates as well as a quick or full diagnosis of the drive. A read test is carried out over the entire memory area. The package is rounded off by a five-year guarantee.

Test environment

Hardware

Test station:

The test candidate:

Comparison models:

Software

Our benchmark course

Our benchmark course aims to answer the following questions:

  • How fast is the SSD reading and writing large files sequentially and reading and writing small files at random?
  • How do fragmented blocks (not to be confused with file fragmentation!) And the resulting read-modify writes affect performance after a heavy write load?
  • How fast is the SSD in a continuous load scenario (steady state)?
  • Can TRIM restore full performance?
  • How effective is garbage collection?
  • How fast is the SSD when certain mixes of large and small blocks occur?

Synthetic benchmarks

The use of synthetic benchmarks cannot be avoided, since only with these the technical limits of the SSDs become visible. They show the maximum achievable.

Benchmark Usage
Iometer (sequential read / write) Maximum read and write rate for large blocks; is only achieved in practice when reading / writing with large files, e.g. when editing video.
Iometer (random read / write) Maximum read and write rate for parallel access to small 4k blocks. These occur most frequently in practice in daily work.
AS SSD We use this widely used benchmark for the sake of completeness.

With these benchmarks we determine the performance in the following states:

Condition Features
fresh All pages in the SSD are empty and have not yet been written to. This is the status upon delivery or after a Secure Erase.
Used All blocks have already been written to at least once. (Only for writing tests)
after heavy load Performance according to a reproduced load scenario through our Iometer server load profiles.
according to TRIM Performance after the blocks have been released by TRIM.

In this way it can be seen whether and to what extent the performance of the SSD is falling and whether TRIM can restore the original performance.

It does not matter whether you copy a few hundred MP3 or video files or simulate this work with Iometer, the effort is the same for the SSD. Differences resulting from the file system of the operating system then affect all SSDs equally, so that the ratios of the performance differences remain the same.

Trace benchmarks

Real life, on the other hand, can be simulated using trace benchmarks such as PCMark or Iometer profiles, which simulate use cases. With these tests, practical accesses are carried out in a reproducible manner.

Benchmark Usage
PCMark7 trace benchmarks PCMark7 simulates various use cases that are primarily aimed at private multimedia.
Iometer workstation profile This profile simulates a heavily used workstation with 8K access. Two thirds of the accesses are read accesses, one third are write accesses. Two thirds of the accesses are random and one third sequential.
Iometer web server profile Mainly data of various block sizes is downloaded from a web server. This profile reproduces such work.
Iometer file server profile This profile simulates the work of a file server from which files of various sizes are downloaded and uploaded. A fifth of the accesses are write accesses.

For practical results, we carry out these tests after the SSD has already been written with load profiles several times and is occupied with active data except for a remaining 10 GB. This gives you the performance values ​​of an SSD that has already been used and is currently mostly full.

Applications

We test less per application itself. There are two main reasons for this: First, the CPU limit falsifies the performance gap between the SSDs. For example, when the SSD has to wait for the CPU to process certain data before the SSD can continue working when the application starts. Due to the CPU limit, the SSDs move closer together than would be the case with faster CPUs later. Second, many applications can only be measured with a stopwatch, which is too imprecise for us, especially since the results are sometimes only tenths of a second apart. However, we carry out our long-serving OpenOffice copy test because it is easy to reproduce. We have only increased the amount of data there by a factor of 12. It is now 3,06 GB of data in over 48.000 files of various sizes that will be duplicated on the test drive.

Continuous load measurements

As described in the section “Load behavior”, SSDs collapse under a continuous random write load if the garbage collection cannot provide free blocks quickly enough. Such a load behavior occurs only rarely in normal home use. For one or the other reader, however, it might be interesting whether an SSD is also suitable for somewhat tougher use. For example, as a data carrier for a virtualizer, where a lot of small accesses can occur in parallel, or as a disk for a database test environment.

For this test, we let go of as many 4k write accesses as possible to the SSD using the Iometer and create a graph that shows the performance over time. We repeat this test after a 30-minute or 12-hour break to see whether the garbage collection was able to provide enough free blocks for high performance during this time. Since Iometer works with a large test file, which is never deleted, but only overwritten, the influence of TRIM in these two repetitions is excluded. The increase in performance through TRIM itself is then measured in a fourth run. This takes place after a quick format, whereby the drive is "trimmed". The test file is then created again.

We would like to point out that this goes well beyond the normal requirements for SSDs for home use. If an SSD does not do so well here, it is therefore not counted negatively. But we want to find out which SSDs stand out from the crowd. In addition, this test makes it easier to see how and whether the garbage collection is working.

MByte / s or IOPS?

Normally, we give the measurement results in megabytes per second. For the profile tests, however, we opted for IOPS (Input / Output Operations per Second = input and output commands per second). An input or output command can mean reading or writing a block. This does not detract from comparability. If a data carrier achieves 128 IO per second in a write test with 1.000 KB blocks, then mathematically this results in 1.000 * 128 KB = 128 MB per second. When an operating system writes MP3 files or videos, it does so in blocks too, and the block sizes ultimately depend on the size of the files and the formatting of the file system. With many small files, this may limit the number of IOPS and with large files the maximum write rate of the SSD. Therefore, it makes sense to use the specification of IOPS wherever there is a high number of read and write operations and / or different block sizes are involved.

In the case of continuous load measurements, the indication in IOPS has the additional advantage that the maximum IOPS information usually advertised by the manufacturers can be compared directly with the real results.

Measurement results

Sequential reading

These two tests determine how quickly large files can be read. While Iometer continuously reads data from the test address range (= size of the SSD minus 10 GB), AS SSD uses test files that are "only" 1 GB in size. We measure sequential read performance while the SSD is in the following states:

Condition Features
fresh All pages in the SSD were blank before the test and had not yet been written to. This is the status upon delivery or after a Secure Erase.
according to load Performance according to a reproduced load scenario through our Iometer server load profiles. This load is higher than with typical home use.
Note: Between the execution of the server load profile and this test, the SSD was given half an hour of idle time for regeneration via garbage collection, as between all other tests.
according to TRIM Performance after the blocks have been released by TRIM.
Iometer - sequential reading
[seq. Read (fresh)]
[seq. Read (after load)]
[seq. Read (after TRIM)]
Samsung 960 Evo 500GB

2273,7

1413,2

2284,9
Toshiba OCZ RD400

1887,9

1287,5

1886,2
ADATA Gammix S10 512 GB

1251,2

931,6

1256,5
Intel 600p 512GB

1231,4

976,7

1251,9
Corsair Neutron XT 480 GB

554,7

547,9

554,5
Corsair Force LX 256GB

554,4

485,5

552,5
WD Blue 500GB

554,3

546,0

554,6
Crucial BX100 250GB

554,0

477,3

552,2
Sandisk Extreme II 240 GB

552,9

530,4

552,4
Samsung 840 Pro 256GB

547,3

546,4

548,9
Samsung 840 Evo 250GB

542,7

542,4

542,8
Samsung 840 120GB

541,9

486,3

534,8
Crucial m550 256 GB

537,1

517,5

536,6
Sandisk Ultra Plus 256 GB

536,7

460,4

536,1
Crucial MX100 256 GB

534,2

490,4

534,3
Crucial m550 1TB

533,3

536,5

533,8
AMD OCZ Radeon R7 240GB

503,6

422,3

503,9
Corsair Neutron GTX 480GB

498,4

479,8

498,9
Sandisk Extreme 240GB

490,4

425,9

492,3
Crucial MX300 1050 GB

483,0

457,9

482,7
OCZ ARC 100 240GB

459,2

389,7

456,3
MByte / s

Since we carry out the sequential read tests under Iometer with a queue length ("Queue Depth") of 1 and a transfer size of 2M, not all drives can achieve their maximum theoretical read speed. The differences in performance with the same queue length can be seen. AS SSD exploits the reading process more optimally.

AS-SSD - sequential reading
[seq. Read (fresh)]
[seq. Read (after load)]
[seq. Read (after TRIM)]
Samsung 960 Evo 500GB

2672,8

2638,2

2639,9
Toshiba OCZ RD400

2131,4

1169,4

1924,5
ADATA Gammix S10 512 GB

1518,7

1530,8

1531,9
Intel 600p 512GB

1508,0

1513,0

1512,0
Corsair Force LX 256GB

527,7

526,7

527,1
Crucial BX100 250GB

527,4

526,0

527,1
Corsair Neutron XT 480 GB

527,3

518,7

526,2
Sandisk Extreme II 240 GB

522,8

521,0

520,0
Samsung 840 Pro 256GB

522,6

522,4

522,2
Crucial m550 256 GB

521,5

520,1

520,4
Sandisk Extreme 240GB

520,5

501,2

493,7
Crucial MX100 256 GB

519,9

519,4

518,8
WD Blue 500GB

518,9

505,3

507,3
Crucial m550 1TB

518,7

515,6

516,2
Samsung 840 Evo 250GB

515,6

513,6

515,4
Corsair Neutron GTX 480GB

515,5

509,2

516,3
Samsung 840 120GB

515,2

513,4

516,1
AMD OCZ Radeon R7 240GB

512,1

510,0

511,8
Sandisk Ultra Plus 256 GB

505,1

503,6

504,6
Crucial MX300 1050 GB

498,0

490,1

498,4
OCZ ARC 100 240GB

449,5

443,1

447,9
MByte / s

Sequential writing

These two tests determine how quickly large files can be written. While Iometer continuously writes data to the test address area (= size of the SSD minus 10 GB), AS SSD uses test files that are "only" 1 GB in size. We measure the sequential write performance while the SSD is in different states:

Condition Features
fresh All pages in the SSD are empty and have not yet been written to. This is the status upon delivery or after a Secure Erase.
Used All blocks have already been written to at least once.
according to load Performance according to a reproduced load scenario through our Iometer server load profiles. This load is higher than with typical home use.
Note: Between the execution of the server load profiles and this test, the SSD was given half an hour of idle time for regeneration via garbage collection, as between all other tests. Since the results sometimes fluctuate very strongly with AS SSD, we specify the corridor between the minimum and maximum value there.
according to TRIM Performance after the blocks have been released by TRIM.
Iometer - sequential writing
[seq. Write (fresh)]
[seq. Write (used)]
[seq. Write (after load)]
[seq. Write (after TRIM)]
Toshiba OCZ RD400

1556,0

1582,6

54,4

1584,8
Samsung 960 Evo 500GB

659,6

658,7

105,7

657,5
Corsair Neutron XT 480 GB

536,4

535,3

39,7

534,2
Samsung 840 Pro 256GB

526,7

528,6

28,0

487,8
Sandisk Extreme II 240 GB

515,2

517,4

126,4

514,9
AMD OCZ Radeon R7 240GB

503,9

502,6

210,1

504,2
Crucial m550 1TB

503,9

501,0

421,6

499,1
Crucial m550 256 GB

498,2

497,8

138,6

499,6
Corsair Neutron GTX 480GB

497,5

495,4

297,3

498,2
Sandisk Ultra Plus 256 GB

484,7

482,5

39,0

483,5
Crucial MX300 1050 GB

436,8

444,1

293,4

440,6
OCZ ARC 100 240GB

427,8

428,0

220,6

429,5
Crucial BX100 250GB

384,0

382,8

140,5

382,9
Crucial MX100 256 GB

342,7

342,4

49,0

342,9
WD Blue 500GB

310,9

298,7

47,2

309,9
Corsair Force LX 256GB

298,9

298,8

125,9

298,9
Samsung 840 Evo 250GB

289,0

289,7

39,3

290,3
Sandisk Extreme 240GB

240,7

252,8

13,7

252,1
ADATA Gammix S10 512 GB

164,2

183,4

280,6

162,7
Intel 600p 512GB

150,4

155,5

239,1

148,1
Samsung 840 120GB

133,4

133,4

27,7

133,1
MByte / s

Since our Iometer test run writes a large amount of data for several minutes, the write rates for this TLC drive are relatively low because the SLC cache is not sufficient for such a large amount of data. It is noticeable that the value (after load) is higher. Intel's 600p behaved in the same way, and both models have the same controller, so that a connection with the way the SLC cache works can be assumed (see next page).

The AS SSD benchmark, on the other hand, writes a smaller amount of data, so it tends to make the higher write rates with SLC cache visible. While users with large amounts of writing (e.g. 4K video editing) should use the Iometer benchmark as a guide, the AS-SSD benchmark is more decisive for most users.

AS-SSD - sequential writing
[seq. Write (fresh)]
[seq. Write (used)]
[seq. Write (after Last_Minimalwert)]
[seq. Write (after Last_Maximalwert)]
[seq. Write (after TRIM)]
Samsung 960 Evo 500GB

1744,6

1763,7

682,1

1712,6

1768,6
Toshiba OCZ RD400

1156,9

912,3

87,2

913,5

856,5
ADATA Gammix S10 512 GB

842,8

874,5

40,9

857,3

845,8
Intel 600p 512GB

544,6

563,6

39,3

557,6

541,2
Corsair Neutron XT 480 GB

509,7

509,8

34,2

459,0

502,9
Samsung 840 Evo 250GB

503,5

502,7

501,0

501,9

503,2
Samsung 840 Pro 256GB

503,0

443,3

39,7

445,9

487,7
AMD OCZ Radeon R7 240GB

501,8

500,2

498,3

499,4

501,8
Crucial MX300 1050 GB

499,8

490,6

357,5

495,8

493,6
Sandisk Extreme II 240 GB

491,1

489,2

289,7

444,0

488,0
WD Blue 500GB

486,3

498,4

94,5

478,6

498,5
Crucial m550 1TB

486,3

485,2

483,1

484,2

485,8
Crucial m550 256 GB

483,6

482,6

481,2

482,5

483,1
Corsair Neutron GTX 480GB

481,1

480,6

398,6

457,7

463,9
Sandisk Ultra Plus 256 GB

458,5

459,4

94,7

273,0

453,5
OCZ ARC 100 240GB

413,7

435,9

434,9

435,4

414,4
Crucial BX100 250GB

366,0

367,9

363,1

367,8

367,4
Crucial MX100 256 GB

332,8

331,7

331,7

335,2

331,5
Corsair Force LX 256GB

286,9

286,3

286,3

287,2

287,1
Sandisk Extreme 240GB

275,4

207,1

115,2

141,0

204,3
Samsung 840 120GB

128,5

128,5

127,3

128,1

128,0
MByte / s

Sequential writing over time

Here we check how the sequential write speed develops over time in order to scrutinize the SLC cache. The controller first writes larger amounts of data to an area that is quickly controlled in SLC mode. If this area is full, the data rate drops accordingly. The size of the SLC cache can be derived from the write rate and the time at which the write rate fell. ADATA advertises the cache as "Intelligent Cache". Some manufacturers combine this with a dynamic adjustment of the cache size, depending on how full the data carrier is. We take the first exemplary measurement when the SSD is only a quarter full:

The GAMMIX S10 can maintain a write rate of just over 15 MB / s for about 800 seconds before the further write processes take place directly in TLC mode. Now we repeat the measurement if only 10 GB are free on the SSD:

The values ​​are practically identical; the size of the cache does not seem to change in this area. Accordingly, one can assume that the SLC cache in our 512 GB model has a size of 12 GB. This will be correspondingly smaller for smaller models. It is noticeable that the controller evidently empties the cache again during the further write process, whereby the write rate increases to the maximum value for a short moment every few seconds.

Random reading

These two tests determine how quickly 4 kilobyte blocks can be read. When comparing the values ​​between Iometer and AS SSD, it should be noted that Iometer works with a queue depth of 4. We measure the read performance in the case of random access while the SSD is in different states:

Condition Features
fresh All pages in the SSD are empty and have not yet been written to. This is the status upon delivery or after a Secure Erase.
according to load Performance according to a reproduced load scenario through our Iometer server load profiles. This load is higher than with typical home use.
Note: Between the execution of the server load profile and this test, the SSD was given half an hour of idle time for regeneration via garbage collection, as between all other tests.
according to TRIM Performance after the blocks have been released by TRIM.
Iometer - random reading
[4K Read (fresh)]
[4K Read (after load)]
[4K Read (according to TRIM)]
Samsung 960 Evo 500GB

143,0

138,5

141,9
Sandisk Extreme II 240 GB

129,9

115,2

129,5
Samsung 840 Pro 256GB

129,6

129,8

129,5
Sandisk Ultra Plus 256 GB

125,2

56,3

125,4
Toshiba OCZ RD400

121,1

121,1

121,1
Crucial m550 256 GB

120,3

120,2

119,6
Samsung 840 Evo 250GB

117,5

118,0

117,8
Crucial MX100 256 GB

117,3

116,8

117,3
Crucial m550 1TB

115,7

116,3

115,9
Corsair Neutron XT 480 GB

114,1

114,1

114,7
Corsair Neutron GTX 480GB

113,2

112,7

113,2
WD Blue 500GB

111,0

101,8

110,8
Samsung 840 120GB

106,7

106,6

106,7
ADATA Gammix S10 512 GB

105,2

105,9

105,7
Crucial BX100 250GB

97,8

98,0

97,9
Corsair Force LX 256GB

95,5

95,7

96,1
Intel 600p 512GB

89,8

89,6

90,0
AMD OCZ Radeon R7 240GB

88,8

88,6

88,0
Crucial MX300 1050 GB

78,1

77,7

78,1
OCZ ARC 100 240GB

76,6

77,0

77,3
Sandisk Extreme 240GB

46,0

55,4

53,1
MByte / s
AS-SSD - random reading
[4K Read (fresh)]
[4K Read (after load)]
[4K Read (according to TRIM)]
Corsair Neutron XT 480 GB

46,1

45,2

45,7
Samsung 840 Evo 250GB

38,1

36,9

37,9
WD Blue 500GB

37,1

36,7

36,7
Samsung 960 Evo 500GB

35,5

34,9

34,0
Sandisk Extreme II 240 GB

34,0

33,7

33,8
Samsung 840 Pro 256GB

33,3

33,0

33,3
Sandisk Ultra Plus 256 GB

32,9

32,8

32,6
Toshiba OCZ RD400

32,5

30,5

32,2
Crucial m550 256 GB

30,5

30,7

30,6
Crucial MX100 256 GB

29,8

29,7

29,7
Crucial m550 1TB

29,6

29,5

29,4
Crucial BX100 250GB

29,1

29,1

29,1
Corsair Force LX 256GB

28,7

28,5

28,5
Corsair Neutron GTX 480GB

28,4

28,1

28,3
Samsung 840 120GB

28,1

28,1

28,2
Crucial MX300 1050 GB

27,5

25,2

27,3
AMD OCZ Radeon R7 240GB

26,8

30,3

26,7
OCZ ARC 100 240GB

26,3

29,6

25,8
ADATA Gammix S10 512 GB

22,6

22,1

22,2
Intel 600p 512GB

22,0

21,9

22,3
Sandisk Extreme 240GB

21,3

23,6

22,2
MByte / s

Random writing

These two tests determine how fast 4 kilobyte blocks can be written. When comparing the values ​​between Iometer and AS SSD, it should be noted that Iometer works with a queue depth of 4. Measurements with a higher queue depth are carried out in the continuous load measurements. We measure the write performance for random accesses while the SSD is in different states:

Condition Features
fresh All pages in the SSD are empty and have not yet been written to. This is the status upon delivery or after a Secure Erase.
Used All blocks have already been written to at least once.
according to load Performance according to a reproduced load scenario through our Iometer server load profiles. This load is higher than with typical home use.
Note: Between the execution of the server load profiles and this test, the SSD was given half an hour of idle time for regeneration via garbage collection, as between all other tests. Since the results fluctuate very strongly with AS SSD, we specify the corridor between the minimum and maximum values ​​there.
according to TRIM Performance after the blocks have been released by TRIM.
[Iometer]
[Web server]
Toshiba OCZ RD400

56475,0
Samsung 960 Evo 500GB

54861,4
Intel 600p 512GB

48183,0
ADATA Gammix S10 512 GB

46259,1
Samsung 840 Pro 256GB

31500,0
Samsung 840 Evo 250GB

30744,1
Samsung 840 120GB

29824,1
AMD OCZ Radeon R7 240GB

28973,9
Crucial m550 1TB

28374,3
OCZ ARC 100 240GB

26441,1
Corsair Neutron XT 480 GB

26439,7
Crucial m550 256 GB

26157,3
WD Blue 500GB

25488,5
Corsair Force LX 256GB

25475,6
Crucial BX100 250GB

24589,5
Crucial MX100 256 GB

24566,7
Sandisk Extreme II 240 GB

24107,4
Corsair Neutron GTX 480GB

24077,3
Crucial MX300 1050 GB

21580,1
Sandisk Extreme 240GB

18938,4
Sandisk Ultra Plus 256 GB

17251,3
IOPS / s
[Iometer]
[File server]
ADATA Gammix S10 512 GB

49590,9
Intel 600p 512GB

47600,4
Samsung 960 Evo 500GB

37232,8
AMD OCZ Radeon R7 240GB

28599,0
Crucial m550 1TB

28219,6
Crucial MX300 1050 GB

26632,6
OCZ ARC 100 240GB

26362,1
Crucial BX100 250GB

23537,5
Corsair Neutron GTX 480GB

22986,5
WD Blue 500GB

21990,4
Sandisk Extreme II 240 GB

20031,7
Crucial MX100 256 GB

17044,0
Sandisk Extreme 240GB

16410,3
Samsung 840 Evo 250GB

15682,3
Samsung 840 Pro 256GB

14102,8
Crucial m550 256 GB

13885,9
Corsair Neutron XT 480 GB

12625,3
Corsair Force LX 256GB

12054,9
Sandisk Ultra Plus 256 GB

11602,3
Toshiba OCZ RD400

11180,0
Samsung 840 120GB

8325,0
IOPS / s
[Iometer]
[Workstation]
ADATA Gammix S10 512 GB

50668,5
Intel 600p 512GB

48088,5
AMD OCZ Radeon R7 240GB

38440,4
OCZ ARC 100 240GB

38000,1
Crucial m550 1TB

35515,2
Samsung 960 Evo 500GB

27848,3
Corsair Neutron GTX 480GB

26852,5
Crucial MX300 1050 GB

26305,3
WD Blue 500GB

22555,5
Sandisk Extreme II 240 GB

21413,8
Sandisk Extreme 240GB

15622,1
Crucial m550 256 GB

13170,2
Corsair Neutron XT 480 GB

12393,1
Sandisk Ultra Plus 256 GB

11320,9
Toshiba OCZ RD400

11256,9
Crucial BX100 250GB

11209,5
Samsung 840 Evo 250GB

10846,4
Corsair Force LX 256GB

10138,8
Samsung 840 120GB

9483,1
Samsung 840 Pro 256GB

7546,2
Crucial MX100 256 GB

7464,0
IOPS / s

Web server, file server, workstation

These profiles simulate simultaneous read and write access as they occur in typical server or workstation applications. We measure the performance as practically as possible when only 10 GB are free on the SSD and all blocks have already been written to at least once by a previous load that was reproducibly identical for all test subjects.

Profile Features
webserver Blocks of various sizes are read from the SSD. This profile also allows good conclusions to be drawn about game partitions, from which usually only the files of the games are loaded into the RAM.
File server This profile simulates the work of a file server from which files of various sizes are downloaded or uploaded. A fifth of the accesses are write accesses.
Workstation This profile simulates a heavily used workstation with 8K access. Two thirds of the accesses are read accesses, one third are write accesses. Two thirds of the accesses are random and one third sequential.

These profiles represent a load of several minutes. Drives that carry out a garbage collection during idle times benefit from a higher level of performance at the beginning of the measurement.

We come to the mixed load tests. It should be pointed out once again that these extreme load situations do not occur during normal use in the home environment. If a drive does not perform well here, it does not mean that it is less suitable for use at home, but only that it cannot be used for purposes other than intended if you want to experiment with server loads yourself or control the resources for test environments again is stingy.

[Iometer]
[File server]
ADATA Gammix S10 512 GB

49590,9
Intel 600p 512GB

47600,4
Samsung 960 Evo 500GB

37232,8
AMD OCZ Radeon R7 240GB

28599,0
Crucial m550 1TB

28219,6
Crucial MX300 1050 GB

26632,6
OCZ ARC 100 240GB

26362,1
Crucial BX100 250GB

23537,5
Corsair Neutron GTX 480GB

22986,5
WD Blue 500GB

21990,4
Sandisk Extreme II 240 GB

20031,7
Crucial MX100 256 GB

17044,0
Sandisk Extreme 240GB

16410,3
Samsung 840 Evo 250GB

15682,3
Samsung 840 Pro 256GB

14102,8
Crucial m550 256 GB

13885,9
Corsair Neutron XT 480 GB

12625,3
Corsair Force LX 256GB

12054,9
Sandisk Ultra Plus 256 GB

11602,3
Toshiba OCZ RD400

11180,0
Samsung 840 120GB

8325,0
IOPS / s
[Iometer]
[Workstation]
ADATA Gammix S10 512 GB

50668,5
Intel 600p 512GB

48088,5
AMD OCZ Radeon R7 240GB

38440,4
OCZ ARC 100 240GB

38000,1
Crucial m550 1TB

35515,2
Samsung 960 Evo 500GB

27848,3
Corsair Neutron GTX 480GB

26852,5
Crucial MX300 1050 GB

26305,3
WD Blue 500GB

22555,5
Sandisk Extreme II 240 GB

21413,8
Sandisk Extreme 240GB

15622,1
Crucial m550 256 GB

13170,2
Corsair Neutron XT 480 GB

12393,1
Sandisk Ultra Plus 256 GB

11320,9
Toshiba OCZ RD400

11256,9
Crucial BX100 250GB

11209,5
Samsung 840 Evo 250GB

10846,4
Corsair Force LX 256GB

10138,8
Samsung 840 120GB

9483,1
Samsung 840 Pro 256GB

7546,2
Crucial MX100 256 GB

7464,0
IOPS / s

HT4U OpenOffice copy test

Our OpenOffice copy test duplicates the OpenOffice installation files on the test drive. Since today's SSDs do this in no time at all, we have increased the amount of data twelve-fold. Ultimately, 3,06 GB in over 48.000 files of various sizes are read on the test drive and immediately written to another location on the test drive.
[Xcopy]
[OpenOffice copy test]
Samsung 840 120GB

50,8
Sandisk Ultra Plus 256 GB

43,2
WD Blue 500GB

39,9
Corsair Neutron XT 480 GB

35,7
Sandisk Extreme II 240 GB

35,3
Corsair Neutron GTX 480GB

34,9
OCZ ARC 100 240GB

34,5
AMD OCZ Radeon R7 240GB

34,3
Samsung 840 Pro 256GB

33,4
Sandisk Extreme 240GB

33,4
Samsung 840 Evo 250GB

32,3
Crucial MX300 1050 GB

32,2
Intel 600p 512GB

31,6
Crucial MX100 256 GB

31,4
Crucial m550 256 GB

30,5
Corsair Force LX 256GB

30,1
Crucial m550 1TB

30,0
ADATA Gammix S10 512 GB

29,9
Crucial BX100 250GB

28,2
Toshiba OCZ RD400

27,8
Samsung 960 Evo 500GB

27,6
Duration in seconds (less is better)

PCMark7 trace benchmarks

PCMark7 simulates various use cases that are aimed primarily at private multimedia. From the memory tests available in PCMark7, we have selected those that show the greatest differences in performance between devices in the most varied of performance classes.
[PCMark, 7]
[Image import]
Samsung 960 Evo 500GB

34,5
Toshiba OCZ RD400

34,1
ADATA Gammix S10 512 GB

33,6
Intel 600p 512GB

32,4
Corsair Neutron GTX 480GB

30,4
Samsung 840 Pro 256GB

30,4
Crucial m550 256 GB

30,3
Crucial m550 1TB

30,3
AMD OCZ Radeon R7 240GB

30,2
Sandisk Extreme 240GB

30,1
OCZ ARC 100 240GB

29,9
WD Blue 500GB

29,8
Crucial MX300 1050 GB

29,4
Samsung 840 Evo 250GB

29,3
Crucial BX100 250GB

28,7
Crucial MX100 256 GB

28,4
Sandisk Extreme II 240 GB

28,2
Corsair Force LX 256GB

27,5
Corsair Neutron XT 480 GB

27,4
Sandisk Ultra Plus 256 GB

26,5
Samsung 840 120GB

21,0
MByte / s
[PCMark, 7]
[Video editing]
Toshiba OCZ RD400

24,5
Samsung 960 Evo 500GB

23,7
Samsung 840 Evo 250GB

23,7
Samsung 840 Pro 256GB

23,7
Intel 600p 512GB

23,6
Sandisk Extreme 240GB

23,6
WD Blue 500GB

23,5
Crucial m550 256 GB

23,4
Crucial m550 1TB

23,4
Sandisk Extreme II 240 GB

23,3
Crucial MX100 256 GB

23,3
ADATA Gammix S10 512 GB

23,3
Samsung 840 120GB

23,2
Corsair Force LX 256GB

23,2
Sandisk Ultra Plus 256 GB

23,2
Crucial BX100 250GB

23,1
Corsair Neutron XT 480 GB

22,8
Crucial MX300 1050 GB

22,7
Corsair Neutron GTX 480GB

22,4
AMD OCZ Radeon R7 240GB

22,3
OCZ ARC 100 240GB

22,3
MByte / s
[PCMark, 7]
[Application start]
Toshiba OCZ RD400

85,2
Intel 600p 512GB

77,1
Samsung 960 Evo 500GB

75,1
ADATA Gammix S10 512 GB

71,8
Crucial MX100 256 GB

69,3
Samsung 840 Pro 256GB

67,5
WD Blue 500GB

66,8
Crucial m550 1TB

63,6
Crucial m550 256 GB

63,2
Corsair Force LX 256GB

62,0
Crucial BX100 250GB

61,6
Samsung 840 120GB

60,9
Sandisk Extreme II 240 GB

60,6
Corsair Neutron XT 480 GB

60,2
Samsung 840 Evo 250GB

59,1
Sandisk Ultra Plus 256 GB

58,3
Sandisk Extreme 240GB

56,8
Corsair Neutron GTX 480GB

55,1
Crucial MX300 1050 GB

54,2
AMD OCZ Radeon R7 240GB

52,4
OCZ ARC 100 240GB

51,8
MByte / s
[PCMark, 7]
[Games]
Toshiba OCZ RD400

18,1
Samsung 960 Evo 500GB

17,8
Intel 600p 512GB

17,6
Samsung 840 Pro 256GB

17,5
ADATA Gammix S10 512 GB

17,4
Samsung 840 Evo 250GB

17,3
WD Blue 500GB

17,3
Sandisk Extreme 240GB

17,2
Corsair Neutron XT 480 GB

17,1
Crucial m550 256 GB

17,1
Sandisk Extreme II 240 GB

17,1
Crucial m550 1TB

17,0
Crucial MX100 256 GB

17,0
Samsung 840 120GB

17,0
Corsair Force LX 256GB

17,0
Sandisk Ultra Plus 256 GB

16,9
Crucial BX100 250GB

16,9
Corsair Neutron GTX 480GB

16,7
Crucial MX300 1050 GB

16,6
AMD OCZ Radeon R7 240GB

16,3
OCZ ARC 100 240GB

16,3
MByte / s

Continuous load curves

This test is based on the "Solid State Storage Performance Test Specification" of the SNIA (Storage Networking Industry Association). It should show the behavior of the SSD under continuous load - and also what minimum performance the user can build on and how stable the performance is in such a case. For this purpose, the SSD is continuously written with 4k random writes with a queue depth of 32. The longer the SSD can maintain its high initial performance and the higher the permanent performance after the break-in, the better. This test scenario is like that Worst case and less important for normal home applications as it tends to target higher loads. This test shows the loss of performance over time with constant load. With lower loads or smaller test areas, the loss of performance will accordingly only occur later!

The strong dynamic after the SLC cache has been used up is the same as with the sequential time courses can also be seen here. As soon as no more free blocks can be exchanged with the spare area, time-consuming blocks fall Read-Modify-Writes on and the performance collapses. By consistently emptying and releasing the SLC cache, the original performance is always available for a brief moment.

The following is a list of the IOPS averages after the disk has settled at a low level. This gives an indication of the minimum performance to be expected when writing many parallel 4K blocks in the absolute worst case under continuous load.

Steady State Performance

Steady state mean

AMD OCZ Radeon R7 240GB

20000,0
OCZ ARC 100 240GB

18300,0
Corsair Neutron GTX 480GB

12300,0
WD Blue 500GB

11700,0
Samsung 960 Evo 500 GB

11200,0
ADATA Gammix S10 512 GB

10400,0
Sandisk Extreme II 240 GB

9900,0
Corsair Neutron XT 480 GB

8660,0
Intel 600p 512 GB

7300,0
Crucial MX300 1050 GB

5858,0
Samsung 840 120GB

5200,0
Samsung 840 Pro 256GB

4900,0
Crucial m550 1TB

4900,0
Crucial m550 256 GB

4200,0
Crucial MX100 256 GB

4200,0
Corsair Force LX 256GB

3900,0
Sandisk Extreme 240GB

3400,0
Samsung 840 Evo 250GB

3400,0
Sandisk Ultra Plus 256 GB

3400,0
IOPS

Price considerations and conclusion

A look at the current pricing shows that the GAMMIX S10 is offered more cheaply than its M.2 rivals from Intel and Samsung:

Model Price comparison of 500/512 GB PCIe SSDs on Geizhals (April 2018)
ADATA GAMMIX S10 512 GB 152€
Intel 600p 512 GB 164€
Samsung 960 EVO 500 GB 187€

Our tests show that this is justified, at least with regard to Samsung's 960 EVO, since the GAMMIX S10 cannot match the high IOPS and sequential read and write rates of the 960 EVO with the fast Polaris controller. In contrast, Intel's 600p is slower than the ADATA model, especially in terms of write rates.

So if you are considering buying a PCI-Express M.2 SSD with an inexpensive 3D TLC NAND, you have various options available: Samsung's 960 EVO is fast in the benchmarks, but only has a three-year guarantee and is more expensive. Intel's 600p is currently also more expensive than ADATA's S10, but not faster, and has encryption on board.

So if you are looking for an M.2 SSD with good read rates but don't want to spend a lot of money, ADATA's GAMMIX S10 is an M.2 model with a very good price-performance ratio and a long warranty, but you have to use encryption dispense.

Their reading and writing performance are more than sufficient for home applications such as operating systems and games. As a small bonus, it also looks pretty chic with its cooler if you have a window in the computer case.

Test scoring ADATA GAMMIX S10 512 GB
Reading performance +
Writing performance o
Endurance +
Guarantee ++
What's in the box o
Price per GB (price comparison April 19, 2018) € 0,30 / GB (512 GB)
Manufacturer product page

Evaluation options: ++ [very good] / + [good] / o [satisfactory] / - [bad] / - [very bad
[ri], April 26, 2018

About David Maul

David Maul is a qualified business IT specialist with a passion for hardware