[nextpage title=”Introduction”]
We tested the Kingston KC1000 480 GiB, a high-end SSD that uses M.2 form factor, PCI Express 3.0 x4 connection, and NVMe standard. It is found on 240 GiB, 480 GiB, and 960 GiB capacities, and its annouced maximum read speed is 2,700 MiB/s and write speed of 1,600 MiB/s.
While the most popular SSDs use the 2.5 inches form factor (which is the same size of a standard laptop HDD), the M.2 form factor is being more and more common. The main reason is that this standard allows both the SATA-600 and the PCI Express x4 connections, that has a higher maximum bandwidth. One of the models that use this standard is the Kingston HyperX Predator. There are also SSDs that uses the PCI Express 3.0 x4 connection, but come as an expansion card, like the Intel SSD 750 Series.
Another highlight refers to the conection specification: traditional SSDs use the AHCI (Advanced Host Controller Interface) standard, that was designed for SATA mechanical hard disk drives. Modern drives like the Kingston KC1000 use the NVMe (Non-Volatile Memory express) protocol, which was developed for SSDs, allowing lower latencies and higher speeds, specially under parallel tasks.
Unlike most recent SSDs, that use TLC (triple-level cell) memories, the Kingston KC1000 uses MLC (multiple-level cell) flash memory, that stores two bits per cell (TLC memories store three bits per cell). MLC Flash memories allow smaller data density than TLC ones, so they have a higher cost per gigabyte, but MLC chips have higher speed and longer lifespan, because there is less cell wearing on the erasing process (executed before writing new data).
This fact reflects on the TBW (see table below), which stands for Total Bytes Written, meaning the amount of data written on the drive before it begin to experience tearing problems.
In this review, we compared the Kingston KC1000 480 GiB to the Samsung 960 PRO 512 GiB, to the Samsung 960 EVO 500 GiB and to the Kingston HyperX Predator 480 GiB, that have a similar capacities and also use PCI Express x4 connection. However, while the both Samsung models use NVMe standard, the Kingston model uses the AHCI standard. Another difference is that the Samsung models use PCI Express 3.0 x4 interface, while the Predator uses PCI Express 2.0 x16 standard.
In the table below, we compared the tested units. All of them use M.2 2280 form factor.
Manufacturer |
Model |
Model # |
Nominal capacity |
Price |
Kingston |
KC1000 |
SKC1000H/480G |
480 GiB |
USD 265 |
Samsung |
960 PRO |
MZ-V6P512 |
512 GiB |
USD 300 |
Samsung |
960 EVO |
MZ-V6E500 |
500 GiB |
USD 220 |
Kingston |
HyperX Predator |
SHPM2280P2H/480G |
480 GiB |
USD 350 |
In the table below, we compared technical specs of the tested drives.
Model | Controller | Buffer | Memory | TBW |
Kingston KC1000 | Phison PS5007-E7 | 2 x 512 MiB | 8 x 64 GiB Toshiba TH58TFG9DFLBA8C | 550 TiB |
Samsung 960 PRO | Samsung Polaris | – | 4 x 128 GiB Samsung V-NAND MLC | 400 TiB |
Samsung 960 EVO | Samsung Polaris | 512 MiB | 2 x 256 GiB Samsung V-NAND TLC | 200 TiB |
HyperX Predator | Marvell 88SS9293 | 2 x 512 MiB | 8 x 64 GiB Toshiba TH58TEG9DDKBA8H | 882 TiB |
[nextpage title=”The Kingston KC1000 480 GiB”]
Figure 1 shows the box of the Kingston KC1000 480 GiB. There are two versions available: only the M.2 card, or with an adaptor to be installed on a conventional PCI Express x4 (or x16) slot.
Figure 1: package
As you see in Figure 2, the model we tested comes with a PCI Express slot adapter, which must be used if your motherboard doesn’t come with an M.2 slot compatible with PCI Express 3.0 x4 bus. The package also includes an adapter to use with slim cases and the serial code for the Acronis True Image program, which allows to copy all the content of a drive to another oner.
Figure 2: package contents
On Figure 3, we see the Kingston KC1000 480 GiB removed from the adapter.
Figure 3: the Kingston KC1000 480 GiB
On the bottom of the PCB (solder side,) there are four flash memory chips and one DDR3 SDRAM chip that works as a cache.
Figure 4: bottom side
Removing the sticker, we see the component side of the PCB. Here you see the controller chip, more four flash memory chips and one more cache memory chip.
Figure 5: component side of the PCB
The controller chip used by the KC1000 is the Phison PS5007-E7, seen in Figure 6.
Figure 6: Polaris controller chip
The Kingston KC1000 uses two DDR3L-1600 Kingston D2516EC4BXGGB chips as cache memory. Each chip has 512 MiB of capacity.
Figure 7: buffer memory chip
The eight flash memory chips are Toshiba TH58TFG9DFLBA8C.
Figure 8: flash memory chip
[nextpage title=”How We Tested”] During our testing procedures, we used the configuration listed below. The only variable component between each benchmarking session was the SSD being tested.
Hardware configuration
- Processor: Core i9-7900X @ 4.6 GHz
- Motherboard: Gigabyte X299 AORUS Gaming 7
- Memory: 64 GiB DDR4-3000, four HyperX Predator 16 GiB modules
- Boot drive: Kingston HyperX Predator 480 GiB
- Video display: Samsung U28D590D
- Power Supply: Corsair CX750
- Case: Thermaltake Core P3
Software Configuration
- Operating System: Windows 10 Home
Benchmarking Software
Error Margin We adopted a 3% error margin in our tests, meaning performance differences of less than 3% cannot be considered meaningful. Therefore, when the performance difference between two products is less than 3%, we consider them to have similar performance.
[nextpage title=”Compressible Data Test”]
As you will have gathered from the previous page, we measured the performance of each drive using CrystalDiskMark 5.
First, we set CrystalDiskMark to “All 0x00 Fill mode” to evaluate the performance of the SSD when dealing with compressible data.
On the sequential read benchmark with QD 32, the Kingston KC1000 480 GiB was 19% slower than the Samsung 960 PRO, 16% slower than the 960 EVO, and 184% faster than the HyperX Predator.
On the sequential write benchmark with QD 32, the Kingston KC1000 480 GiB was 10% faster than the Samsung 960 PRO, 28% faster than the 960 EVO, and 120% faster than the HyperX Predator.
On the random read test with 4 kiB blocks and QD 32, the Kingston KC1000 480 GiB was similar to the Samsung 960 PRO, 14% faster than the 960 EVO, and 131% faster than the HyperX Predator.
On the random write benchmark with 4 kiB blocks and QD 32, the Kingston KC1000 480 GiB was on a technical tie with the Samsung 960 PRO, being 12% faster than the 960 EVO, and 104% faster than the HyperX Predator.
On the sequential read benchmark, the Kingston KC1000 480 GiB was 7% slower than the Samsung 960 PRO, 9% faster than the 960 EVO, and 111% faster than the HyperX Predator.
And on the sequential write benchmark, the Kingston KC1000 480 GiB performed similarly to the Samsung 960 PRO and to the 960 EVO, being 63% faster than the HyperX Predator.
On the random read benchmark with 4 kiB blocks, the Kingston KC1000 480 GiB was 162% faster than the Samsung 960 PRO, 120% faster than the 960 EVO, and 224% faster than the HyperX Predator.
On the random write benchmark with 4 kiB blocks, the Kingston KC1000 480 GiB was 24% slower than the Samsung 960 PRO, 29% slower than the 960 EVO, and 39% faster than the HyperX Predator.
[nextpage title=”Incompressible Data Test”]
For the second test, we set CrystalDiskMark to the default mode, which uses incompressible data. As both results were similar for all the tested SSDs, we choose to show only the uncompressible data results.
On the sequential read benchmark with QD 32, the Kingston KC1000 480 GiB was 22% slower than the Samsung 960 PRO, 18% slower than the 960 EVO, and 163% faster than the HyperX Predator.
On the sequential write benchmark witn QD 32, the Kingston KC1000 480 GiB was 46% slower than the Samsung 960 PRO, 38% slower than the 960 EVO, and 5% faster than the HyperX Predator.
On the random read test with 4 kiB blocks and QD 32, the Kingston KC1000 480 GiB was similar to the Samsung 960 PRO, 15% faster than the 960 EVO, and 206% faster than the HyperX Predator.
On the random write benchmark with 4 kiB blocks and QD 32, the Kingston KC1000 480 GiB was similar to the Samsung 960 PRO, 5% faster than the 960 EVO, and 100% faster than the HyperX Predator.
On the sequential read benchmark, the Kingston KC1000 480 GiB was 42% slower than the Samsung 960 PRO, 27% slower than the 960 EVO, and 70% faster than the HyperX Predator.
And on the sequential write benchmark, the Kingston KC1000 480 GiB was 60% slower than the Samsung 960 PRO, 58% slower than the 960 EVO, and 31% faster than the HyperX Predator.
On the random read benchmark with 4 kiB blocks, the Kingston KC1000 480 GiB was 20% slower than the Samsung 960 PRO, 34% slower than the 960 EVO, and similar to the HyperX Predator.
On the random write benchmark with 4 kiB blocks, the Kingston KC1000 480 GiB was 49% slower than the Samsung 960 PRO, 47% slower than the 960 EVO, and similar to the HyperX Predator.
[nextpage title=”32 GiB write test”]
One of the main disadvantages of TLC Flash memories is its low write speed. Most SSDs that use this kind of memory compensates this by including in its controller chip a small amount of fast SLC Flash memory, acting as a write cache. So, in these models, write operations with small amount of data are fast, because the data are stored in the SLC cache and later, when the drive is idle, the controller transfer them to the TLC memory chips. But when you write a big amount of data (more than the SLC cache), the speed drops significantly.
In order to benchmark it, we used CrystalDiskMark 5, on sequential write modes, with two repetitions and 32 GiB test file. Let’s check the results.
On the sequential write and queue depth of 32, the Kingston KC1000 was 25% slower with 32 GiB than with 1 GiB.
On the simple sequential write benchmark, the KC1000 480 GiB was 13% faster with 32 GiB than with an 1 GiB file.
[nextpage title=”Conclusions”]
As usual, our tests result on several conclusions. The first one is that the Kinsgton KC1000 480 GiB achieved a higher performance with compressible data than with unconpressible data, which means its controller uses data compression to improve the performance.
Compared to the HyperX Predator, which is the older model from the same manufacturer, we can say there was a huge performance improvement, of more than 100% on most tests.
Comparing the KC1000 to its competitors from Samsung, the 960 EVO and the 960 PRO, it achieved lower maximum transfer rates, but it was similar on some tests, and faster in other ones.
However, the fact it uses MLC (and not TLC) memories gives it a better endurance. As you see in the table in Page 1, its TBW (Total Bytes Written, which is the amount of data you can write to the SSD before it could experience problems due to wearing) is 550 TiB, which is almost three times the TBW of the Samsung 960 EVO, for example.
Besides that, it doesn’t have the same problem seen on the 960 EVO, where there is a major performance drop when writing a big amount of data.
Because of that, even not being the fastest SSD available, it presents a good balance between performance and endurance, which makes it a great choice for users that need to write a big amount of data dialy.
One last detail: according to the CrystalDiskInfo program, the KC1000 reached more than 80 degrees Celsius when under operation. So, if your motherboard offers a heatsink for the M.2 slot, it is a good idea to use it, or at least to provide a good airflow on the SSD.
Leave a Reply