NVMe SSD Storage Efficiency: Intel Z370 vs. AMD X470

Today, we're investigating high-speed storage performance by comparing AMD and Intel mainstream desktop chipsets with the latest and greatest ADATA XPATA SX8200 960 GB SSD. The SX8200 is an NVMe SSD with an insane sequential read and write performance of 3.2 GB / s and a write performance of 1.7 GB / s. Because it is a very fast NVMe SSD solution that sells at an attractive price, it is one of the most popular drives by enthusiasts, making it perfect for comparing the storage performance of AMD's X470 and Intel Z370 chipsets.

The SX8200 series launched a few months ago includes the latest Silicon Motion SM2262 controller that supports 8 NAND channels, an ARM Cortex-R5 quad-core, NVMe 1.3, LDPC ECC, RAID and more.

Four 64-layer Micron 3D TLC NAND memory chips and two Nanya DDR3 DRAM chips are connected to the controller and act as high-speed buffers. Like most NVMe SSDs, the SX8200 family uses the M.2 2280 form factor and uses the PCIe 3.0 x4 interface for maximum performance. The series offers models with 240 GB, 480 GB and 960 GB and, as already mentioned, is very competitive.

Intel was the first to introduce the SM2262 controller with its 760p series. Today, the 1 TB model is available for $ 370. HP has since undercut them with the EX920, which only costs $ 300 for 1 TB. ADATA has failed to beat HP as the 960 GB SX8200 costs $ 350.

For those of you who are interested in the much cheaper 240GB and 256GB models, ADATA is more competitive. The 240GB model only costs $ 90, while HP requires $ 110 and Intel $ 115. ADATA also offers a 5-year warranty and 160 terabytes for the base model, 320 terabytes for the 480 GB version and massive 640 terabytes for the 960 GB model.

Test notes

We use the Asrock Z370 Taichi with three M.2 slots for testing on the Intel Z370 platform. Like most Z370 motherboards, they are connected to the chipset. The Z370 chipset supports a total of up to 30 high-speed input / output paths, 6 of which are intended as USB 3.0 connections. So 24 must be split for PCIe, SATA and USB 3.0. The Taichi has 34 lanes that don't work. Only lanes are shared here. So if you happen to use all M.2 slots, half of the SATA ports will be disabled.

The three PCIe 3.0 x16 slots of the Z370 Taichi are directly connected to the CPU and therefore do not share any bandwidth with any of the M.2 slots. This is not a big problem for the Z370 platform, since the Z370 chipset connects to the CPU via DMI 3.0, which enables a throughput of 3.93 GB / s. The lower H310 chipset, on the other hand, uses the older DMI 2.0 and is therefore limited to a bandwidth of 2.0 GB / s.

This is a problem for high-speed NVMe SSDs as it can significantly slow performance and it is also important that those who own an NVMe SSD be careful where they are attached.

Even on newer AMD X470 motherboards, the X470 chipset only offers PCIe 2.0 lanes, so it has the same bandwidth limitations as the Intel H310 chipset. The Asrock X470 Taichi used for testing has two M.2 slots. The upper slot, which is referred to as M2 1, is directly connected to the CPU and therefore has a PCIe 3.0 x4 connection. The second slot, labeled M2 2, is connected to the chipset and is therefore limited to the PCIe 2.0 x4 bandwidth.

How much does this affect performance when using a single NVMe SSD like the ADATA SX8200? Let's find out if we …

Benchmarks

First of all, we have the sequential read and write performance in the AS SSD benchmark. This is a challenging benchmark for SSDs because no compressible data is used for these tests. Therefore, this is often viewed as a worst-case scenario. Where other benchmarks show a sequential read performance of 3 to 3.2 GB / s, we are limited to 2.7 GB / s, which is still very fast.

In terms of write performance, you expect the SX8200 specifications to be reached at just under 1.7 GB / s. Since the platforms Z370 and X470 gave very similar results when using a PCIe 3.0 x4 connection, we noticed when running the SSD via the chipset on the X470 card that the performance is severely limited. Read throughput is reduced by 50%, while write bandwidth is reduced by almost 20%.

Here we see that the limited bandwidth of the X470 chipset configuration is not as effective because the performance of reading and writing individual 4K files is much slower. Even so, we still see a 13% drop in writing performance.

In this test 64 requests for 4 KB of data are made simultaneously. Although this is more of a server load, general computing usually makes few requests at the same time. However, this is a better test to maximize the random 4K read and write performance of SSDs. Here, when we are connected to the X47 chipset, we see that throughput is limited by almost 30% when considering write performance.

Here is a brief overview of the access time. Note that less is better here. The Ryzen platform has an advantage here when it is directly connected to the CPU, and this has helped to slightly reduce read access time.

Next we have the ISO file test, which is a benchmark for hard disk copying. Here the Ryzen system was a bit faster than the Core i7 system when connected to the CPU. However, connecting the SSD to the chipset reduced performance by almost 25% and limited throughput to 1.4 GB / s.

The program copy test contains a large number of small, uncompressed files, which significantly reduces throughput. This time we only see a 10% reduction in throughput for the Ryzen platform when connecting through the chipset, although this is still a decent reduction in throughput.

Then we have the game copy test, which contains a mix of small and large, compressed and uncompressed files. This time we see a 13% reduction in throughput for the Ryzen system when connecting through the chipset. However, it should be noted that even the more optimal CPU connection for the Ryzen platform was still 8% slower than the Intel Z370 system with the 8700K.

The ATTO hard drive benchmark uses compressible data, so we see a peak in sequential read performance of around 3 GB / s in this test. As soon as we have reached a file size of 128 KB, the performance of Ryzen and Core i7 is almost the same, at least if the SSD is connected directly to the CPU of the Ryzen system.

The Ryzen system was 20% slower when working with files less than 128 KB, which is a pretty significant margin. Of course it was much slower again when it was connected via the chipset, especially for the tests with larger file sizes.

When measuring write performance, the Z370 platform with the 8700K was faster this time during the ATTO Disk Benchmark tests.

closing remarks

With a typical high-performance NVMe SSD today, reading performance on an AM4 motherboard can be reduced by up to 50% if you install it in the wrong M.2 slot. For the most part, throughput is reduced by at least 10%, and a 20% reduction is often observed.

To be honest, this kind of throughput reduction in everyday tasks, including playing and loading games, is unlikely to be noticed. So for most of you, this is not a serious problem. If you still paid good money to ensure lightning-fast read and write performance, you can also make sure you put it in the right slot. The motherboard manufacturers who specify the bandwidth for M.2 slots make this work a little easier. So read the manual to make sure you get all the GB you should have.

This is a little easier for Intel users because the chipset has the full bandwidth to the CPU, so the slowdown is minimal there.

ADATA's XPG SX8200 is an extremely affordable NVMe SSD, especially the 240 and 480 GB models. The next step would be to get your hands on a second drive for RAID testing. That would undoubtedly provide some interesting results on these platforms.

Continue reading

Purchasing links

Leave a Reply

Your email address will not be published. Required fields are marked *