If you've just bought a shiny new or an in the Black Friday sales, then you might be wondering just how well it can fling data between the storage and the graphics card. Well, it turns out that 3DMark has been wondering the same thing because it's just released a to test DirectStorage performance.

If [[link]] you're wondering what DirectStorage is, just think of it as a software toolkit that when used in games can help speed up the loading of data, between the storage drive and graphics card's VRAM, as well as reducing the load on the CPU.

Normally, such data transfers are handled in a very linear fashion, resulting in a bottleneck with modern SSDs and CPUs. DirectStorage is an API that provides more control over this, by allocating a large pool of system memory for the incoming data and then using multiple data transfers in parallel. The most recent version of DirectStorage also includes an option to let the GPU, rather than the CPU, handle any data decompression.

The new benchmark in 3DMark is a so-called feature test, so it doesn't produce a final score. Instead, you get estimates of how many gigabytes per second a given data transfer has achieved. The test runs in three stages: Without DirectStorage, then with it enabled, and finally with the use of GPU decompression (aka GDeflate).

In the images below, you can see a sample test run produced by UL Benchmarks, the makers of 3DMark, and a run . I took the screen capture before the final result was posted, as I wasn't sure how long that display would hang around once finished (not very long, as it turned out).

Image 1 of 2

Note how much data per second is able to be transferred from the [[link]] system RAM to the graphics card's VRAM when using DirectStorage. Also, note how big the effective bandwidth is when using GDeflate.

At this point, you're probably thinking 'Why doesn't every game use DirectStorage?' and it's a very good question. Some do, of course, such as but even that doesn't use the GDeflate option.

The reason why so many games don't use it is because the time to load data isn't being held back by the normal process; other factors are the bottleneck. Horizon Forbidden West doesn't use GPU compression because doing so would reduce the frame rate of the game, as the GPU is too busy rendering the huge open world.

It still benefits from using DirectStorage, of course, which is why it loads up saves so darn quickly. However, all of the data decompression is done via the CPU.

Now, if you've run the new 3DMark test a few times and noticed that the results aren't as high as you expected, bear in mind that read/write transfer speeds for SSDs are always quoted as theoretical maximums. They're achievable in very short bursts but not for a sustained transfer, though how long that window is depends on the model.

Some motherboards will reduce the number of PCIe lanes that a GPU can use if the main M.2 SSD slot is filled, and that can also affect the results of the test. That's the case with mine, so the RAM-to-VRAM result is being affected by the fact that only eight out of the 16 PCIe lanes are active.

Anyway, 3DMark's DirectStorage test is pretty much the maximum throughput your gaming PC or SSD is likely to get. It's not free unless you already have the previously released storage benchmark DLC for 3DMark, but for a handful of pounds, bucks, or dollarydoos, you can check out your system's storage chops