Post your 3DMark DirectStorage results

[QuietBob]

DirectStorage feature test is free if you already own the Storage Benchmark component of 3DMark. For those that have the standard version of the suite, Storage Benchmark DLC can be purchased separately for measly 3 USD/EUR.

The test focuses on the benefits of Microsoft DirectStorage API, which enables accelerated transfer of game assets from the storage medium to video memory, as well as faster asset decompression. This should result in much quicker level loading times, and a smoother overall experience due to less hitching and more even frame times.

I've already done a detailed analysis of DirectStorage and its contributing factors. It required quite a lot of effort to assess the data and aggregate the results in a spreadsheet, while manipulating specific variables in an app that is meant for developers and not exactly user friendly. But this standardized benchmark needs just one click and a few minutes to run.

Please post a screenshot with your results. I'm also interested in seeing the detailed statistics that the test provides:

Here's my result performed on the system in my profile. I have no idea why the component fields are blank, it did detect them properly during the benchmark run. In any case, it was a 5800X3D, a 7900XTX with stock clocks, and a 1TB WD Black SN850X.

 

[Tomorrow]

 

[Arctucas]

I scored 114.9% faster in DirectStorage feature test

Intel Core i9 processor 14900K, NVIDIA GeForce RTX 4070 Ti SUPER x 1, 49152 MB, 64-bit Windows 11}

www.3dmark.com

 

[rusTORK]

Original result:

I scored 129.7% faster in DirectStorage feature test

AMD Ryzen 7 5800X3D, NVIDIA GeForce RTX 3080 Ti x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

But it also might be like this:

 

[Kapone33]

Is there not another thread for this topic?

 

[QuietBob]

But it also might be like this:

How did you get those numbers in the second test? The graph looks different.

Is there not another thread for this topic?

I thought this was a more appropriate place for sharing the results. You're welcome to copy yours here.

 

[rusTORK]

How did you get those numbers in the second test?

I think i run test twice. I was unable to get result close to this one. I think it's bad data.

 

[BlaezaLite]

 

[QuietBob]

Ah, now I see what happened. The benchmark got an update on December 18, adding the previously missing statistic (RAM->VRAM DS on uncompressed) to the first graph:

DirectStorage Feature Test v1.0.6.1

• Fixed an issue with some systems with non-Latin character sets.
• Minor visualisation improvement in the result page, added RAM to VRAM score without GDeflate.

View attachment 376306

Some incredible numbers here, though the memory transfer rates seem inflated?

Ok, so I re-ran the test. Some transfers are a bit higher, but it's still not showing the missing number 16.61 GB/s in the first graph
Third time's the charm! Had to re-start Steam, otherwise the benchmark wouldn't update:

 

[BlaezaLite]

Ah, now I see what happened. The benchmark got an update on December 18, adding the previously missing statistic (RAM->VRAM DS on uncompressed) to the first graph:

DirectStorage Feature Test v1.0.6.1

• Fixed an issue with some systems with non-Latin character sets.
• Minor visualisation improvement in the result page, added RAM to VRAM score without GDeflate.

Some incredible numbers here, though the memory transfer rates seem inflated?

Ok, so I re-ran the test. Some transfers are a bit higher, but it's still not showing the missing number 16.61 GB/s in the first graph
Third time's the charm! Had to re-start Steam, otherwise the benchmark wouldn't update:

View attachment 376346

I've a pcie gen 5 boot drive, must be why my rates are fast. I dunno!

 

[QuietBob]

I've a pcie gen 5 boot drive, must be why my rates are fast. I dunno!

Sure, that explains the higher SSD->RAM transfers Though the benefit of DS seems to be also much higher with a Gen5 drive.
But compare your RAM->VRAM DS off with @Arctucas on much faster 8000cl38 DDR5. Your result is still higher. Maybe they changed the way it's calculated with the update?

 

[BlaezaLite]

Sure, that explains the higher SSD->RAM transfers Though the benefit of DS seems to be also much higher with a Gen5 drive.
But compare your RAM->VRAM DS off with @Arctucas on much faster 8000cl38 DDR5. Your result is still higher. Maybe they changed the way it's calculated with the update?

I'll run it again...

I'm so lost lol. I turned off edge and a few background programs and got this...

 

[QuietBob]

I'll run it again...

I'm so lost lol. I turned off edge and a few background programs and got this...
View attachment 376358

Nice scores! Guess we'll have to wait for more results with DDR5/Gen5 for comparison.

 

[BlaezaLite]

Nice scores! Guess we'll have to wait for more results with DDR5/Gen5 for comparison.

My 'puter isn't called "AMDWeapon" for nothing, lol

 

[rusTORK]

Here is another result: GPU Boost Lock activated in EVGA Precision X1 (it will keep high clocks during benchmark run).

All results are higher, but there one strange thing... Take a look on second screen.

CPU (or motherboard) decided to silently overclock itself (on 101,86 MHz).

 

[Kapone33]

 

[stahlhart]

NVIDIA GeForce RTX 4090 video card benchmark result - Intel Core i7-13700K Processor,ASUSTeK COMPUTER INC. ROG STRIX Z690-E GAMING WIFI

UL is misreporting the drive capacity; it's actually 4Tb.

@kapone32 : you've got a driver interfering with Direct Storage; you can check for this by running "fsutil bypassIo state c:\" from an elevated command prompt. I had this same issue, and in my case it turned out to be disk imaging software (O&O). Uninstalling the application corrected the problem. Apparently, third party virus scanners can cause this as well.

 

[Paramount]

I just surprised how VP4300L with no DRAM get better results than Soldigim P44 pro

 

[QuietBob]

I just surprised how VP4300L with no DRAM get better results than Soldigim P44 pro

I don't think it does. If you look at the transfer rates, they're all higher with the Solidigm. But there's something wrong with the SSD->RAM DS on result, it should be higher with DS enabled than without.

Perhaps some chance background activity lowered the transfer?

 

[INSTG8R]

DirectStorage feature test is free if you already own the Storage Benchmark component of 3DMark. For those that have the standard version of the suite, Storage Benchmark DLC can be purchased separately for measly 3 USD/EUR.

The test focuses on the benefits of Microsoft DirectStorage API, which enables accelerated transfer of game assets from the storage medium to video memory, as well as faster asset decompression. This should result in much quicker level loading times, and a smoother overall experience due to less hitching and more even frame times.

I've already done a detailed analysis of DirectStorage and its contributing factors. It required quite a lot of effort to assess the data and aggregate the results in a spreadsheet, while manipulating specific variables in an app that is meant for developers and not exactly user friendly. But this standardized benchmark needs just one click and a few minutes to run.

Please post a screenshot with your results. I'm also interested in seeing the detailed statistics that the test provides:

View attachment 375987
View attachment 375988View attachment 375989

Here's my result performed on the system in my profile. I have no idea why the component fields are blank, it did detect them properly during the benchmark run. In any case, it was a 5800X3D, a 7900XTX with stock clocks, and a 1TB WD Black SN850X.

My spec is pretty much identical to yours for the most part(I'm using a 970 Evo Plus, oof which just clicked is a Gen 3)so I would probably score about the same. I'm sure my own curiousity will get the better of me and I'll end up buying to see if that's true
Edit: @QuietBob I just noticed you are on Win 10? I was under the impression DS was only availble on Win 11. I mean if I'm wrong that's great!

 

[BlaezaLite]

Ran it again with a gen 4 Seagate Firecuda

My last drive. Unimpressed with speed.

 

[QuietBob]

I was under the impression DS was only availble on Win 11. I mean if I'm wrong that's great!

Yes, DS works in Win10. The only difference in Win11 is the optional BypassIO mode which is supposed to further reduce CPU overhead when copying data.

But it doesn't seem to provide a noticeable advantage when looking at the limited data we have here.

 

[stahlhart]

NVIDIA GeForce RTX 3070 Ti video card benchmark result - Intel Core i5-12600K Processor,Micro-Star International Co., Ltd. MPG Z690 EDGE WIFI DDR4 (MS-7D31)

Windows 10 and DDR4 for this one, and the SSD actually is 2Tb here. Also...

"8Gb of VRAM"

 

[INSTG8R]

Yes, DS works in Win10. The only difference in Win11 is the optional BypassIO mode which is supposed to further reduce CPU overhead when copying data.

But it doesn't seem to provide a noticeable advantage when looking at the limited data we have here.

Well I grabbed it out of curiosity and to see how I stack up. Thanks for the clarification. I’m running it now

So my Gen3 OS drive vs my Gen4 Game drive

 

[QuietBob]

Ran the test with a Gen3 SSD in the same system and I got weird results. DS is of little benefit here unless GPU decompression is used.

Maybe it's because the drive is attached to the chipset? I'll test the same drive attached to the CPU in a different system.