Super cool 😎

https://www.3dmark.com/3dm/68469427?

Waay above the average for Samsung SSD 980 PRO 1 TB with 2817 points.

Seems as if things like CPU / RAM and OS have a too big influence. You are not testing a drive but rather a whole system.

Originally posted by crossbone:

https://www.3dmark.com/3dm/68469427?

Waay above the average for Samsung SSD 980 PRO 1 TB with 2817 points.

Seems as if things like CPU / RAM and OS have a too big influence. You are not testing a drive but rather a whole system.

https://www.3dmark.com/strg/4990

There is a difference from model to model, but your performance is greatly affected by the amount of free space on the drive during the test.

Note: I had stock GPU drivers and my RAM was running at 2933 MT/s Gear 2 due to ongoing BIOS issues.

Yes, this is true. Whole system does affect storage performance. This is one of the reasons why we did not have comparison data immediately - we need quite a bit of data for it.

On some Samsung SSDs the firmware version is also a big factor.

Note that the usual "within 3% is normal variance" rule of 3DMark does not apply to the storage test - the variance for it is larger.

Originally posted by UL_Jarnis:

Yes, this is true. Whole system does affect storage performance. This is one of the reasons why we did not have comparison data immediately - we need quite a bit of data for it.

On some Samsung SSDs the firmware version is also a big factor.

Note that the usual "within 3% is normal variance" rule of 3DMark does not apply to the storage test - the variance for it is larger.

I have a Enterprise NVME Samsung Pseries Installed through a U.2 Will this benchmark properly benchmark it? Also have you taken into consideration RAID NVME setups?

Yes, U.2 is still NVME. Just different form factor.

In theory RAID should not matter, but I frankly do not have enough experience with various modern RAID configs to say with confidence how they all behave. The test reads and writes data on normal NTFS filesystem and measures the latency and bandwidth for completing those tasks and rest is up to the OS/driver/hardware stack to perform.

I got 3300 on my 1tb 980 pro, 500 higher than those results, maybe test 980 dont have updated firmware?

Just dl'd the ssd test, but it hangs at 5% "preparing data. I have 3 ssd drive with way more than 30 Gb free. Seems buggy to me

Let it prepare the data. This step can take up to 10 minutes depending on the speed of the drive (on fast SSDs it should be less than 5 minutes)

We will improve the messaging of the progress in a future update, but at this time just let it run and it will eventually step to the actual test. It is processing 30GB of data on that step to generate the sandbox and it takes a while.

Originally posted by Isvelte:

I got 3300 on my 1tb 980 pro, 500 higher than those results, maybe test 980 dont have updated firmware?

It is actually likely, I haven't checked the exact drive what firmware it has. But indeed 980pro has three different firmware variants out there that have quite substantial performance differences. Also other parts of the system (motherboard chipset, storage driver, OS) can affect the result, as can the state of the drive (how full it is). These are all variables that unavoidably affect storage performance.

3DMark Storage Test is just a measuring stick - the thing being measured is quite complex and in general if you see variance or "odd" results, the first reaction should be "what could be causing this unexpected storage performance?" and not "what is wrong with this test?" as the test has been extensively tested and is very close cousin of the PCMark 10 storage test that has been out for quite awhile already. I'm not saying that there couldn't be issues with the test, but it is unlikely at this point and the far more likely explanation is that for some reason the setup is not performing as you expect and the measurement is real.

Originally posted by UL_Jarnis:

Let it prepare the data. This step can take up to 10 minutes depending on the speed of the drive (on fast SSDs it should be less than 5 minutes)

We will improve the messaging of the progress in a future update, but at this time just let it run and it will eventually step to the actual test. It is processing 30GB of data on that step to generate the sandbox and it takes a while.

running it to a Kingston NVMe SSD with 200Gb free waited 23 mins and no increase on 5 % sorry can't afford to wait hrs

That is definitely too long and something is then wrong. Do you have more than one drive to figure out if this is specific to this one drive or not, ie does this occur also on some other drive?

Can you send the result file from the failed run after cancel (found in Documents/3DMark/, most recent result file) to ul.benchmarksupport@ul.com

http://www.3dmark.com/strg/10591
I have a pair of KIOXIA EXCERIA PLUS G2 1TB(R20) Raid0, which only score 2506 in the test.
the result is much above par in other test such as CrystalDiskMark.
I assume there is problem with Raid system?

https://www.3dmark.com/compare/strg/12190/strg/12197#

Both are WD SN850 NVMe drives on a 12700K w/ DDR5 5200 CL38 running XMP.

Left drive is 1TB on CPU PCIe x4, right drive is 2TB on chipset PCIe x4.