UL Announces 3DMark SSD Storage Benchmark

[Uskompuf]

For more than 20 years, 3DMark has been gamers' first choice for benchmarking the latest graphics cards and processors. Today, we're taking 'The Gamer's Benchmark' into a new area with the 3DMark Storage Benchmark, a dedicated component test for measuring the gaming performance of SSDs, hybrid drives, and other storage devices.

With fast modern SSD storage, loading times are shorter, levels restart faster, and there are fewer interruptions to your gameplay. PC gamers can now choose from a wide range of high-performance storage options from the fastest PCI Express 4.0 and NVMe devices down to cheaper SATA SSDs and high-capacity hybrid drives.





Unfortunately, many of the tools for measuring storage performance were developed when HDDs were the most common drive type. And it's hard to relate results from those synthetic benchmarks to real-world performance. The 3DMark Storage Benchmark is a dedicated component test that measures the gaming performance of the fastest modern PC storage hardware. It supports all the latest storage technologies and focuses on practical, real-world gaming performance.

Real-world gaming performance

The problem with many storage tests is that they use artificial, synthetic workloads to measure performance under ideal conditions. Results from these tests are hard to relate to practical, everyday needs, which is why the 3DMark Storage Benchmark focuses on measuring real-world gaming performance.

Storage activity consists of input and output operations. It is possible to record these operations while the storage device is performing a task. These recordings are called traces.

The 3DMark Storage Benchmark uses traces recorded from popular games and gaming-related activities to measure real-world gaming performance, such as:

• Loading Battlefield V from launch to the main menu.
• Loading Call of Duty : Black Ops 4 from launch to the main menu.
• Loading Overwatch from launch to the main menu.
• Recording a 1080p gameplay video at 60 FPS with OBS (Open Broadcaster Software) while playing Overwatch.
• Installing The Outer Worlds from the Epic Games Launcher.
• Saving game progress in The Outer Worlds.
• Copying the Steam folder for Counter-Strike : Global Offensive from an external SSD to the system drive.

Compare the performance of the latest SSDs

The 3DMark Storage Benchmark is compatible with all modern storage devices and can be used to test both internal and external drives.

The test produces a 3DMark Storage Benchmark Score as a measure of performance. As usual with 3DMark, a higher score means better performance. Here are a few reference scores for context.

• Intel Optane SSD 900P 280 GB (PCI Express 3 M.2) - 4,241
• Samsung SSD 980 PRO 500 GB (PCI Express 4 M.2) - 2,854
• WD_BLACK SN750 NVMe 500 GB (PCI Express 3 M.2) - 2,014
• Samsung SSD 860 EVO 1 TB (SATA III) - 1,193

The test also produces bandwidth and average access time metrics.

Pricing & Availability

The Storage Benchmark DLC is available now for 2.99 USD on Steam and the UL Benchmarks website for 3DMark Advanced Edition while Professional Edition owners with a valid annual license will receive the benchmark as a free update.

View at TechPowerUp Main Site

 

[dj-electric]

This is cool. Last known decent storage benchmark from them was all the way pack on PCMark 8.

Really wish for a variaty of tests on that to measure different storage stengths

 

[watzupken]

In my opinion, this is another meaningless benchmark. From what I can glean from the benchmark results, I feel the few tests that impacts the result is likely the last test where files are being copied over. The rest like loading of games, I think there are plenty of test results that shows very little difference between good SSDs. So the benchmark results make you feel like you have a crap SSD, but in reality it works perfectly fine for people just thinking about game/ application loading time.

 

[piloponth]

I don’t know Rick, benchmark that takes away precious write cycles from my SSD. I will pass.

 

[W1zzard]

Stopped reading at "traces". I'm sure they compress the whole flow of time, so that there's no pauses in-between disk activity

 

[khogan111]

That Intel Optane 900P SSD from 2017 may really be ultra fast, I have never seen one.

However, seeing a 5 year old niche Intel product with a much higher score than a Samsung 980 Pro reminded me why I don't trust Bapco and PCMark series of benchmarks: somehow, Intel branded products score much higher than other brands. Somehow, 0.1 second difference in a specific product feature in Intel branded product leads to hundreds or thousands of higher score in the overal score for the Intel branded product.

On the other hand, whatever additional features products from other brands have, they don't get "special points".

 

[The Quim Reaper]

You don't really need a benchmark to tell you that the difference between the slowest & fastest SSD's, in real world use, is about 2 seconds at most.

If spending $100 extra to save all of 2 seconds, Is the sort of thing that you can't live without and gets you aroused, then have at it.

 

[Ferrum Master]

That Intel Optane 900P SSD from 2017 may really be ultra fast, I have never seen one.

However, seeing a 5 year old niche Intel product with a much higher score than a Samsung 980 Pro reminded me why I don't trust Bapco and PCMark series of benchmarks: somehow, Intel branded products score much higher than other brands. Somehow, 0.1 second difference in a specific product feature in Intel branded product leads to hundreds or thousands of higher score in the overal score for the Intel branded product.

On the other hand, whatever additional features products from other brands have, they don't get "special points".

Optane is miles better than conventual SSD we are being sold off. It is fast, durable and does not rely on SLC emulation trickery and doesn't get slower even when 99% full.

It is faster than the 980PRO as top linear write speeds do not matter at all and can't be used as a metric.

 

[WhoDecidedThat]

Stopped reading at "traces". I'm sure they compress the whole flow of time, so that there's no pauses in-between disk activity

Isn't this a mostly read speed benchmark?

 

[jigar2speed]

I don’t know Rick, benchmark that takes away precious write cycles from my SSD. I will pass.

Ditto my thoughts, plus not to forget if the SSD is running on last legs or has a vulnerable hardware you risk your data and drive.

 

[WonkoTheSaneUK]

That Intel Optane 900P SSD from 2017 may really be ultra fast, I have never seen one.

I have one. I won it as a door prize at that year's annual "Citizen Con" - The yearly (except for the last couple, obviously) conference for video game, Star Citizen.
Intel sponsored that year's event to launch the SSD, and gave the company 200 or so to give away.

 

[Ferrum Master]

I have one. I won it as a door prize at that year's annual "Citizen Con" - The yearly (except for the last couple, obviously) conference for video game, Star Citizen.
Intel sponsored that year's event to launch the SSD, and gave the company 200 or so to give away.

I bought the drive and sold off the Sabre Raven on fleabay for a hundred

My score is even higher than the put result.

 

[GabrielLP14]

Stopped reading at "traces". I'm sure they compress the whole flow of time, so that there's no pauses in-between disk activity

Yeap probably right, which means if the drive has its pSLC Cache full, you'l see a degradated performance

 

[Tsukiyomi91]

I only know that owning an SSD, regardless of whether it uses SATA or the "slower" PCIe Gen3 x4 speeds is plenty enough for me as I just want storage drives faster than say... a SSHD.

 

[AVATARAT]

It's good to know that you need 30GB on C:\ drive for this Benchmark and you cannot change to other drive!

 

[Bruno Vieira]

I bought the drive and sold off the Sabre Raven on fleabay for a hundred

My score is even higher than the put result.

You have a faster CPU. CPU and mem also matters a lot in game loding benchmarcks. Its a system benchmarck focused on storage.

 

[Ferrum Master]

You have a faster CPU. CPU and mem also matters a lot in game loding benchmarcks. Its a system benchmarck focused on storage.

My memory set is ancient PoS if we compare on average... CPU maybe, but there are a lot better single threaders out there. I tested my other drives, they are in line with the numbers.

The test is actually a heavy read test, the pSLC cache exhaustion shouldn't matter that much, unless it is really small. Here actually the inherited flaws of current NAND controller architectures creep out.

Optane isn't about the speed, like many argue here, it is about stability and how you can hammer it and it remains the same. It is a perfect OS drive.

 

[Tomorrow]

Optane is miles better than conventual SSD we are being sold off. It is fast, durable and does not rely on SLC emulation trickery and doesn't get slower even when 99% full.

It is faster than the 980PRO as top linear write speeds do not matter at all and can't be used as a metric.

Also unlike NAND based SSD's Optane does not really benfit from parralelism. A 100GB Optane is as fast as a 1TB one assuming the board and controller are same. NAND based ones get faster the higher the capacity (and number of chips). Intel P5800X is literally the best SSD out there. Crushing 4K r/w and PCIe 4.0 compliant using 2nd gen Optane. Unfortunately availability in EU is nonexistant and price is absurd.

 

[csendesmark]

I see no temperature logging,
Why? This way 3DMark SSD SB does not give us anything new...

 

[Braegnok]

The benchmark takes awhile to setup, run. On first benchmark result it shows typical graphs as all other 3DMark Benchmarks show. http://www.3dmark.com/strg/5512

 

[freeagent]

Pretty sweet!

Would be sweeter if they didn’t make you pay for it.. because you already bought the program, and chances are you paid for the other “extras”..

 

[looniam]

price 10/10
graphics 1/10

 

[Ferrum Master]

Also unlike NAND based SSD's Optane does not really benfit from parralelism. A 100GB Optane is as fast as a 1TB one assuming the board and controller are same. NAND based ones get faster the higher the capacity (and number of chips). Intel P5800X is literally the best SSD out there. Crushing 4K r/w and PCIe 4.0 compliant using 2nd gen Optane. Unfortunately availability in EU is nonexistant and price is absurd.

Yeah... it costs and arm and leg... but if you look at GPU prices... well maybe not. At least this one really is designed not to screw you up and you really get something what you paid for, the best of best in all realms, like some are scared here that the benchmark only does deteriorate their drive on their aging QLC's or TLC's It cracks me up.

I am also considering NVME 22110 sized older Optanes, if some emerges up around me. I don't need the pcie4 bursts, as they give you nothing.

We are forgetting the Samsung Z-NAND also, Sammy is really late this time to reply to the P5800X. It would be fun to see someone sporting Samsung Z-SSD SZ985.

 

[tehehe]

If you want real ssd performance tests look at this: https://www.snia.org/tech_activities/standards/curr_standards/pts
There are several open source tools that try to do this to a varying degree of success, but are still miles better than this lol worthy "test".
https://www.snia.org/sites/default/files/SSS_PTS_Client_v1.2.pdf read at least up to page number 20 to get an idea why this "test" is shit.

 

[Vecix6]

It's a really boring test nothing to see. It took about 25 minutes to show only 2 result numbers... If the test measures suposed time to load Battlefield V, why not show this in time?