News The DirectStorage Advantage: Phison IO+ SSD Firmware Preview

[Admin]

We put Phison's new DirectStorage-enabling SSD firmware to the test.

The DirectStorage Advantage: Phison IO+ SSD Firmware Preview : Read more

 

[-Fran-]

Just out of curiosity... How would a SAS or sATA HDD behave in these tests? Even in RAID 0 would be interesting. More than anything, just to know how far these two are from each other.

EDIT: "these tests" as in the QD32+ and 64KB+ blocks.

Regards.

 

[alceryes]

I think DirectStorage will only make a mediumish splash with gaming, and only in the mid to low tier space.
PCIe 4+ is lightning quick. Put together a PCIe 4+ performance system and you're loading up NVMe-optimized games in 7 seconds or less anyway. Yes, DS could potentially take that down to 3 seconds but, meh.

The unexpected gains of DS will be in the mid-performance gaming systems. Not only will games load much quicker on middling-performance storage mediums but, if you were sometimes hitting a CPU bottleneck due to a mid-performance CPU, DS may be exactly what you need to relieve 3-4% of the CPU workload by moving the asset decompression stage from the CPU to the GPU.

...and, it's free so, yeah, good stuff all around.

 

[salgado18]

I think DirectStorage will only make a mediumish splash with gaming, and only in the mid to low tier space.
PCIe 4+ is lightning quick. Put together a PCIe 4+ performance system and you're loading up NVMe-optimized games in 7 seconds or less anyway. Yes, DS could potentially take that down to 3 seconds but, meh.

The unexpected gains of DS will be in the mid-performance gaming systems. Not only will games load much quicker on middling-performance storage mediums but, if you were sometimes hitting a CPU bottleneck due to a mid-performance CPU, DS may be exactly what you need to relieve 3-4% of the CPU workload by moving the asset decompression stage from the CPU to the GPU.

...and, it's free so, yeah, good stuff all around.

The big deal is not full level loading, but constant incremental loading of assets during games. Something like what Rage tried to do before SSDs. As an example, don't load all textures at once, load only the ones you need at the current scene, and when the player moves you load what you need. That would be very hard on the CPU, but with DS it would be a lot more efficient.

 

[elforeign]

With the SK Hynix P41, are there any firmware improvements in the pipeline to access the benefits of Directstorage or will it require a new drive with a new controller fit for purpose? I recently bought one for a new build and am using it as my primary SSD, but I lack insight into this technology.

 

[itsmedatguy]

The big deal is not full level loading, but constant incremental loading of assets during games. Something like what Rage tried to do before SSDs. As an example, don't load all textures at once, load only the ones you need at the current scene, and when the player moves you load what you need. That would be very hard on the CPU, but with DS it would be a lot more efficient.

It's interesting I believe that Unreal 5 is doing something like this using an atlas to lookup assets, which seems to have lowered the overhead for streaming in what's needed, because Unreal 5 seems capable of doing this kind of thing off of a standard 2.5" SSD

 

[salgado18]

It's interesting I believe that Unreal 5 is doing something like this using an atlas to lookup assets, which seems to have lowered the overhead for streaming in what's needed, because Unreal 5 seems capable of doing this kind of thing off of a standard 2.5" SSD

A stupid example to represent the idea, in old GTA's you only got a few of the cars on the streets, which caused the game to never show a car, but once you got it the game showed that car a lot suddenly. In UE5 Matrix demo, every car is unique, because they are loaded on the fly. I think that's the big advancement of this tech.

 

[gggplaya]

I think DirectStorage will only make a mediumish splash with gaming, and only in the mid to low tier space.
PCIe 4+ is lightning quick. Put together a PCIe 4+ performance system and you're loading up NVMe-optimized games in 7 seconds or less anyway. Yes, DS could potentially take that down to 3 seconds but, meh.

The unexpected gains of DS will be in the mid-performance gaming systems. Not only will games load much quicker on middling-performance storage mediums but, if you were sometimes hitting a CPU bottleneck due to a mid-performance CPU, DS may be exactly what you need to relieve 3-4% of the CPU workload by moving the asset decompression stage from the CPU to the GPU.

...and, it's free so, yeah, good stuff all around.

I think you'll see a benefit in higher clutter object density in scenes and larger openworlds games. Also, more unique objects throughout the map as well.

 

[alceryes]

The big deal is not full level loading, but constant incremental loading of assets during games. Something like what Rage tried to do before SSDs. As an example, don't load all textures at once, load only the ones you need at the current scene, and when the player moves you load what you need. That would be very hard on the CPU, but with DS it would be a lot more efficient.

Partial level loading has been a thing for decades(?)
But, quicker asset access will benefit things like pop-in and more detail distant textures, definitely.

 

[gggplaya]

Partial level loading has been a thing for decades(?)
But, quicker asset access will benefit things like pop-in and more detail distant textures, definitely.

Correct, but loading more map sections are typically disquised as a long tunnel, a long road or highway, or a warp portal etc.... Direct Storage and super fast SSD's will eliminate the need for that.

 

[hannibal]

What I would like to see is Phison with and without this firmware update. What goes up, what goes down...

 

[Giroro]

Correct, but loading more map sections are typically disquised as a long tunnel, a long road or highway, or a warp portal etc.... Direct Storage and super fast SSD's will eliminate the need for that.

Fast storage, even SATA SSDs, already eliminated the need for long loading hallways. So why do we need Direct Storage?

I don't totally understand why reading data with Direct Storage would wear out a SSD faster, but the performance results aren't showing meaningful improvement, so I'm not convinced it will be worth it. Not that storage benchmarks would catch reduced CPU overhead.
(If it didn't cause extra wear then, fine, who cares).

How big or badly designed is Forsaken that it needs a constant 4GBps stream of data from storage, instead of just keeping most the assets in RAM? I understand why the Microsoft's XseX|S or the PS5 might need this tech, because those consoles are ridiculously limited in shared system memory. But on PC? I'm not seeing it.
A system with enough RAM shouldn't be constantly paging out to virtual memory on the hard drive - and games probably shouldn't be designed in a way that would require massive storage bandwidth when free memory (either RAM or GPU) is still available.
Why would requiring a an exceptionally high-speed gaming optimized SSD be better a better concept than having people put that money into more RAM with a normal nvme SSD?

One area where Direct Storage might actually be exciting, is if it improves the access speeds of RAM drives (using a pool of RAM as if it were storage). Obviously I would much rather that games directly take advantage of my "unfillable" 32GB of RAM- but it would be nice to see people get some better gains out of what is currently a terribly inefficient workaround.

 

[-Fran-]

Fast storage, even SATA SSDs, already eliminated the need for long loading hallways. So why do we need Direct Storage?

I don't totally understand why reading data with Direct Storage would wear out a SSD faster, but the performance results aren't showing meaningful improvement, so I'm not convinced it will be worth it. Not that storage benchmarks would catch reduced CPU overhead.
(If it didn't cause extra wear then, fine, who cares).

How big or badly designed is Forsaken that it needs a constant 4GBps stream of data from storage, instead of just keeping most the assets in RAM? I understand why the Microsoft's XseX|S or the PS5 might need this tech, because those consoles are ridiculously limited in shared system memory. But on PC? I'm not seeing it.
A system with enough RAM shouldn't be constantly paging out to virtual memory on the hard drive - and games probably shouldn't be designed in a way that would require massive storage bandwidth when free memory (either RAM or GPU) is still available.
Why would requiring a an exceptionally high-speed gaming optimized SSD be better a better concept than having people put that money into more RAM with a normal nvme SSD?

One area where Direct Storage might actually be exciting, is if it improves the access speeds of RAM drives (using a pool of RAM as if it were storage). Obviously I would much rather that games directly take advantage of my "unfillable" 32GB of RAM- but it would be nice to see people get some better gains out of what is currently a terribly inefficient workaround.

It matters more on mobile and low power devices like notebooks and thermally constrained devices, I'd say. Hence why Consoles got it implemented first in full.

For Desktop, I'd say the biggest advantage is just better texture streaming for seamless loading (less stutter) when the CPU has to do the lift and shift for the GPU when it needs a lot of textures not cached or light enough. Think worlds with a bazillion small textures or ultra quality ones (heavy). The RAM and VRAM usage will go up for sure when games start making good use of it (I hope).

Regards.

 

[crunchylayer4]

sounds like SAS style commands for PCIe

 

[LuxZg]

What I would like to see is Phison with and without this firmware update. What goes up, what goes down...

I'd also like test repeated without added heatsink. They measure 8W peak, and I really don't think it absolutely needs that cooling. I also don't believe typical DirectStorage game will keep that 4-5Gbps sustained over 4 hours. I mean come on, do you actually have any game except maybe MS Flight Sim that would benefit of continuous data stream of that size? Stop to talk to NPC, take cover not to get ass kicked, match ends and you are back in lobby, etc. I've never played a game that has 4h of pure action and never easing up for a bit every 15-20min. (Note: fragging 500 monsters in Serious Sam style still shouldn't require that much data unless each of 500 monsters is unique.. or whatever)

 

[AgentBirdnest]

Idiot question here... if I'm understanding this review right - my Samsung 980 Pro won't quite be good enough to run Forspoken on medium, because it can only stream at 3.5GB/s ?
I actually bought it with Forspoken and DirectStorage in mind, but it looks like it won't be enough? Is this game being gated to only the highest-end consumer SSDs? That can't be right?
Help!! lol.

 

[alceryes]

Idiot question here... if I'm understanding this review right - my Samsung 980 Pro won't quite be good enough to run Forspoken on medium, because it can only stream at 3.5GB/s ?
I actually bought it with Forspoken and DirectStorage in mind, but it looks like it won't be enough? Is this game being gated to only the highest-end consumer SSDs? That can't be right?
Help!! lol.

It's too early to tell.
I doubt those requirements will stick for just medium settings at 4k. That would literally lock over half three quarters of the gammers out there, even ones with great rigs, at low settings.

Edit - ...unless Microsoft gave the developers a pile of cash to make the storage requirements unreasonably high just so they could showcase DS, and by extension Windows 11.

 

[bit_user]

Fast storage, even SATA SSDs, already eliminated the need for long loading hallways. So why do we need Direct Storage?

From what the article says, I think the idea is that they want to decompress assets to temporary storage on the SSD, from where they can be streamed to the GPU on-demand (I can't see why else you'd need to sustain 4 GB/s of reads, unless it's decompressed). The effect should be enabling larger, richer, more detailed worlds.

but the performance results aren't showing meaningful improvement,

Click through the iometer results. Those are pretty massive wins.

How big or badly designed is Forsaken that it needs a constant 4GBps stream of data from storage, instead of just keeping most the assets in RAM?

How much RAM do you want to require @ max quality? Modern desktops cap it at 128 GB, but you experience performance degradation above 64 GB. And the game doesn't get to use all of that for asset caching - maybe 48 GB or so would be safe to assume.

games probably shouldn't be designed in a way that would require massive storage bandwidth

Intuitively, I agree. But you'd really have to get into what sorts of features this enables and what impact that has on resource requirements. For that, you ought to be haunting game dev sites.

Why would requiring a an exceptionally high-speed gaming optimized SSD be better a better concept than having people put that money into more RAM with a normal nvme SSD?

If the SSD is used to cache decompressed or runtime-generated data, then it lessens the burden on the CPU and/or GPU. If spending a little more on a SSD gives you the same experience as spending a lot more on a faster CPU and/or GPU, then it could be good value for money.

 

[bit_user]

It's interesting to see how it stomps all the other drives in iometer, yet delivers only middling latency numbers. In computing, there's often a tradeoff between throughput and latency, so that tallies.

But, then we get to some of the low QD sequential benchmarks, and it's a little surprising to see where it lands on some. In particular, the Crystal Disk Mark 1 MB QD1 Read benchmark stands out. Also, I'd have expected to see it do much better on the Crystal Disk Mark Random Read 4 KB QD256. But, maybe the latter is simply hitting the limits of the underlying NAND and there's not much the firmware can do about it.

I also don't really understand why the drive partially recovers after falling off the cliff on sustained sequential write @ 425 seconds or so. Did it literally fill up all of the space @ lower density, and that forces it to have to simultaneously migrate data from lower-density to higher-density packing, while it continues to write? And once that's finished, it's then able to write at natively higher-density? That's the only explanation I can come up with.

 

[jackt]

I see many security / privacy possible troubles here ! 😑

• If the gpu can read our data, then talk to the multiple drivers background processes, then the drivers talk to nvidia... nvidia can read our data ?
• If someone makes a fake video card device (or virtual device?), it could read our data ?
• Does this require a software to work ? or one is suggested maybe ? Then the ssd creator can read our data.

And these are just a few possible problem, there are more probably !

 

[RodroX]

We will see how better or not it is for gaming, when theres actual launched games for PC to test. If theres more than the only one announced for now.

 

[escksu]

I really don't see how using GPU to perform the decompression helps. If we are in the old days where CPUs only have 1-2 cores, I can understand. But today, we have 4-8 or even more cores. So, how does offloading this task to GPU helps? esp. when bottleneck in games is mostly with GPU rather than CPU.

And then, GPU usually have very limited RAM compared to CPU (most cards are just 8GB). So, if part of it has to be used for storing data from SSD, wouldnt there be even less memory available for games?

Btw, the I would say the main benefit of this is due to the queue depth direct storage uses. NAND is known to be slow at low queue depth.

 

[escksu]

I also don't really understand why the drive partially recovers after falling off the cliff on sustained sequential write @ 425 seconds or so. Did it literally fill up all of the space @ lower density, and that forces it to have to simultaneously migrate data from lower-density to higher-density packing, while it continues to write? And once that's finished, it's then able to write at natively higher-density? That's the only explanation I can come up with.

I believe this is might be due to the way the SLC cache works.

 

[gggplaya]

Fast storage, even SATA SSDs, already eliminated the need for long loading hallways. So why do we need Direct Storage?

That's not true, Call of Duty Zombies for example requires portals to move from map to map. With SSD's the portal loading time is low and seems more natural than older magnetic HDD's. But there isn't a seamless move to the next map, it requires a portal. With a more seamless transition, it'll enable very large open world maps with far more ground objects clutter and unique assetts.

 

[drajitsh]

Fast storage, even SATA SSDs, already eliminated the need for long loading hallways. So why do we need Direct Storage?

I don't totally understand why reading data with Direct Storage would wear out a SSD faster, but the performance results aren't showing meaningful improvement, so I'm not convinced it will be worth it. Not that storage benchmarks would catch reduced CPU overhead.
(If it didn't cause extra wear then, fine, who cares).

How big or badly designed is Forsaken that it needs a constant 4GBps stream of data from storage, instead of just keeping most the assets in RAM? I understand why the Microsoft's XseX|S or the PS5 might need this tech, because those consoles are ridiculously limited in shared system memory. But on PC? I'm not seeing it.
A system with enough RAM shouldn't be constantly paging out to virtual memory on the hard drive - and games probably shouldn't be designed in a way that would require massive storage bandwidth when free memory (either RAM or GPU) is still available.
Why would requiring a an exceptionally high-speed gaming optimized SSD be better a better concept than having people put that money into more RAM with a normal nvme SSD?

One area where Direct Storage might actually be exciting, is if it improves the access speeds of RAM drives (using a pool of RAM as if it were storage). Obviously I would much rather that games directly take advantage of my "unfillable" 32GB of RAM- but it would be nice to see people get some better gains out of what is currently a terribly inefficient workaround.

I strongly agree. For my system I went for 2 x16GB Ram because that is where i found the best price/GB ratio.i have DDR3600 C16. I could have gone for 64 GB, but I have not seen my previous system manage to utilise 16 GB @100% in most cases-- except serious multitasking like 3 spreadsheets, a word document, a pdf, browser with 10 tabs, + antivirus etc.
Also, i strongly agree that either the game is completely non-optimised or has been deliberately tuned to exaggerate direct strorage.