News MSI PCIe 5.0 SSD Debuts with 12 GBps Reads, 10 GBps Writes

[Admin]

Spatium M570 Pro uses a Phison M26 controller and Micron 232-layer NAND.

MSI PCIe 5.0 SSD Debuts with 12 GBps Reads, 10 GBps Writes : Read more

 

[hotaru251]

kind of w/e about any improved speed for ssd's.

theres very little real world benefit atm over pcie 4.0's speeds and these will consume more pwr and output more heat for basically nothing. (on top of ofc higher price tag)


directstorage games are gonna get loaded what... 1.5seconds faster than now? (only take liek 5-10 sec to load as is) not really game changer in itself.

 

[TechieTwo]

So the PCI 5.0 SSDs are going to be available almost a year later than advertised in '22. Bad marketing for sure. Apparently they don't expect consumers to be queued up to buy these.

 

[PlaneInTheSky]

I won't buy an SSD that needs a massive heatsink. Data integrity is very sensitive to heat.

PCI 4.0 is more than fast enough, PCI 5.0 SSD are just running way too hot.

 

[Leptir]

theres very little real world benefit atm over pcie 4.0's speeds and these will consume more pwr and output more heat for basically nothing. (on top of ofc higher price tag)

directstorage games are gonna get loaded what... 1.5seconds faster than now? (only take liek 5-10 sec to load as is) not really game changer in itself.

I agree, I really don't see any real world benefits except for some very specific and unlikely scenarios. The 990 Pro (PCIe 4.0) leaves this drive far behind in random performance and that's what counts most of the time. I'm sure that eventually 5.0 drives will be superior, but for 2023 I plan to upgrade my 980 Pro with the 4 tb 990 Pro as soon as it becomes available, not a 5.0 drive.

 

[DavidLejdar]

Then again, when one has a big browser cache, and reloads some tabs from it...

Such is not an usual scenario of course. More likely the case for many is that there is a swap file... And in any case, whether one can make use of increased bandwith, that can be checked with a hardware monitor (to see whether i.e. Gen4 drive tops out during everyday use, assuming the CPU is not a bottleneck about that).

For me, generally quite appealing is that Gen5 drives can have up to 32 TB capacity. And the increased bandwidth I may not make full use of soon. But the way I see it, if I end up spending quite some on a new CPU soon (3D V-Cache), and later on possibly also a new GPU, then saving a few bucks on the drive, which may end up being a bottleneck, not worth that saving. Which drive will perform how, and at what price, that isn't clear yet of course.

And this is talking about moving towards 4K gaming though, where the performance isn't just about a loading screen, with games a bit more demanding than e.g. CS and PUBG. By which I just mean to point out that many a gamer sure may not see much a difference between using a Gen3 or Gen5 drive, but there are use cases where it is nice that Gen5 is coming, in particular as it enables CPUs and GPUs to grow further.

 

[USAFRet]

theres very little real world benefit atm over pcie 4.0's speeds

"theres very little real world benefit atm over pcie 3.0's speeds"

FTFY

 

[dstarr3]

kind of w/e about any improved speed for ssd's.

theres very little real world benefit atm over pcie 4.0's speeds and these will consume more pwr and output more heat for basically nothing. (on top of ofc higher price tag)


directstorage games are gonna get loaded what... 1.5seconds faster than now? (only take liek 5-10 sec to load as is) not really game changer in itself.

In an ideal world, non-volatile storage would be as fast as RAM and thus RAM would no longer be necessary. Any step towards that is worthwhile.

 

[watzupken]

"theres very little real world benefit atm over pcie 3.0's speeds"

FTFY

Totally agree with this. The increase in the sequential transfer rate does nothing for most users, other than running some benchmark to see a higher number. This is akin to the infamous megapixel race with cameras in the past. Most users won't see any difference in picture quality, other than those who few that prints the shots for big posters. And to be honest, even if one is to compare the responsiveness of their system between 1 that uses a SATA 3 SSD and other with a PCI-E 4.0 NVME SSD, I don't think anyone can discern the difference. Difference in game loading time is also negligible.

 

[Jack1923]

In an ideal world, non-volatile storage would be as fast as RAM and thus RAM would no longer be necessary. Any step towards that is worthwhile.

Yes, but...in an ideal world, non-volatile storage would also have the infinite write endurance of RAM. Which you would definitely need if you were going to use it as a RAM replacement.

 

[Co BIY]

Lots of very good thoughts here.

I just won't buy anything named Spatium.

 

[halfcharlie]

I don't know why people who just casually browse the web and play games even care about PC hardware. They're certainly the only people who seem to comment most of the time, if it was up to them we'd still be on Windows 98 with dial-up internet and floppy discs on 15" CRT monitors because 'it works so that's good enough why improve'. The benefits are obvious to anyone who actually does work on a PC, particularly creative/production work. Every second counts when frequently saving, or caching, or exporting/writing etc etc, just one example is saving a project to my regular HDD storage drive taking 20 seconds, the same size project being saved to my working drive an old gen3 m.2 taking 5 seconds, I have to do that every few minutes as a quicksave for safety so you better believe it adds up. Again that is just one small example, and I'm frequently running out of space on my 2x1TB m.2's and having to move files to storage, so higher capacity drives are also great.

 

[nitrium]

There is still an extremely slow crawl in improving QD1 speeds (which is really all that matters for 99% of consumer's real-world usage). QD's almost never go beyond 4 in Windows, except in synthetic tests (which is why those are generally almost useless for most people). The only tests that really matter to the vast majority of end-users imo are the Windows boot time and various video game/application load times, since those represent the typical scenario of an SSD in a typical consumer environment.

 

[ezst036]

I won't buy an SSD that needs a massive heatsink. Data integrity is very sensitive to heat.

PCI 4.0 is more than fast enough, PCI 5.0 SSD are just running way too hot.

The increasing need for heatsinks just means they need to migrate these SSDs back into PCI-E add in cards like the format of the old RevoDrives. Depending on the size of heatsinks and other things(some even look like they may block video cards), it's just a poor placement for this kind of thing.

 

[Nikolay Mihaylov]

There is still an extremely slow crawl in improving QD1 speeds (which is really all that matters for 99% of consumer's real-world usage). QD's almost never go beyond 4 in Windows, except in synthetic tests (which is why those are generally almost useless for most people).

Which is why I agree with the poster above that there is very little gain from PCIe v4 over v3. Also, it is why I format my data drives with 16KB or 32 KB (for 1TB and 2TB drives respectively; just a rule of thumb, don't take it as gospel) cluster sizes. It should guarantee that even for a heavilly fragmented files, the smallest IO sizes wil be higher than the worst case scenario 4KB. I cannot claim that I feel the difference in performance, but at least I am under the impression that I'm doing something

 

[HundredthIdiot]

There is still an extremely slow crawl in improving QD1 speeds (which is really all that matters for 99% of consumer's real-world usage). QD's almost never go beyond 4 in Windows, except in synthetic tests (which is why those are generally almost useless for most people). The only tests that really matter to the vast majority of end-users imo are the Windows boot time and various video game/application load times, since those represent the typical scenario of an SSD in a typical consumer environment.

Right. The Q1T1 time is slower than the 4.0 drive (980 Pro) that I have.

I really despair for the quality of tech journalism when almost zero "reviewers" point this out. I'm not sure whether it's incompetence or subtle pressure from manufacturers.

The other thing I discovered from my own use and then had to search hard to read about is the massive degradation in performance as the drives fill up. Almost all tests are done with an empty drive. So if you buy a 2TB drive maybe only 1.5TB is usable. What's the degradation curve? Don't know. Who publishes such a curve? Nobody, as far as I can tell.

May as well get the manufacturers to write these articles, or fire the "journalists" and get ChatGPT to spew out reprocessed press releases.

 

[Jaxstarke9977]

I'll take one. My mobo has a PCIe 5.0 x4 M.2 slot itching for one of these mama jammas.

 

[YouFilthyHippo]

This is so stupid. Once we hit PCIe-3.5GBps, we should have stopped focussing on speed. You aren't going to notice a difference. Even the biggest AAA games will load a half second faster, and for what? The extra heat and more power consumption? Focus should shift towards longesvity, lifespan, efficiency, more practical usecases. We know software outgrew harddrives, and that was a problem. We have solved that problem with SSDs. But now these retarded NVME speeds are completely impractical and useless. 99.999999999999% of consumers will not find a practical use case for these speeds. Just leave speed alone and find more practical things to focus on. Let's focus on dropping the cost of production, manufacturing, and thus price, and increase longevity. Once you're past 3.5GBps, really, no one cares

 

[USAFRet]

I don't know why people who just casually browse the web and play games even care about PC hardware. They're certainly the only people who seem to comment most of the time, if it was up to them we'd still be on Windows 98 with dial-up internet and floppy discs on 15" CRT monitors because 'it works so that's good enough why improve'. The benefits are obvious to anyone who actually does work on a PC, particularly creative/production work. Every second counts when frequently saving, or caching, or exporting/writing etc etc, just one example is saving a project to my regular HDD storage drive taking 20 seconds, the same size project being saved to my working drive an old gen3 m.2 taking 5 seconds, I have to do that every few minutes as a quicksave for safety so you better believe it adds up. Again that is just one small example, and I'm frequently running out of space on my 2x1TB m.2's and having to move files to storage, so higher capacity drives are also great.

Why are you still using HDD's? I gave that up years ago. All my house systems are SSD only.
The only spinning drives are the ~85TB in my NAS.

And why might you assume that most of us out here do not do "work" on our systems?

The problem here is we are deep into diminishing returns.

ex: Doubling drive speed
10 sec -> 5 sec - Big change
5 sec -> 2.5 - Still big and noticeable
2 sec -> 1 sec - Yes, faster
1 -> 0.5 - hmmmm
0.5 -> 0.25 - Did something change?
0.2 -> 0.1 - I spent money for WHAT? (We're down here now)


Performance doubled with each iteration, but it is now near impossible to tell.

The big sequential number is what we see advertised, and are expected to treat it like the Second Coming. But what really needs to change is the tiny 4k performance.
There, the difference between a PCIe 4.0 and SATA III SSD isn't nearly as great as the woowoo sequential numbers indicate.

That's where most of our data lives....even us "content creators".

I've had people state that a system with a PCIe 4.0 drive (980 Pro) will boot up in 1/2 the time as one with a 3.0 (970 EVO). And be very adamant about it because "simple math".
It doesn't work like that.


Here's 2 of the drives in my system.
980 Pro and 860 EVO, both 1TB.

Looking at that..."Well of course the 980 is 10 times faster, so everything happens 10 times faster!!"
If you were copying a large sequential block between 2 980 Pro's, then yes mostly.
Otherwise, not so much.

Down in the last two rows...the 4k...that is where most of our data lives. And the differences aren't so great.

 

[jp7189]

Right. The Q1T1 time is slower than the 4.0 drive (980 Pro) that I have.

I really despair for the quality of tech journalism when almost zero "reviewers" point this out. I'm not sure whether it's incompetence or subtle pressure from manufacturers.

The other thing I discovered from my own use and then had to search hard to read about is the massive degradation in performance as the drives fill up. Almost all tests are done with an empty drive. So if you buy a 2TB drive maybe only 1.5TB is usable. What's the degradation curve? Don't know. Who publishes such a curve? Nobody, as far as I can tell.

May as well get the manufacturers to write these articles, or fire the "journalists" and get ChatGPT to spew out reprocessed press releases.

Anandtech used to cover all these points. Thay have been my go to site to in depth SSD reviews, but they do so few nowadays.

 

[Deleted member 14196]

I used to use a spatinum when I chewed tobacco. No wait that’s a spittoon.

 

[DingusDog]

Meh. PCIe 4.0 speeds are plenty fast. These drives will be a horrible value.

 

[dan_L]

Spatium M570 Pro uses a Phison M26 controller and Micron 232-layer NAND.

MSI PCIe 5.0 SSD Debuts with 12 GBps Reads, 10 GBps Writes : Read more

Now what about these large and tall heat sinks on these forthcoming PCIe 5.0 NVMe SSDs? The new x670 motherboards include built-in heat sinks. One would think that these built-in heatsinks ought to be adequate for these new PCIe 5.0 SSDs, wouldn't one? The bottom line is will we need to use the heat sinks from the SSD manufacturers or from the motherboards? Some NVMe slots, frankly, won't be useable if we have to use these huge heatsinks from the manufacturers due to their placement on the motherboards so close to the CPU and video cards.

 

[nitrium]

I really despair for the quality of tech journalism when almost zero "reviewers" point this out. I'm not sure whether it's incompetence or subtle pressure from manufacturers.

The "75% full" benchmarks would be extremely useful, and easy to test for. You are right. If you run benchmarks (or even real-world usage), they change depending on how full the drive is. Real-world scenarios should be the primary focus.

 

[2Be_or_Not2Be]

.... these specs remind of why Optane was so good at QD=1.

Oh well, no use crying over spilled milk and all that...