News Crucial T700 SSD Preview: Fastest Consumer SSD Hits 12.4 GB/s

[Admin]

The Crucial T700 looks to be the fastest PCIe 5.0 NVMe SSD that will be available on the market, arriving as early as May. Its passive cooling is a nice addition to what is an incredibly fast drive.

Crucial T700 SSD Preview: Fastest Consumer SSD Hits 12.4 GB/s : Read more

 

[edzieba]

This SSD is not intended to be used without a heatsink, such as in a laptop or PlayStation 5.

It would be illuminating to test without a heatsink anyway. The current drive only starts to throttle after 0.2-0.25 TB continuous writes, so based on testing of previous drives without heatsinks, it's quite likely the only effect from operating this drive without a heatsink would be reducing that threshold to 'only' 100 GB or so. I would expect any testing that does not involve throwing around continuous hundreds of GB of data would not see any significant (or any) performance impact from removing the cosmetic greebly.

 

[JarredWaltonGPU]

It would be illuminating to test without a heatsink anyway. The current drive only starts to throttle after 0.2-0.25 TB continuous writes, so based on testing of previous drives without heatsinks, it's quite likely the only effect from operating this drive without a heatsink would be reducing that threshold to 'only' 100 GB or so. I would expect any testing that does not involve throwing around continuous hundreds of GB of data would not see any significant (or any) performance impact from removing the cosmetic greebly.

If you're doing stuff that doesn't throttle, you probably also don't need to buy a Gen5 SSD. I mean, PS5 doesn't support Gen5 anyway, so there's literally zero performance benefit. Anyway, the nature of our SSD benchmarks is that a lot of the tests are of short enough duration that they won't hit the throttling mark. The problem is stuff like CrystalDiskMark that takes quite a bit longer to run and will thus start to throttle hard on later tests. Which results are "valid"? If we take the maximum performance measured in a given test, we can make it look like having a heatsink doesn't matter.

But here's a good indicator of the worst-case scenario for the Crucial T700: with and without a heatsink in our write saturation and ATTO testing. Throttling only kicks off in ATTO at 2MiB block size, meaning everything less than that wasn't hit the thermal limit. But that's only if you start the drive from a "rested" (not hot) state. For IOMeter Write Saturation, starts after 25 seconds of sustained writes. Granted, given the speed, that's still enough time to write about 200–250 GB! 🙃







It's important to note that Crucial explicitly states that a heatsink or some form of cooling is required. Running it without a heatsink effectively voids the warranty and could result in a damaged drive. Don't buy a Gen5 drive and use it without a heatsink, in other words. (Testing the T700 in our Asus board with the motherboard heatspreader results in effectively identical performance to Crucial's own cooler, incidentally. YMMV depending on your particular mobo, heatspreader, and case configuration, naturally.)

 

[atomicWAR]

I feel like we're hitting the point of diminishing returns for SSD speeds in consumer PCs. This extra speed is nice in theory but between the heat and the fact I can barely tell the difference in my PCIe 3.0 drives that max out the bus vs my PCIe 4.0 drives that do the same. Game load times are probably the easiest place beside large file transfers to spot the speed ups and even there the difference feels almost moot when its a 11 sec game load time vs a 9 sec one (or something like that). Plus we went from small passively cooled Pop tart sized 2.5" drives to cool little 'sticks of spearmint gum' that could be slid in anywhere on your motherboard with nothing or maybe a small heat sink, to now requiring large heat sinks and in some instances a fan as well to keep things cool.

Until manufacturers do more with solid state cooling to keep things cool, quiet and compact (or something different than the current heat sinks)...I think most folks will stick to drives their built in motherboard heat sinks can handle making a lot of the PCIe 5.0 nvme drives unattractive. I could be wrong but I know I choose to go all 3.0/4.0 and avoid waiting for 5.0 drives for the reasons of heat and speed I brought up. I did get some beefy passive after market coolers for my two fastest 4.0 drives that max the bus out due to throttling issues I was facing with stock motherboard heat sinks and a couple smaller light wait ones due poor airflow the built in motherboard heat sink had beneath my gpu. I placed my slowest coolest drives here but they again they throttled without better cooling (running one drive was ok but running both drives they over heat the shared heat sink). Until better more compact, quiet cooling is in place I think fast PCIe 5.0 nvme drives will face some adoption issues among home users.

 

[JarredWaltonGPU]

I feel like we're hitting the point of diminishing returns for SSD speeds in consumer PCs. This extra speed is nice in theory but between the heat and the fact I can barely tell the difference in my PCIe 3.0 drives that max out the bus vs my PCIe 4.0 drives that do the same. Game load times are probably the easiest place beside large file transfers to spot the speed ups and even there the difference feels almost moot when its a 11 sec game load time vs a 9 sec one (or something like that). Plus we went from small passively cooled Pop tart sized 2.5" drives to cool little 'sticks of spearmint gum' that could be slid in anywhere on your motherboard with nothing or maybe a small heat sink, to now requiring large heat sinks and in some instances a fan as well to keep things cool.

Until manufacturers do more with solid state cooling to keep things cool, quiet and compact (or something different than the current heat sinks)...I think most folks will stick to drives their built in motherboard heat sinks can handle making a lot of the PCIe 5.0 nvme drives unattractive. I could be wrong but I know I choose to go all 3.0/4.0 and avoid waiting for 5.0 drives for the reasons of heat and speed I brought up. I did get some beefy passive after market coolers for my two fastest 4.0 drives that max the bus out due to throttling issues I was facing with stock motherboard heat sinks and a couple smaller light wait ones due poor airflow the built in motherboard heat sink had beneath my gpu. I placed my slowest coolest drives here but they again they throttled without better cooling (running one drive was ok but running both drives they over heat the shared heat sink). Until better more compact, quiet cooling is in place I think fast PCIe 5.0 nvme drives will face some adoption issues among home users.

Yeah, Gen5 SSDs are very much less about typical consumer use and more for specific use cases like servers. Still, I'll say that verifying a Steam install (or doing a big update to an installed Steam game) can go a lot quicker on a Gen4 or Gen5 SSD. 🙃

There's also the question of how much DirectStorage could make faster SSDs useful in games. I suspect even if done properly, we're still looking at a current maximum of maybe 10-20GB of data for a game engine, level, and textures. So, you could potentially do 20GB in under two seconds with Gen5, versus three seconds with a fast Gen4, versus five seconds with a fast Gen3. But when I launch stuff like Red Dead Redemption 2 and it takes over a minute to get into the game, any of those sounds amazing!

 

[atomicWAR]

Yeah, Gen5 SSDs are very much less about typical consumer use and more for specific use cases like servers. Still, I'll say that verifying a Steam install (or doing a big update to an installed Steam game) can go a lot quicker on a Gen4 or Gen5 SSD. 🙃

There's also the question of how much DirectStorage could make faster SSDs useful in games. I suspect even if done properly, we're still looking at a current maximum of maybe 10-20GB of data for a game engine, level, and textures. So, you could potentially do 20GB in under two seconds with Gen5, versus three seconds with a fast Gen4, versus five seconds with a fast Gen3. But when I launch stuff like Red Dead Redemption 2 and it takes over a minute to get into the game, any of those sounds amazing!

Yeah I couldn't agree more. In the enterprise/server space SSDs speeds will always be hungry for more bandwidth/IO. And with direct storage we may see some uses down the line but for now I think any fast 4.0 SSD is more than enough for most home users.

 

[USAFRet]

I feel like we're hitting the point of diminishing returns for SSD speeds in consumer PCs.

We hit that going from PCIe 3.0 to 4.0.

In some use cases, SATA III SSD to NVMe.

The big jump was HDD to SSD.

 

[DavidLejdar]

Nice stuff. With current-gen software, it may not seem worth it to get it for. But the technical possibility to load 64 GB into RAM in about 5 seconds (and if needed, reload), that's that. And e.g. both RTX 4090 and RX 7900 XTX have 24 GB of VRAM. So it is not like there would be nothing to load the data to.

Whether in particular video game devs will make full use of it, such as to depict e.g. cities to not look like Stalinist Moscow (that is using one type of texture, which gets reused on various walls, and therefore doesn't need to load much when such a texture fills half the screen, with three types of balconies reused to make it look not as monotone) - that isn't clear of course. But the option is there, including moving towards 8K, and not using artificial loading screens, such as an elevator ride between 2 areas as the only option to move between these two areas.

 

[Amdlova]

I have a adata pcix 4x placed on pcix 3 showing on software pcix 2... it's working that matters.
Pci 5 only will be good for server. For desktop for now only heat and compatible issues. Maybe two years beyond to be slim and the software can use that speed.

 

[TechieTwo]

...
Pci 5 only will be good for server. For desktop for now only heat and compatible issues. Maybe two years beyond to be slim and the software can use that speed.

There are no heat nor compatibility issues as long as you are using a PCIe 5 M.2 mobo. Don't be fooled by the elephant sized heatsinks that some companies are supplying with Gen 5 SSDs. In most if not all cases the mobo supplied SSD heatsink/cover plate is more than enough cooling to prevent throttling. The large heatsinks are a marketing ploy.

While the Gen 5 SSDs are almost 7 months late they have merit for some desktop applications. They will not offer any practical advantage over PCIe 3 or 4 unless you are moving a lot of data at once.

 

[PBme]

I feel like we're hitting the point of diminishing returns for SSD speeds in consumer PCs.

It isn't about cooling or anything else being needed to help them max out. It is diminishing returns as this max speed is only for sequential writes for large massive video file to another drive that is this fast. And even that is very time limited. For all the smaller writes these drives aren't showing meaningful difference two pcie 3 drives. Just a very specific and not common use case is what is being increased and advertised. Sets folks up for upgrade disappointment.

 

[tootercomputer]

The Crucial T700 looks to be the fastest PCIe 5.0 NVMe SSD that will be available on the market, arriving as early as May. Its passive cooling is a nice addition to what is an incredibly fast drive.

Crucial T700 SSD Preview: Fastest Consumer SSD Hits 12.4 GB/s : Read more

Long time fan of Tom's Hardware, since 2000, and this is my first post.

I recently completed a long overdue system upgrade and initially purchased a mobo with a pci-e 5.0 NVME slot, anticipating the arrival of 5.0 drives. Unfortunately, the mobo died within an hour and I replaced it with a board capable of "only PCI-e 4.0." I have a Samsung 990 Pro installed for my OS drive. Seeing this review, I was expecting to be bummed out by how quickly my 990 Pro was going to be outdated. But, well, the 5.0 stats in this article for most measures were not all that impressive to me. Perhaps each succeeding generation of PCI-e drives indeed produces only diminishing returns as stated by another poster, unless one is regularly moving around large files frequently, and those seconds in speed improvement start to add up.

On one test (DiskBench 50Gb File Folder), the T700 was 21% faster than the 990 Pro. But I'm not sure how that translates into real-world significance. It reminds me of the distinction between statistically significant differences and clinically significant differences (I was in the social sciences): the former are important and look good, but the latter are what count practically and often represent a genuine meaningful change and improvement.

Don't get me wrong, I'm all for storage progress (going from a 5400 rpm hdd to a 7200 rpm was very exciting, going to the 10,000rpm raptor was mind-boggling, and moving then to solid was was damn near religious!!), and I will keep monitoring the improvements and jump when the time is right. For now, as a mere PC enthusiast, I think I'll be happy for the near future with my PCI-e 4.0 990 Pro on my new Z790 system with an i5-13600k modestly overclocked and 6000Ht Corsair memory. Compared to my last build in 2016, I mean, stuff is just flying around and there are some lights spinning and flashing some very cool colors, and well heck, I'm just having fun.

 

[FunSurfer]

Where is te temperature benchmark? I saw a benchmark of this SSD on another site and the temperature reached about 80C after 15 minutes of use with heatsink and performance went down, hoped someone will verify this, maybe it was faulty heatsink

 

[Amdlova]

Where is te temperature benchmark? I saw a benchmark of this SSD on another site and the temperature reached about 80C after 15 minutes of use with heatsink and performance went down, hoped someone will verify this, maybe it was faulty heatsink

My adata without use stays on 45°c when have high usage goes to 68... 80°c I think will be normal to pcix 5 gen ssds high speed high Temps xD

 

[Colif]

incoming "my raid 5 PCIe drives aren't running any faster" posts

 

[JarredWaltonGPU]

Where is te temperature benchmark? I saw a benchmark of this SSD on another site and the temperature reached about 80C after 15 minutes of use with heatsink and performance went down, hoped someone will verify this, maybe it was faulty heatsink

We're looking at adding thermal results in the future, but it's an extra step and hasn't generally been part of our SSD reviews. The Write Saturation test shows 15 minutes of sustained writes (and I actually tested for two hours), with no degradation in speed after the initial "folding" period following the pSLC getting filled up. So, if someone else measured higher temperatures and reduced performance, I'm not sure what workload they're using, but my test data doesn't show a drop off.

(FYI, I'm now doing the SSD testing while Shane does the writing. Keeps things "in house" with the full-time staff.)

 

[JarredWaltonGPU]

On one test (DiskBench 50Gb File Folder), the T700 was 21% faster than the 990 Pro. But I'm not sure how that translates into real-world significance. It reminds me of the distinction between statistically significant differences and clinically significant differences (I was in the social sciences): the former are important and look good, but the latter are what count practically and often represent a genuine meaningful change and improvement.

The DiskBench Copy is probably the most "real world" if you're doing something like updating a large Steam game. Steam will often read / update / write a bunch of files during the patching process (same goes for Epic Games Store). But this is more of a special case workload that can be very storage intensive rather than something people will do on a daily or even weekly basis. Your Samsun 990 should continue to be plenty fast for quite some time, and it cost quite a bit less than these new Gen5 drives.

 

[FunSurfer]

We're looking at adding thermal results in the future, but it's an extra step and hasn't generally been part of our SSD reviews. The Write Saturation test shows 15 minutes of sustained writes (and I actually tested for two hours), with no degradation in speed after the initial "folding" period following the pSLC getting filled up. So, if someone else measured higher temperatures and reduced performance, I'm not sure what workload they're using, but my test data doesn't show a drop off.

(FYI, I'm now doing the SSD testing while Shane does the writing. Keeps things "in house" with the full-time staff.)

I'm sorry I am not sure the time was 15 minutes, this is the benchmark:

Crucial T700 preview: The fastest PCIe 5.0 SSD on the planet... for what that's worth

Headline grabbing 12,000 MB/s speeds can't hide the fact there's little to tempt gamers.

www.pcgamer.com