Review Crucial 2TB T500 SSD Review: The All-Around Gen 4 SSD

[Admin]

The Crucial T500 hits all the right notes with good performance, efficiency, and support. The heatsink option is also a nice touch. You’ll have to wait for a 4TB SKU, and the pricing needs some adjustment, though.

Crucial 2TB T500 SSD Review: The All-Around Gen 4 SSD : Read more

 

[waltc3]

Always want to point out that the so-called endurance numbers are far less important than the warranty period for all of these NVMe drives. If the drive craps out at 400TB written, but you are 7 years along, too bad... Most vigorous usage for these drives might cap out as high as 150TB-200TB in five years, so I only look at the warranty period when I buy an NVMe drive--the longer it is the better. I'm still using a SS 960 EVO 256-GB NVMe I bought ~ 4 years ago, and it's still sitting on 21TBs written, and the first 18 months or so it was my boot drive, which saw a lot of writing through beta OS installs. Today it runs games, while my boot drive is an SS 980 Pro sitting on 31Tb written. Interesting that the "endurance" number for the 960 EVO is 75TBs, and after 4+ years of constant use I'm not anywhere near that... BTW, I use my system for 8-12 hours a day. So, yeah...

 

[dimar]

What's the deal with TBW? It keeps going down with newer models.. Shouldn't they make improvement with TBW before moving to higher speeds?

 

[Makaveli]

What's the deal with TBW? It keeps going down with newer models.. Shouldn't they make improvement with TBW before moving to higher speeds?

yup seems as the flash gets faster the endurance is less.

My Gen 1 Corsair MP600 has 1,800 TBW
All of the Newer 1TB drives are at 600 TBW
And even the 2TB drives are at 1200 TBW

 

[JarredWaltonGPU]

yup seems as the flash gets faster the endurance is less.

My Gen 1 Corsair MP600 has 1,800 TBW
All of the Newer 1TB drives are at 600 TBW
And even the 2TB drives are at 1200 TBW

The thing is, realistically, no one (as a regular consumer) should ever come anywhere near the TBW figures. Like, I have a desktop PC that I've been using for 3.5 years, with a 2TB SSD. I've used it A LOT, installing and deleting games, filling it up and deleting stuff, etc. After that much time, it currently has 175 TB of total writes, according to CrystalDiskInfo. It's rated for 1280 TBW, so I'm not even 15% of the way through the endurance.

And the SSD is now "slow" compared to modern stuff. By the time it's five years old, I'll definitely be thinking about retiring the PC, or maybe putting in a larger SSD. Basically, outside of QLC (which has much worse endurance than TLC), you really have try very, very hard to hit the endurance rating within five years.

As another example, I have a 2TB Samsung 850 Pro SATA SSD. That means I've had it about eight years (!), and while I've stored plenty of stuff on it, I haven't hit it nearly as hard as the OS drive. SMART says it currently has 161TB of writes. Funny thing is, being from 2015, it's only rated for 300TBW. Still, I'm only at a bit more than half it's endurance, and it is very much just slow backup storage for me these days.

 

[Makaveli]

The thing is, realistically, no one (as a regular consumer) should ever come anywhere near the TBW figures. Like, I have a desktop PC that I've been using for 3.5 years, with a 2TB SSD. I've used it A LOT, installing and deleting games, filling it up and deleting stuff, etc. After that much time, it currently has 175 TB of total writes, according to CrystalDiskInfo. It's rated for 1280 TBW, so I'm not even 15% of the way through the endurance.

And the SSD is now "slow" compared to modern stuff. By the time it's five years old, I'll definitely be thinking about retiring the PC, or maybe putting in a larger SSD. Basically, outside of QLC (which has much worse endurance than TLC), you really have try very, very hard to hit the endurance rating within five years.

As another example, I have a 2TB Samsung 850 Pro SATA SSD. That means I've had it about eight years (!), and while I've stored plenty of stuff on it, I haven't hit it nearly as hard as the OS drive. SMART says it currently has 161TB of writes. Funny thing is, being from 2015, it's only rated for 300TBW. Still, I'm only at a bit more than half it's endurance, and it is very much just slow backup storage for me these days.

I agree most of us don't come close to that in normal desktop usage.

This is my host writes on the MP600 been in use since Dec 2019 as an OS drive until recently replaced by my 850X



WD SN850X purchased in august as new OS drive.

 

[Amdlova]

For OS I use a intel optane 280gb will last forever. For other things solidgm ssd with drivers from solidm

 

[Sleepy_Hollowed]

If this drive has their previously excellent power-loss safety features, this would be a go-to drive for consumers that care about that.

It's hard to test and the only people that care about that are those using it for work, school or both, but Crucial's top end drives used to be top notch for consumers even amongst premium brands, and they saved me more than once.

 

[Makaveli]

Will be curious to see how this drive is in a few months after a couple firmware updates and price drops.

 

[Not_Available]

I'm part of Amazon Vine, which is a group of customers who can order a lot of items for "free" in order to review them. For tax purposes companies must give a "fair market value" for their products. (We pay income tax based on that)

Crucial listed the fair market value for the Crucial T500 2TB (with heatsink) at $156.95. So I'd expect the first few sales to be around there.

 

[NWCherokee]

Does the T500 have 256 AES hardware encryption support (SED self encrypting drive) that would allow it to use hardware encryption with Windows bitlocker? Or is it strictly OPAL encryption only?

 

[TJ Hooker]

Does the T500 have 256 AES hardware encryption support (SED self encrypting drive) that would allow it to use hardware encryption with Windows bitlocker? Or is it strictly OPAL encryption only?

OPAL encryption is AES. It looks like the OPAL spec technically allows AES 128 or 256, but I don't think I've ever seen an implementation that wasn't AES 256.

 

[Ty_Bower]

I'm very concerned about the sustained write performance / saturated cache recovery, as shown about halfway through the review. I had an old Crucial MX100 SSD (512GB, 2014) that was otherwise a perfectly fine drive, but would completely fall on it's face after ~40 seconds of sustained write. It sucked big time, and left an unpleasant memory that will be difficult to forget. It's no fun at all when your system goes "hurr, durr" and the whole world stops turning until the cache recovers.

I'm glad there are reviewers that check this often overlooked aspect of storage. I think I'll skip this T500. Maybe Crucial will get it right on their next model.

 

[shaolin95]

So as a primary OS drive and main apps (photoshop, Premiere) etc... will you take this T500 over a 990 Pro or 850x?
This is for a new build not the pc on my Sig.

 

[phxrider]

I'm struggling to understand why this is being pushed as a laptop SSD (for example, Tom's currently has it as "best SSD for laptops")... The power consumption is worse than a 990 Pro and double a P31 Gold. It seems to me that a MP44/US75 type drive would still be the one to have in a PCIe 4.0 laptop despite the slight drop in performance (which will likely not be felt in real world use), and the P31 in 3.0 (if you're limited by the BUS, you might as well minimize power consumption IMO), at least if the laptop is going to be used as a laptop.

 

[JarredWaltonGPU]

I'm struggling to understand why this is being pushed as a laptop SSD (for example, Tom's currently has it as "best SSD for laptops")... The power consumption is worse than a 990 Pro and double a P31 Gold. It seems to me that a MP44/US75 type drive would still be the one to have in a PCIe 4.0 laptop despite the slight drop in performance (which will likely not be felt in real world use), and the P31 in 3.0 (if you're limited by the BUS, you might as well minimize power consumption IMO), at least if the laptop is going to be used as a laptop.

Please bear in mind that our desktop testing does not have all power saving features enabled. It's a "worst case" result, though in some cases even the "best case" results may not be that great. Phison E26 drives for example simply have very high idle power draw and no real way to fix that. Most laptops by default have much more aggressive M.2 power saving features enabled and will often drop into the sub-100mW range when idle.

The Phison E25 used in the T500 does much better, and while the idle power and average power consumption aren't the lowest we've seen, the inclusion of a DRAM cache with overall very good performance in a single-sided drive are important considerations. There are indeed drives that use less power on average, but most of those are DRAM-less solutions and/or slower drives. This is also why I personally would generally prefer the E25 solutions over the more efficient and "newer" E27T solutions.

Thermals were good on the T500 as well, peaking at 53C during our heaviest workload. Of course, that's in a desktop with good airflow and may not fully represent what you'd see in a laptop, but on a like-for-like basis it's better than many other single-sided drives, including all the Maxio MAP1602 solutions.

 

[phxrider]

Please bear in mind that our desktop testing does not have all power saving features enabled. It's a "worst case" result, though in some cases even the "best case" results may not be that great. Phison E26 drives for example simply have very high idle power draw and no real way to fix that. Most laptops by default have much more aggressive M.2 power saving features enabled and will often drop into the sub-100mW range when idle.

The Phison E25 used in the T500 does much better, and while the idle power and average power consumption are the lowest we've seen, the inclusion of a DRAM cache with overall very good performance in a single-sided drive are important considerations. There are indeed drives that use less power on average, but most of those are DRAM-less solutions and/or slower drives. This is also why I personally would generally prefer the E25 solutions over the more efficient and "newer" E27T solutions.

Thermals were good on the T500 as well, peaking at 53C during our heaviest workload. Of course, that's in a desktop with good airflow and may not fully represent what you'd see in a laptop, but on a like-for-like basis it's better than many other single-sided drives, including all the Maxio MAP1602 solutions.

Thanks for the explanation - I'm mainly going off of Tom's Hardware's average power consumption listings, which are (all 2TB):
P31 Gold - 2.28w
Silicon Power US75 - 3.04w
T500 - 4.42w
990 Pro - 4.03w

4.42w is definitely not one of the lowest averages Tom's has measured.... What am I missing? I'm sure there is "more to it", but I'm not getting it from the numbers.

Also, I am becoming less focused on DRAM, it used to be a big deal to me until I looked deeper at what the differences show up in, and realized they are mostly with moving very large amounts of data being fed to the device as fast as possible, which is rare in real world use (how often are you really copying tens or hundreds of GB between SSDs on the same system, or between folders on the same SSD)? I can confirm from having a 990 Pro and a US75 (and actually a P31 Gold too) in different systems I use regularly, there's very little real world difference and most users would never even know it had one vs the other. Bring a bit of a nerd for this stuff, I appreciate the theoretical differences and enjoy comparing them, but in a laptop I'd go for whatever eats my battery the least while still offering very good (not ultimate) performance, except maybe for a few very specific, IMO rare use cases.

 

[JarredWaltonGPU]

Thanks for the explanation - I'm mainly going off of Tom's Hardware's average power consumption listings, which are (all 2TB):
P31 Gold - 2.28w
Silicon Power US75 - 3.04w
T500 - 4.42w
990 Pro - 4.03w

4.42w is definitely not one of the lowest averages Tom's has measured.... What am I missing? I'm sure there is "more to it", but I'm not getting it from the numbers.

Also, I am becoming less focused on DRAM, it used to be a big deal to me until I looked deeper at what the differences show up in, and realized they are mostly with moving very large amounts of data being fed to the device as fast as possible, which is rare in real world use (how often are you really copying tens or hundreds of GB between SSDs on the same system, or between folders on the same SSD)? I can confirm from having a 990 Pro and a US75 (and actually a P31 Gold too) in different systems I use regularly, there's very little real world difference and most users would never even know it had one vs the other. Bring a bit of a nerd for this stuff, I appreciate the theoretical differences and enjoy comparing them, but in a laptop I'd go for whatever eats my battery the least while still offering very good (not ultimate) performance, except maybe for a few very specific, IMO rare use cases.

I said "while the idle power and average power consumption aren't the lowest we've seen" initially, but that was supposed to say aren't — I've edited it now.

I totally get where you're coming from. I've used the Neatac NV790 4TB, Samsung 990 Pro 4TB, and Crucial T700 4TB outside of the SSD test PC. There are a few things where the T700 clearly wins (like validating a Steam game install), but most of the time they all feel about the same. It's just the Neatac costs like $250, the Samsung $320, and the Crucial $450 or whatever.

I've also used a lot of laptops that come with an absolutely bottom of the barrel garbage 1TB or 512GB QLC SSD. Those suck so bad that just about anything will be a massive improvement. I haven't tried putting one of the DRAM-less Maxio MAP1602 drives into a laptop, because I worry they'll run too hot, but at least they're not QLC. Blech.

 

[Order 66]

would a DRAM cache improve file decompression speeds? I have a 7700x, and a 980 pro 2tb, and cinebench r23 takes a bit of time to decompress.

 

[phxrider]

I've also used a lot of laptops that come with an absolutely bottom of the barrel garbage 1TB or 512GB QLC SSD. Those suck so bad that just about anything will be a massive improvement. I haven't tried putting one of the DRAM-less Maxio MAP1602 drives into a laptop, because I worry they'll run too hot, but at least they're not QLC. Blech.

I know what you're talking about with the bottom of the barrel laptop SSDs, nearly every corporate laptop comes with one. I've run Crystal on mine for giggles and it's abysmal, WAY slower than the P31 Gold in my personal laptop. But in practice there are no situations where I'm tapping my fingers waiting on one while the other blows through the task. I don't game on the work laptop so I can't speak to load times or validating a large game or anything like that.

My MAP1602 drive is in a Minisforum UM780 XTX mini PC, and it reads a constant 38C. I'm actually starting to doubt the reading's validity because it NEVER moves. And it benches almost as fast as the 990 Pro in Crystal, except in one of the tests it gets clobbered. For browsing, MS Office and AAA gaming, I can't tell the difference.

 

[mouk]

Please consider benchmarking storage used in enclosures accessed by ThunderBolt 3 or newer.

 

[JarredWaltonGPU]

Please consider benchmarking storage used in enclosures accessed by ThunderBolt 3 or newer.

TB3 would drastically reduce performance, as it's PCIe 3.0 x4 equivalent. Steady state performance wouldn't get hurt too much, but other tests it would really hurt throughput on PCIe 4.0 and 5.0 drives. It would be a bit like reviewing new high-end GPUs by testing them with a Core i9-9900K (ie, CPU limited bottlenecks, where with storage it's interface bottlenecks).

The other big issue with using TB3 (or even TB4) is that the controller on both the host device and the external device are both factors. You can end up with very different results with the various options. We do have external drive reviews where we test this stuff, but we haven't done a lot with standard M.2 SSDs in an enclosure as the amount of effort required will likely not result in much benefit.