Crucial MX100 vs 550 vs Samsung Pro

[omega3]

Comparison benchmarks Crucial M550 vs 840 Pro 256GB

According to the above benchmarks, the M550 256GB would be equally good as the 840 Pro 256GB.. is that even possible as i thought the sammy pro's where above & beyond??

 

[Cerb]

The 840 Pro has a bit lower latency for very small QDs. The M550 takes lots of IOPS in stride better. No, the Samsung isn't above and beyond anything. It was clearly tuned for disk-intensive content creation workloads, where it shines, while the M500 and M550 were closer to server drives, in their performance.

The one from the past that was above and beyond others was the Corsair Neutron, that has finally been supplanted by the Samsung 850 Pro and Sandisk Extreme Pro.

 

[omega3]

The 840 Pro has a bit lower latency for very small QDs. The M550 takes lots of IOPS in stride better. No, the Samsung isn't above and beyond anything. It was clearly tuned for disk-intensive content creation workloads, where it shines, while the M500 and M550 were closer to server drives, in their performance.

The one from the past that was above and beyond others was the Corsair Neutron, that has finally been supplanted by the Samsung 850 Pro and Sandisk Extreme Pro.

Well, the benchmarks over here show the evo's better then both crucials.. also in things like IOPS and storagemark.. don't understand those benches can be so opposite.

However, when having to choose between the MX100 or M550.. it seems the latter has better write speed.. does it make sense to pick the 550 over the mx100?

 

[CiPHER]

The benchmarks can be opposite because the EVO uses a technique that allows its NAND to be used as SLC instead of MLC/TLC. This means, however, that the high speeds are only temporary, which will distort a lot of benchmarks.

The MX200/M600 is a bit more sophisticated, since it has a dynamic size for the dynamic write acceleration technology, as Micron calls it. This means that if you use the SSD <50% capacity, you are basically using it as an SLC drive.

 

[omega3]

The benchmarks can be opposite because the EVO uses a technique that allows its NAND to be used as SLC instead of MLC/TLC. This means, however, that the high speeds are only temporary, which will distort a lot of benchmarks.

The MX200/M600 is a bit more sophisticated, since it has a dynamic size for the dynamic write acceleration technology, as Micron calls it. This means that if you use the SSD <50% capacity, you are basically using it as an SLC drive.

Does the MX100 also have the dynamic write you mention?

Also, i just found out that the reason the mx100 vs 550 being slower has to do with the size.. in smaller sizes the mx100 just has lower write specs.

This being said.. i have the choice between the 256GB mx100 and 550 at the same price.. does it make sense to choose the older 550?

 

[CiPHER]

MX100 is the budget version of the M550. They use the same Marvell-controller, but use different NAND; but MX100 uses cheaper NAND iirc.

But even so, the MX100 currently is the best SSD that i recommend to consumers, because of its good controller chip with many advanced features including 'RAID5' bitcorrection (Micron calls it RAIN - like RAID but not Disks but with NAND) and power-safe capacitors, which is rare for a consumer-grade SSD. It is also very cheap so no reason not to buy it really. The MX100 512GB is almost capped for SATA/600 when dealing with sequential writes; though the Samsung push a little bit more juice out of the interface. But really, if you want high MB/s then buy multiple smaller drives and easily get 2GB/s+ speeds on an addon HBA/controller and about 1200-1800MB/s on your chipset HBA/controller.

The MX200 is basically the M600. It has the dynamic write feature that Micron wanted after Samsung having so much success selling their TLC SSDs for more money than Micron's MLC drives. Samsung has done very well at cutting the costs, and their marketing-oriented technical features cause it to perform well in benchmarks and thus be popular by consumers. It is still one of the best consumer-grade SSDs next to Crucial's offerings, but you could say they are the OCZ-light. Very popular, but not really the best choice.

Remember, SSDs vary very little in performance, but they do vary in other areas such as features, reliability and price.

 

[Face2Face]

I posted my results in the other Crucial MX100 thread, so I guess I'll post it again here.

My MX100 256GB vs. a friends 850 Pro 512GB - Not bad for $100 SSD.

 

[coercitiv]

My MX100 256GB vs. a friends 850 Pro 512GB - Not bad for $100 SSD.

In more realistic workloads, the Samsung is anywhere from 20% to 150% faster in combined loads.

However, to put it in a very simple sentence, if you don't know why you need the 850 Pro, you should get the Crucial drive.

 

[CiPHER]

Still it's better to get two Crucial 256, than to buy one Samsung 512, if you want high MB/s for unrealistic workloads like very heavy video editing with large 4K video files etc.

Realistic I/O is very much lighter than all the benchmarks that get shown to people. In fact, the 25MB/s random read performance that basically every SSD has, is the prime reason an SSD feels so fast. This was the area that harddrives had been lacking in terms of performance. In reality consumer workloads use alot more reads and the writes are much more sporadic and not as-fast-as-the-ssds-can-take-it.

All the Trace&Replay benchmarks that AnandTech and other sites use, removes the idle time between I/O-request during I/O-capture - and replays the I/O at maximum speed with given queue depth. That is fine, but that means it does not mimic a realistic workload.

 

[Hulk]

What do you think the SLC mode endurance of the MX200 would be like? 10,000 writes? 20,000 writes? If you keep it less than half full the GB you could write to it would be off the charts I would assume?

 

[CiPHER]

Yes, it's an incredible feature as it allows more flexible usage of the NAND. Probably Samsung will follow though, with similar functionality. But personally I'd like better parity protection; like triple parity (3/138) instead of single parity bitcorrection. That would make it even more suitable as server SSD.

 

[coercitiv]

Still it's better to get two Crucial 256, than to buy one Samsung 512, if you want high MB/s for unrealistic workloads like very heavy video editing with large 4K video files etc.

Why would I not buy 2x Samsung 256?

Again, the Crucial drive is the best option for most consumers, but why stretch these comparisons?

 

[Ramses]

Having had an mx100, 840 evo and 850 pro, all 500gb, there is jack all difference in performance for regular daily normal computer stuff that I can tell. All three worked perfectly, zero issues. While I might avoid the 840 till some final word comes on the speed issues they developed, although mine patched and seemed to test out fine (I actually had two), I wouldn't loose sleep over it if I got a good deal on em. The MX100 was fine, the 850 Pro is great but expensive. I'd buy on price/size if I was buying right now.

 

[aman74]

Why would I not buy 2x Samsung 256?

Again, the Crucial drive is the best option for most consumers, but why stretch these comparisons?

Price/Performance. You were concerned about performance so one would think you'd want the best and cheaper to boot? You could almost buy a third drive with the cost savings and still be faster.

 

[omega3]

Having had an mx100, 840 evo and 850 pro, all 500gb, there is jack all difference in performance for regular daily normal computer stuff that I can tell. All three worked perfectly, zero issues. While I might avoid the 840 till some final word comes on the speed issues they developed, although mine patched and seemed to test out fine (I actually had two), I wouldn't loose sleep over it if I got a good deal on em. The MX100 was fine, the 850 Pro is great but expensive. I'd buy on price/size if I was buying right now.

Regarding "daily normal computer stuff".. do you also see improvement with a browser like Chrome with 15-20 tabs open (which i do alot) or is that purely a CPU and RAM thing where SSD won't make a difference compared to HDD?

I read somewhere that a browser constantly writes small files to disk, so then ssd would benifit, but i don't see why a browser would constantly do that.

 

[aman74]

In this review of the MX200, which I imagine is similar?...they're saying it slows down 30% when it's half full?

http://www.hardwarecanucks.com/foru...68695-crucial-mx200-500gb-1tb-ssd-review.html

 

[Ramses]

Regarding "daily normal computer stuff".. do you also see improvement with a browser like Chrome with 15-20 tabs open (which i do alot) or is that purely a CPU and RAM thing where SSD won't make a difference compared to HDD?

I read somewhere that a browser constantly writes small files to disk, so then ssd would benifit, but i don't see why a browser would constantly do that.

I regularly run two browsers with 20+ tabs for work and personal stuff, usually Chrome anymore. It's been so long since I had a regular hard drive, and at that it was a 10K raptor which was no slouch among spinning disks, but I'd be astounded if you didn't see a positive difference with an SSD as it fetches cached images and such. Lots of RAM is the big thing with many tabs open, I bust 8gb in use regularly, 16gb is my minimum to not have to curtail my usage habits.

 

[CiPHER]

Why would I not buy 2x Samsung 256?

Well, performance wise a RAID0 of two Samsungs 256GB would be better than RAID0 of two Crucial MX100 @ 256GB because only the 512GB has almost-capped SATA/600 write performance. The 256 and especially 128GB versions of the MX100 are quite a bit lagging in this area. Something to be fixed with MX200 with its dynamic write.

But regardless of performance, i have some concerns about the software protection used in Samsung SSDs when it concerns anything other than regular desktop usage. This has to do with the feature that reverts the SSD to an earlier state. For desktop this is not a problem at all. But what if some SSDs in a RAID0 array revert, while others don't? Then you have a filesystem spanned across all SSDs with parts new and parts old data. Probably all you'll see is some filesystem check when you boot. But this can really hurt things potentially as the filessytem is only designed to cope with lost recent writes AFTER a flush request. The flush request (FLUSH CACHE in ATA-terminology) is holy to maintain integrity, and Samsung SSDs do not honor it.

When everything reverts, so does the metadata. The result will reflect a filesystem that is consistent (i.e. the metadata matches the stored data). Perhaps lost recent writes but the 64MiB journal takes care of that.

But if parts revert, your filesystem will experience a condition where parts of the metadata may not match the data, and parts of the metadata may not even match other metadata. This could be deadly if you have important things stored. Though i guess the filesystem check will do its work.

So what could be the result? Well, it could mean Windows won't boot anymore, for example, and you get some blue screen when booting. But people will think oh its power supply oh its the BIOS oh its the videocard, etc. A reinstall and it works again. Hmm maybe it was the driver i installed last week, yes that could be it! I can cry about these kind of things, which i regularly see on the Dutch techforums that i roam.

Again, the Crucial drive is the best option for most consumers, but why stretch these comparisons?

Well i consider them both excellent SSDs for desktop usage. But the Crucial MX100 is quite a bit (20%) cheaper than Samsung EVO, which is a TLC SSD. And the Crucial is slightly better in my opinion thanks to its hardware protection. But i agree the quality difference for desktop users is negligible. But the 20% price difference is hard to deny, unless the difference is much smaller in the US?

 

[coercitiv]

But regardless of performance, i have some concerns about the software protection used in Samsung SSDs when it concerns anything other than regular desktop usage.

I had some concerns regarding data integrity with using stripped array vs single drive to begin with, but I figured we were comparing performance only.

As to the journal system Samsung uses, although I understand your line of reasoning, I would also like to see at least some form of proof/source that makes it a real concern. Fact is even with the partial hardware protection Crucial has, a power loss can lead to corrupted files. There's no going around that unless you have full power loss protection.

Price/Performance. You were concerned about performance so one would think you'd want the best and cheaper to boot? You could almost buy a third drive with the cost savings and still be faster.

If the Samsung drive is significantly faster than the Crucial drive under heavy workloads, how do you reckon pitting stripped arrays of the same drives will reverse the situation?

 

[mrpiggy]

As to the journal system Samsung uses, although I understand your line of reasoning, I would also like to see at least some form of proof/source that makes it a real concern. Fact is even with the partial hardware protection Crucial has, a power loss can lead to corrupted files. There's no going around that unless you have full power loss protection.

Here's an article that discusses journaling issues with SSD's..

http://lwn.net/Articles/349970/

Relevant section:

It is not the lack of journaling for data which has created grief for Pavel and others, though. The nature of flash-based storage makes another "interesting" failure mode possible. Filesystems work with fixed-size blocks, normally 4096 bytes on Linux. Storage devices also use fixed-size blocks; on traditional rotating media, those blocks are traditionally 512 bytes in length, though larger block sizes are on the horizon. The key point is that, on a normal rotating disk, the filesystem can write a block without disturbing any unrelated blocks on the drive.

Flash storage also uses fixed-size blocks, but they tend to be large - typically tens to hundreds of kilobytes. Flash blocks can only be rewritten as a unit, so writing a 4096-byte "block" at the operating system level will require a larger read-modify-write cycle within the flash drive. It is certainly possible for a careful programmer to write flash-drive firmware which does this operation in a safe, transactional manner. It is also possible that the flash drive manufacturer was rather more interested in getting a cheap device to market quickly than careful programming. In the commodity PC hardware market, that possibility becomes something much closer to a certainty.

What this all means is that, on a low-quality flash drive, an interrupted write operation could result in the corruption of blocks unrelated to that operation. If the interrupted write was for metadata, a journaling filesystem will redo the operation on the next mount, ensuring that the metadata ends up in its intended destination. But the filesystem cannot know about any unrelated blocks which might have been trashed at the same time. So journaling will not protect against this kind of failure - even if it causes the sort of metadata corruption that journaling is intended to prevent.

As long as the SSD has enough power loss protection to write the entire blocks, which Crucial power loss solution should have, it won't have the surrounding block corruption as described in the article.

 

[Emulex]

doesn't the newer crucial drives use 16KB pages? resulting in larger erase time latency?

The 850-pro uses 1st gen V-NAND which is based on 40~NM fab which is far more stable than current 19nm MLC as far as page-erase cycles..

 

[Cerb]

doesn't the newer crucial drives use 16KB pages? resulting in larger erase time latency?

Yes, and I doubt it. The program and erase times have gone up a little at a time as they have been shrinking the memory, and were similar between page and block sizes when they offered multiple sizes on prior generations.

 

[coercitiv]

Here's an article that discusses journaling issues with SSD's..

That article discusses issues with filesystem journaling, not FTL journaling.

PS: this FTL journaling is getting interesting, will have to do some reading during next weekend