Crucial MX100 only 154MB/s read??

[omega3]

Regarding the MX100, i found a review bij CDRLabs where they notice that even after leaving the drive idle for a couple of hours so that trim and garbage collection should get the drive back up in speed, it only climbed up to 154.1MB/s. Isn't that rather slow for an SSD drive?

A secure erase gets it back to around 350MB/s.

PS. aside, can anybody mention how much read a 5400 rpm drive can get.. I read reviews mentioning up to 100MB/s which seems a bit much, no?

 

[ggadrian]

Black magic just reported 259.3MB/s read and 203MB/s write over a SATA2 interface for my 256GB MX100, 154MB/s is unacceptably slow, unless the disk is completely full or something like that.

 

[Hulk]

Regarding the MX100, i found a review bij CDRLabs where they notice that even after leaving the drive idle for a couple of hours so that trim and garbage collection should get the drive back up in speed, it only climbed up to 154.1MB/s. Isn't that rather slow for an SSD drive?

A secure erase gets it back to around 350MB/s.

PS. aside, can anybody mention how much read a 5400 rpm drive can get.. I read reviews mentioning up to 100MB/s which seems a bit much, no?

A couple of other review sites noted the same lack of trim performance from both the MX100 and MX200. The BX100 recovered well.

 

[omega3]

A couple of other review sites noted the same lack of trim performance from both the MX100 and MX200. The BX100 recovered well.

Do you have links about the B100 trim performance?

I still don't get the MX100 could slow down that much. Alot of conflicting reviews in that respect. Not sure what to believe anymore.

And as mentioned before.. can a classic 5400 rpm go to 100MB/s.. seems a bit much no?

 

[ggadrian]

Do you have links about the B100 trim performance?

I still don't get the MX100 could slow down that much. Alot of conflicting reviews in that respect. Not sure what to believe anymore.

And as mentioned before.. can a classic 5400 rpm go to 100MB/s.. seems a bit much no?

Yes, in sequential reads it can get about 100MB/s, although some drives stay closer to 85-90MB/s

 

[omega3]

Yes, in sequential reads it can get about 100MB/s, although some drives stay closer to 85-90MB/s

But what really matters in every day performance.. sequential or 4K (QD32) benchmarks?

 

[Hulk]

Do you have links about the B100 trim performance?

I still don't get the MX100 could slow down that much. Alot of conflicting reviews in that respect. Not sure what to believe anymore.

And as mentioned before.. can a classic 5400 rpm go to 100MB/s.. seems a bit much no?

Here's one comparing the MX200 and BX100. The MX200/100 use the same controller.
http://www.legitreviews.com/crucial-mx200-bx100-series-ssd-review_158644/7

I read the same thing in a few MX100 reviews as well but can't find them at the moment. I actually had a MX100 on order and while doing further research found out about this and cancelled the order.

 

[omega3]

Here's one comparing the MX200 and BX100. The MX200/100 use the same controller.
http://www.legitreviews.com/crucial-mx200-bx100-series-ssd-review_158644/7

I read the same thing in a few MX100 reviews as well but can't find them at the moment. I actually had a MX100 on order and while doing further research found out about this and cancelled the order.

The bench below from that site really confuses me .. it seems the 850 Evo recovers very well but why does it start so low like only around 50MB/s where Sandisk and OCZ are more like 350 and 250MB/s.. the 850 Evo can't be that bad, huh??

And the Crucial MX100 (brown line) actually recovers VERY well BUT starts the LOWEST at only 25MB/s.. can anybody clarify this cause i'm really confused by this bench below regarding both MX100 and 850 Evo

 

[Cerb]

Regarding the MX100, i found a review bij CDRLabs where they notice that even after leaving the drive idle for a couple of hours so that trim and garbage collection should get the drive back up in speed, it only climbed up to 154.1MB/s. Isn't that rather slow for an SSD drive?

If it's 256GB model, yes. You don't mention whether yours is exactly that size, or not.

 

[omega3]

If it's 256GB model, yes. You don't mention whether yours is exactly that size, or not.

Yes i was referring to the 256GB model which has a max of 550MB/s on paper.. why does it only get 150MB/s in the review above?

Also, regarding my post 2 up with the graph and my comments regarding MX100 and 850 Evo which both start very low IO in that graph in comparison to sandisk and ocz which i didn't consider to be better. What does that graph mean?

 

[Hulk]

The way I see it the drives should have been hammered (degraded) to the point where they were performing very badly. Then they're given a break and they should recover. The MX100/200 don't seem to recover very well.

The three drives that are performing well the whole time either weren't hammered hard enough or have more performance consistency than the other drives I would assume.

 

[omega3]

The way I see it the drives should have been hammered (degraded) to the point where they were performing very badly. Then they're given a break and they should recover. The MX100/200 don't seem to recover very well.

The three drives that are performing well the whole time either weren't hammered hard enough or have more performance consistency than the other drives I would assume.

Actually, according to the graph above, the MX100 (greenish line) seems to recover equally well as the BX100 (see how the line goes up) .. it bothers me more that it's begin point - so before the drive was hammered - is so low (50MB/s).

What i also don't understand is the lowest begin point for the 850 Evo (25MB/s), however that drive recovers the best.

Would love everybody's opinion on that.

 

[Cerb]

It's a low-end consumer drive. If you intend to perform large amounts of random writes across the whole LBA set, drives with pSLC caches should (and appear to, except for the MX200 in that test) do better. There are any number of reasons why that drive might not. Maybe it's using lower grade MLC, and throttles to protect it. Maybe the controller just sucks at it. Who knows. That's what the higher-end drives are for.

When not abusing them, they seem to do fine (pages 3-5). If you want to know how well some 5400 RPM HDD would do, in a similar test, the best answer is that most reviewers aren't going to be patient enough to try to find out. Some new ones can do around 150MBps sequential, but that's in comparison to the fresh/TRIMed 500+ MBps of SSDs, which is partly SATA itself.

Precondition Phase:
1. Write to the drive sequentially through up to the reported capacity with random data.
2. Write the drive through a second time (to take care of overprovisioning).

Degradation Phase:
1. Run writes of random size between 8*512 and 2048*512 bytes on random offsets for 10 minutes.
2. Run performance test (one pass only).
3. Repeat 1 and 2 for 8 times, and on each pass increase the duration of random writes by 5 minutes.

Steady state Phase:
1. Run writes of random size between 8*512 and 2048*512 bytes on random offsets for 50 minutes.
2. Run performance test (one pass only).
3. Repeat 1 and 2 for 5 times.

Recovery Phase:
1. Idle for 5 minutes.
2. Run performance test (one pass only).
3. Repeat 1 and 2 for 5 times.

That is not a reasonable test, as far as what to expect from the drive, actually using it. It's a test to see what drives are really good, and what drives royally screw up. The results are in between the 4K read and 4K write results from CDM and AS-SSD, which is reasonable.

For the 5400 RPM HDD comparison in the OP, BTW, now that I am at the right PC, here's a WD Purple (hence the read/write disparity) 3TB's CDM results:

-----------------------------------------------------------------------
CrystalDiskMark 3.0.3 Shizuku Edition x64 (C) 2007-2013 hiyohiyo
                           Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

           Sequential Read :   106.163 MB/s
          Sequential Write :   106.998 MB/s
         Random Read 512KB :    20.968 MB/s
        Random Write 512KB :    54.652 MB/s
    Random Read 4KB (QD=1) :     0.280 MB/s [    68.4 IOPS]
   Random Write 4KB (QD=1) :     0.947 MB/s [   231.2 IOPS]
   Random Read 4KB (QD=32) :     0.696 MB/s [   169.9 IOPS]
  Random Write 4KB (QD=32) :     0.953 MB/s [   232.6 IOPS]

  Test : 1000 MB [D: 64.4% (1798.9/2794.4 GB)] (x5)
  Date : 2015/04/10 23:07:37
    OS : Windows 7 Professional SP1 [6.1 Build 7601] (x64)

Chances are, it would score at 0.5-1MBps in the same kind of testing.

 

[omega3]

It's a low-end consumer drive. If you intend to perform large amounts of random writes across the whole LBA set, drives with pSLC caches should (and appear to, except for the MX200 in that test) do better. There are any number of reasons why that drive might not. Maybe it's using lower grade MLC, and throttles to protect it. Maybe the controller just sucks at it. Who knows. That's what the higher-end drives are for.

When not abusing them, they seem to do fine (pages 3-5). If you want to know how well some 5400 RPM HDD would do, in a similar test, the best answer is that most reviewers aren't going to be patient enough to try to find out. Some new ones can do around 150MBps sequential, but that's in comparison to the fresh/TRIMed 500+ MBps of SSDs, which is partly SATA itself.

That is not a reasonable test, as far as what to expect from the drive, actually using it. It's a test to see what drives are really good, and what drives royally screw up. The results are in between the 4K read and 4K write results from CDM and AS-SSD, which is reasonable.

For the 5400 RPM HDD comparison in the OP, BTW, now that I am at the right PC, here's a WD Purple (hence the read/write disparity) 3TB's CDM results:

-----------------------------------------------------------------------
CrystalDiskMark 3.0.3 Shizuku Edition x64 (C) 2007-2013 hiyohiyo
                           Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

           Sequential Read :   106.163 MB/s
          Sequential Write :   106.998 MB/s
         Random Read 512KB :    20.968 MB/s
        Random Write 512KB :    54.652 MB/s
    Random Read 4KB (QD=1) :     0.280 MB/s [    68.4 IOPS]
   Random Write 4KB (QD=1) :     0.947 MB/s [   231.2 IOPS]
   Random Read 4KB (QD=32) :     0.696 MB/s [   169.9 IOPS]
  Random Write 4KB (QD=32) :     0.953 MB/s [   232.6 IOPS]

  Test : 1000 MB [D: 64.4% (1798.9/2794.4 GB)] (x5)
  Date : 2015/04/10 23:07:37
    OS : Windows 7 Professional SP1 [6.1 Build 7601] (x64)

Chances are, it would score at 0.5-1MBps in the same kind of testing.

Isn't it so that 4K and also 512KB test results are actually more relevant then sequential?

Aside, SSD has the mentioned benifts however I'm still not convinced that an SSD makes much diff vs hdd when using chrome even with 15 tabs open.. its said that ssd is better for multitasking and multiple tabs in chrome could actually be seens as multitasking since each tab runs on its own.. however i don't see why ssd would benifit from multitasking?

 

[Cerb]

Any time synchronous IO has to hit the drive, the SSD can do it in a fraction of a millisecond, while the HDD takes over 10 milliseconds, and as the queue grows, can take much longer. How that may practically affect your regular usage of the PC is your judgment to make.

 

[omega3]

Any time synchronous IO has to hit the drive, the SSD can do it in a fraction of a millisecond, while the HDD takes over 10 milliseconds, and as the queue grows, can take much longer. How that may practically affect your regular usage of the PC is your judgment to make.

We all know SSD starts up progs faster but i see claims that for example webpages will load faster in a brower, esp with multi tabs, however i don't see the logic behind those claims. That's what i was wondering about.

 

[Cerb]

We all know SSD starts up progs faster but i see claims that for example webpages will load faster in a brower, esp with multi tabs, however i don't see the logic behind those claims. That's what i was wondering about.

Reads tend to be synchronous. That is, they block the program from continuing until complete. Writes tend not to be, but then sometimes flushes get called, making them similar. If the browser is having to wait on the disk, it will do so less with an SSD. IME, there's was a difference, but a pretty minor one, and that was with upgrading multiple CPU generations at the same time (not in general, but specifically with Chrome). The biggest difference was during the browser's session restoring after a reboot, for me. YMMV.

SSDs speed up random access to block storage by orders of magnitude, compared to HDDs. Anything that is significantly bottlenecked by file access will get a big boost. Anything not so bottlenecked will not. In some cases, different users of the same programs may, depending on their own efficiencies and work flows, be affected differently. Look at the disk busy time in resource monitor. Any time it's in the 80-100% range, an SSD would speed things up dramatically. But, like in the other thread: if you're making a new computer, get an SSD, and not one too small for your needs. Any with a good reputation will do the job well enough. With prices of 240GB and larger going below $0.35/GB on a regular basis, and staying <$0.45/GB daily, I think it's hard to argue against the value of going with an SSD for your C: drive, today.

 

[VirtualLarry]

With prices of 240GB and larger going below $0.35/GB on a regular basis, and staying <$0.45/GB daily, I think it's hard to argue against the value of going with an SSD for your C: drive, today.

This. I've seen 240/256GB SSDs for $70 AR. I picked up a couple of M500 240GB drives for $79.99 ea. at Newegg some months ago when they were fire-saleing them for a week or two.

 

[omega3]

If it's 256GB model, yes. You don't mention whether yours is exactly that size, or not.

I looked over this previous post by you but why does it make difference if it's the MX100 256 or 512GB model?

 

[Cerb]

I looked over this previous post by you but why does it make difference if it's the MX100 256 or 512GB model?

The 512GB model is faster than the 256GB model, which is faster than the 128GB model.

 

[AlienTech]

Regarding the MX100, i found a review bij CDRLabs where they notice that even after leaving the drive idle for a couple of hours so that trim and garbage collection should get the drive back up in speed, it only climbed up to 154.1MB/s. Isn't that rather slow for an SSD drive?

A secure erase gets it back to around 350MB/s.

PS. aside, can anybody mention how much read a 5400 rpm drive can get.. I read reviews mentioning up to 100MB/s which seems a bit much, no?

NCQ makesa HUGE difference in speeds when the drive fills up. I used a 2TB green drive for years and it worked fine until it got to like 70-80% full at which point it slowed down to a crawl. I noticed this for other drives as well.. BUT not for like the 2TB 7200 drives until over 95% full. Going from a 5400 to 7200 showed a huge difference.. Going from a 7200 to SSD was not noticable. Number of platters matter.. Lower is better.. That 1 MS difference between 5 platters and 2 platters shows up.. The 5400 to 7200 difference is only like 50MB speeds, is not noticable except for the problem with slow downs after the drive gets filled. Not sure why the 7200 does not show that. Many drives dont have NCQ.. If they had it on 160GB drives from a decade ago then obviously its not because it is expensive or difficult to do, they disable it because they dont want people to get lower priced drives. 32MB ram or 64MB ram also shows like a 10% improvement, the bigger difference will show up in marginal techonological companies tryign to maximize profits but using below average parts. IE platters that are rejected for higher end drives and would have a lot of sectors remapped and have to be run at lower RPM's. Although the remapped sectors are on the same track, that extra 32MB RAM to cache the rest of the data will show up while with the 32MB RAM that data will have to be re-read to be sent over.

 

[Cerb]

Many drives dont have NCQ.

False. Almost all drives starting from the mid/late-00s had support for NCQ. I'm pretty sure no SATA 3Gbps drives lacked it, only most 1.5Gbps ones.

Also, NCQ makes a huge difference no matter how filled up the drive is.

 

[AlienTech]

False. Almost all drives starting from the mid/late-00s had support for NCQ. I'm pretty sure no SATA 3Gbps drives lacked it, only most 1.5Gbps ones.

Also, NCQ makes a huge difference no matter how filled up the drive is.

I know this is the internet and everyone is an expert and only lacking in practical experience, having 5 2TB hard drives and a 3 TB hard drive bought last year says you are wrong. Since I cant attach a screen shot here is the info on it. Also I did nto say anything about NCQ making a difference in speed and how filled up the drive is, I said it made a difference because in the RPM. Does not matter who makes the drive, I noticed it with Samsung, WDC, Seagate drives all behaving the same way. It should not matter but it does for some reason. The difference is very very noticeable. Also I bet no one ever thinks getting an error somehow when you use the computer normally would make a difference, well it does.. Its the way windows works.. Once it gets an error it drops the speeds until you reset the values. not many people know this, but it used to be a huge deal with PATA drives where an error would force windows to stop using DMA and slow down speeds drastically.. We dont even think about this with SATA drives but it seems that problem still persists and for some reason Smart values for such things makes windows behave differently. Since I dont know enough about it other than to notice it makes a difference I cant say more on it right now since it IS happening to me as I am moving large amounts of files over and a speed drop of 50% means it taking twice as long and it matters when you are talking about 8 hours to do something.


Model : ST3000DM001-1E6166
Firmware : SC48
Serial Number : ######
Disk Size : 3000.5 GB (8.4/137.4/3000.5/3000.5)
Buffer Size : Unknown
Queue Depth : 1
# of Sectors : 5860533168
Rotation Rate : 7200 RPM
Interface : Serial ATA
Major Version : ATA8-ACS
Minor Version : ATA8-ACS version 4
Transfer Mode : SATA/300 | SATA/600
Power On Hours : 1064 hours
Power On Count : 97 count
Temperature : 41 C (105 F)
Health Status : Good
Features : S.M.A.R.T., APM, 48bit LBA
APM Level : 8080h [ON]
AAM Level : ----

My 5TB toshiba says Features : S.M.A.R.T., APM, 48bit LBA, NCQ, I could post the other 5 drives values as well. And it is because people become experts and think they know that causes such confusion since we take it for granted that if we do the same thing, our results will match and when it dont we think we did something wrong rather than we should not have taken the info without further research ourself.

 

[coercitiv]

I know this is the internet and everyone is an expert and only lacking in practical experience, having 5 2TB hard drives and a 3 TB hard drive bought last year says you are wrong. Since I cant attach a screen shot here is the info on it.

Model : ST3000DM001-1E6166
Firmware : SC48
Serial Number : ######
Disk Size : 3000.5 GB (8.4/137.4/3000.5/3000.5)
Buffer Size : Unknown
Queue Depth : 1
--snip--

Are you talking about the drive in this specsheet?

 

[Cerb]

I know this is the internet and everyone is an expert and only lacking in practical experience, having 5 2TB hard drives and a 3 TB hard drive bought last year says you are wrong. Since I cant attach a screen shot here is the info on it.

...and it supports NCQ, so I don't get it. Lack of NCQ support was rare by the time we had 133GB platters.

Also I did nto say anything about NCQ making a difference in speed and how filled up the drive is

That was your first sentence.

I said it made a difference because in the RPM.

Which it doesn't, in any relative sense. 7200RPM v. 5400RPM, all else being equal, can be a >30% difference. Comparing drives made for low power v. high performance, it can be much more. It also makes quite a difference with 0 RPM. The 7200 RPM drives get it just as bad, or not, compared to 5400 RPM. The data you need to read and write makes a big difference. Push the 7200 RPM drive, and the difference will be big. Like say, updating a VM while trying to use the host desktop, before going to SSDs...