Odd SSD performance

[SMSSMS]

So I purchased a crucial MX100, 512GB, as an upgrade for my laptop (inspiron 7537). I cloned my old HDD and installed and it worked fine.

I used CrystalDiskMark and the benchmark results showed similar results to all that's out there. So everything should be fine.

However, when I tried to make a copy of a music folder (3GB), the copy rates start really good at (350MB/s), but then after 20% of the data is copied, then the performance degrades significantly, it goes down to 20 MB/s and then exhibits a sawtooth pattern going up from 20 to 120MB/s then down to 20 again and so on.

What's the matter?! Is this normal? (btw, the SSD is 70% full)

http://oi61.tinypic.com/fzclt1.jpg

 

[Coup27]

I've never seen that pattern before which does look odd, however remember that high speed writes are only for sequential large files. If you copy a 5GB movie for example it can sustain the write at high speeds because the file is a large sequential file. Once you start copying lots of small random files, such as thumbs.db from your picture, read and write speeds will tank.

At this point I'm more inclined to think that your music folder has a mixture of file sizes in it, some large and sequential and some small and random and the SSD is flirting between the two.

 

[CiPHER]

Could it be you have any antivirus/security-software installed that might be causing this?

You are copying FROM and TO the SSD, right? If you would involve a harddrive, this would have been expected.

 

[SMSSMS]

I've never seen that pattern before which does look odd, however remember that high speed writes are only for sequential large files. If you copy a 5GB movie for example it can sustain the write at high speeds because the file is a large sequential file. Once you start copying lots of small random files, such as thumbs.db from your picture, read and write speeds will tank.

I just tried to make a copy of a 2GB film, the performance still tanks, it starts at 400MB/s then goes down to 60MB/s and stays there, but the zigzag pattern is gone, see pic:

Could it be you have any antivirus/security-software installed that might be causing this?

I disabled then uninstalled my antivirus (Avast), still, the same problem. And yes, I am copying from and to the SSD

 

[SMSSMS]

Would any of you two kindly redo the same test I did above (make a copy of a 2GB film to the same folder) if you have an SSD running windows 8.1? I would appreciate it to find out if it's just my SSD.

Also, I ready somewhere that those SSDs would only have a few blocks to right on ready, so when copying large amount of data, it has to recover, could that be it?

 

[SMSSMS]

ok, problem solved, I reached out to Crucial customer support, they advised me to change the SATA driver from Intel to the generic Microsoft one, I did that and it worked! copy rates are constant and very high now (400MB/s for the film, and 200MB/s for the music folder)! Thanks folks!

I can't believe intel's driver actually did that :/

 

[CiPHER]

SSDs would only have a few blocks to right on ready

I do not understand.

But if your question is degraded SSD performance, that is extremely rare to affect sequential reads. Nothing can go wrong with sequential reads.

You should check the SMART to see if you have cabling problems (UDMA CRC Error Count). Did you do this already? I suspect something else is your problem though. You could try and use the SSD in another system and repeat your benchmarks.

ok, problem solved, I reached out to Crucial customer support, they advised me to change the SATA driver from Intel to the generic Microsoft one, I did that and it worked! copy rates are constant and very high now (400MB/s for the film, and 200MB/s for the music folder)! Thanks folks!

I can't believe intel's driver actually did that :/

Hmm, it is weird that the Intel drivers would be responsible for such low performance. I suspect some kind of issue, maybe something that is pretty unique. It would be very strange if Intel's drivers would give low speeds on SSDs on all systems. Generally, the Intel drivers are very good.

 

[SMSSMS]

well, Crucial customer support recommended it, so I guess it's a known problem.

 

[Coup27]

I don't believe there are any inherent issues with Intel's RST drivers. Can you please post your full system specs and driver versions?

 

[SMSSMS]

I agree with you it is weird.

My computer is a Dell Inspiron 15 (7537), core i7 4th gen, 8gb ram http://www.dell.com/us/p/inspiron-15-7537/pd

The Intel SATA driver (the one I replaced) version is: 13.2.4.1000 dated 25-Jul-14

 

[Coup27]

It is possible that Dell have meddled with the implementation of the SATA ports and how they operate. If is not uncommon for OEM's to "tweak" things to work how they see fit which then causes incompatibilities with standard software and drivers.

It depends how much time you want to put into this. Did you format the system before using it? If not and if it were me, I would obtain a stock Windows ISO from Digital River and reinstall the system from scratch using your OEM licence key and re-build the system and see if the issue is still present using the Intel RST driver. If it isn't, then it's possibly some Dell bloat which was on the system from new. If it does, then Dell have probably meddled with the hardware and you're just going to have to live with the MSACHI driver.

 

[Elixer]

It is possible that Dell have meddled with the implementation of the SATA ports and how they operate. If is not uncommon for OEM's to "tweak" things to work how they see fit which then causes incompatibilities with standard software and drivers.

It depends how much time you want to put into this. Did you format the system before using it? If not and if it were me, I would obtain a stock Windows ISO from Digital River and reinstall the system from scratch using your OEM licence key and re-build the system and see if the issue is still present using the Intel RST driver. If it isn't, then it's possibly some Dell bloat which was on the system from new. If it does, then Dell have probably meddled with the hardware and you're just going to have to live with the MSACHI driver.

This don't sound plausible.
If dell did do something, the standard drivers will be the first ones with the issue, since, they deviated from the standard.
Now, in this case, since crucial did tell OP to switch to MS, then, they must know there is an issue with the intel drivers, and I guess it is easier to tell them to use MS, then to avoid a certain intel revision.
It could also be a case of APM enabled on the laptop as well...

 

[Ketchup]

I am glad Crucial was able to help you find a solution, but I am a little curious as to why you are using the copy command within the same drive, unless you really do want to have more than one copy. The move (cut) function would be a lot faster.

 

[SMSSMS]

It depends how much time you want to put into this. Did you format the system before using it? If not and if it were me, I would obtain a stock Windows ISO from Digital River and reinstall the system from scratch using your OEM licence key and re-build the system and see if the issue is still present using the Intel RST driver. If it isn't, then it's possibly some Dell bloat which was on the system from new. If it does, then Dell have probably meddled with the hardware and you're just going to have to live with the MSACHI driver.

I hope it won't come to that, it's working fine now, and my curiosity to find out what is the problem isn't enough this tiresome and time consuming process.

If dell did do something, the standard drivers will be the first ones with the issue, since, they deviated from the standard. Now, in this case, since crucial did tell OP to switch to MS, then, they must know there is an issue with the intel drivers, and I guess it is easier to tell them to use MS, then to avoid a certain intel revision.

Exactly! When they told me to do that I only did it because I want them to run out of excuses and admit I have a faulty unit. But then I did it and it worked!

I am glad Crucial was able to help you find a solution, but I am a little curious as to why you are using the copy command within the same drive, unless you really do want to have more than one copy. The move (cut) function would be a lot faster.

You are correct, this copy operation is pointless, I was merely testing SSD, nothing else. Wanted to enjoy the sight of crazy fast copy rates after years of HDD agony

 

[Ketchup]

... You are correct, this copy operation is pointless, I was merely testing SSD, nothing else. Wanted to enjoy the sight of crazy fast copy rates after years of HDD agony

Gotcha. Another fun thing I have found is installing an OS on a VM. Takes no time at all.