My 1TB Evo is very slow compared to my Corsair Performance Pro 256GB. Why?

Hi All,

I bought the corsair perf pro 256GB when I bought the Lenovo W520 around 20 months ago thanks for this forum.

This SSD is very, very fast when performing operations on VMWare (SharePoint, Visual Studio etc).

Recently, I bought the Samsung 1TB EVO because of the size as the corsair couldn't hold many VMWare images. But, I'm very disappointed with it's performance compared to the corsair.

It seems as if I'm running HDD. Very slow, sluggish, compiling and deploying code to SharePoint takes time. I'm very disappointed. I'm thinking of putting back my corsair SSD.

Why is the EVO slow compared to the corsair? I expect the EVO to be faster hence it's a newer drive

 

Did you do a clean install or clone? What O/S are you running? How much RAM do you have? How full is the EVO? Have you tried (though I'm NOT recommending) RAPID?

Is the partition properly aligned? Do you have Intel RST installed (12.8 WHQL)? Have you left at least 30% of the capacity as 'unallocated'? (Anand still recommends leaving 25% as 'unallocated' too).


Even with all of the above getting a 'pass'; If you are using much more than 50% of the EVO's capacity, I would expect it to be slower than any 'normal' SSD.

Why? Because the TurboWrite tech it uses to hit 400MB/s (for short bursts) just hides the fact that it is really a 150MB/s drive in sustained, real world use. Yeah; the Corsair IS much faster and I would agree that the Samsung 'sluggishness' would be noticeable in your workloads (too).

If the questions I start out with can be answered satisfactorily (above) - and there is no performance improvement - I'd be looking to replace the decidedly consumer EVO (with the faux SLC nand) with at least the Corsair you have (if not the SanDisk Extreme II).

See:
AnandTech | Samsung SSD 840 EVO Review: 120GB, 250GB, 500GB, 750GB & 1TB Models Tested


Hope some of this helps.

Good luck.

 

Can you put some benchmarks here to compare to others EVO?

If you feel it is like an HDD then for sure there is some issue.

 

I did a clean install. Never clone before. Running Windows 7 Ultimate 64 bit i7-2760QM 32GB Ram. No RAPID . I still have 720GB free. It's not partitioned. I've never done any benchmark on both SSDs before.

 

Have you installed the Magician software, under the support tab then downloads tab? Support - Solid State Drives MZ-7TE1T0 | Samsung Memory & Storage

EDIT: Check especially that the Firmware is up to date.

 

No Magician software & No firmware update yet. I'll do a firmware update but I don't think it'll help

 

We'll see soon

It's a fast rated drive! Need to find out quickly if it's faulty and needs exchanged/replaced RMA.


EDIT: Also, let us know if the AHCI or IRST (one or the other) driver is loading/loaded. AHCI will list in device manager under IDE ATA/ATAPI. IRST Intel Rapid Storage Tech would be a different device category (can't remember). System Mfr kind of determines which one that will be using default install.

 

It's listed as AHCI under device manager. I'll update the firmware this weekend. Thanks

 

How can I update the firmware of my EVO SSD with erasing the data?

 

I don't think you need to erase the data, but I would definitely backup and verify that the backup is good.

 

Do you have any link\instruction on how to update firmware?

 

You'll need to install the Magician software and run it through the Firmware updater there. Only do it that way. Again - backup first!

I have to check out - it's 215AM here in California. Tired!

 

Hi chukwe,

Per your PM, glad to hear the Firmware update went easily. Me and everyone else is surely wondering if the performance improved afterwards???

I'll preface the following with the fact that with all my Samsung SSDs, I can't use Magician software because of: 1. the OEM firmware weirdness associated; and 2. I can't put my 840 into the primary bay because of #1. So I can't install it.

I'm not familiar enough with the new EVO's to really be able to comment on what the best settings are (and there are new one's) and I can't find enough screenshots other than this one: http://www.samsung.com/global/busin...s/Samsung_Magician_421_Installation_Guide.pdf

I would think there are quite a few settings - some just for EVO's, but that it should do many of them for you after picking a profile that suits your usage.

Please let us know how this goes

 

The evos will have weak write performance, theres no ifs or maybes. Its because of the TLC flash, the Samsung cache only helps for frequently used stuff and only about 1-2gb worth.

 

^They're not 840 Pro's, but they should not feel like an HDD either.
AnandTech | Samsung SSD 840 EVO Review: 120GB, 250GB, 500GB, 750GB & 1TB Models Tested

Well except for maybe the 120's.

 

Sorry for the misunderstanding here. I said it feels like a HDD because I was comparing it with my corsair performance pro. Surely it's much more faster than my HDD, but the corsair SDD screams. I never seen any SSD as fast as it.

I never updated the firmware for once nor applied any settings. I wish there can make a 1TB of this corsair model

 

So...

Did the firmware update and Magician optimizations help at all?

 

I can't see any improvement. I'll just have to live with it. Thanks

 

Sorry to hear that. Maybe you can try to RMA if that is an option.
The only other suggestion would be to try and dig into some of the performance settings.
But as has been said before, it is TLC flash made for mainstream SSDs. With limitations.

 

First,
If you don't have Intel RST running now; try installing and see if you see an improvement (I have in every system I've been able to install it on - not in 'benchmarks' but in real world, day to day use).

See:
https://downloadcenter.intel.com/Detail_Desc.aspx?DwnldID=23060


I would recommend downloading the 'floppy' version (x64) and doing a manual install from device manager, . Running the SetupRST.exe file from the link above will install the driver and the tray icon for Intel RST and is not normally needed.


Secondly,
I would be Shrinking the partition of your C: drive to 70% or less (actually, with your responses taken into consideration; 50% or less). With the partition 'shrunk', I would then create a new partition on the 'unallocated' part of the drive, do a 'quick format' - leave the computer idle (do not touch it at all) for at least an hour or two, then, I would delete the partition you created and leave it 'unallocated'. What you have accomplished will be to TRIM the 'unallocated' capacity and hopefully this will give you a quantifiable speedup in your workflow.


Thirdly,
I would strongly consider (clean installing) the Windows 8.1 Pro O/S. Yes; it is that good.


If you need further details on the above, give a shout here. If you do the above and the situation doesn't change for the better: the moral of the story is: SSD's are not created equal (contrary to popular belief).


Good luck.

 

If it's new and running like a hard drive then something is wrong and no software is going to fix it. I would RMA the drive and get it replaced ASAP. There should be a completely noticeable difference between the two.

I have several Samsung 840 based on TLC and performance is not that bad. Even their 120GB which is severely gimped compared to the 240GB and faster versions. No current gen SSD should run anywhere near HDD performance.

Here's my Samsung 840 120GB SSD



Here's Samsung 1TB Evo (source ssdreview)

 

If you are within RMA period then open a ticket up and still have a couple days to work with this one before throwing in the towel for a refund.

^ tilleroftheearth and others, even the reviews and tests, have strongly suggested the need for that unallocated over-provision space >=30%. My personal experience with Samsung's is the same: without it even the OS can feel sluggish at times, this even having plenty of spare room on the OS partition spanning the entire drive space. And I can't imagine that it is filled to the brim already.
^ tilleroftheearth also mentioned the RST; I don't know/understand why you would not be running this.

^ HTWingNut may have missed your post correcting the HDD comparison.
Though he suggests it still shouldn't feel that bad; I can't comment since I haven't any experience with the TLC drives nor with the software you need to run.

Getting an HDD to MLC SSD improvement, then moving to TLC SSD isn't exactly an improvement, save for the space gain. What you're experiencing may be, that slower ability compounded by the amount of work to be processed. And if you added more work, then even worse symptoms.

 

Read those reviews closely. It's all about writes and result from testing after hammering the drive, not with regular use, allowing the drive time to GC on a regular basis. Read performance is barely affected. So yes for "most" users the OP is not necessary. I have many SSD's in service with no OP and years of use, TB's of writes and not a single performance dip. Why? Because they get moderate use like "most" users do.

Tiller hammers his drives all the time, and from that perspective, I say yes, OP can make a difference. It's like a race car driver suggesting NOS to get their cars running as fast as they can to a daily commuter. They're two different things. Or on the opposite end, a daily commuter recommending to the race car driver to keep under 60 mph to conserve gas mileage and longevity of the car. Sure each can help in certain conditions, but not for everyone.

Will it hurt to OP? Of course not. You just lose that much more storage space is all. And even as much as SSD prices have dropped, to most users that get a 128-256GB SSD, they can't afford to lose 30% of their storage capacity.

 

HTWingNut,

If I was the only one that noticed this newfound responsiveness by OP'ing, then I would agree with you (and all the others that point this out each time), but everyone that I have ever done this for has noticed this performance increase (not 'raw' performance increase: rather, the 'sustained' performance/responsiveness increase that OP'ing affords a system).

See:
http://forum.notebookreview.com/sol...ntel-x-25m-slow-write-speeds.html#post9405544


As I've mentioned many, many times before: the state of almost all SSD's at this point is that without 'losing' 30% of their capacity - there is almost no point in buying/spending the premium for having an SSD in you system, because the difference (responsiveness) is so pronounced.


Sure, you can discount my over-the-top usage case model - but you can't discount clients like the one I've linked above that is keeping an almost decade old system running with a 1st/2nd gen SSD with 'just enough' OP'ing used - along with a modern O/S like Windows 8 Pro x64.


The 'cost' of losing 30% capacity is simply the cost of doing business with an SSD at this time.




...

 

I have a 120G and i can assure you it is not like an HDD.

 

Point of correct here. When I say it feels like an HDD, I meant it feels slow as an HDD compared to my Corsair performance Pro under intensive usage executing SharePoint Applications in VMWare Workstation.

On normal usage on my laptop, I don't notice any performance difference. The EVO is fast as any SSD.

Regarding create partitions etc, I don't think it's necessary. I used up 95% of my 256GB Corsair, no partition etc and it's was still performing extremely well.

There is nothing wrong with my 1TB EVO. I'm just surprised with it's performance under heavy usage compared to my Corsair. That's all

 

My OEM 830 performs extremely well with or without an HDD as a working data drive.
Going from 0 to ~5% then to ~30 over-provision, there was a large noticeable difference in system responsiveness and smoothness in both steps.

Adding an 840 as system drive really upped the performance; even though it couldn't stay, due to technical restrictions it has to be relegated to working data drive position. Still the system out performed the previous configuration.

These are subjective comparisons. Never a benchmark ran; didn't need to, I could see and feel the improvements.

Have to purchase the right tool for the job, or reinforce the weaker introduction. Barring returning the drive and getting an MLC, make the best of what's there.

If the Corsair was never used with ~30 unallocated over-provision then there is no way to know if even that drive was performing to its potential.

I thought the purpose of the thread was to try and resolve an issue or improve a less than ideal situation?

 

That is not a great example. That user just did an hdparm secure erase and did a 20% OP based on your recommendation, but hadn't been running with OP all this time. It has nothing to do with having OP. If there were two systems running same drive one with 20% OP the other not then OK, there'd be enough data to determine that but there isn't. I don't think there's anything wrong with OP if a user chooses to do so. But your assessment is that the sky is falling if they don't and that's far from the case.

And the point that there is no point in buying an SSD if you don't OP is totally false. So all my systems, clients' systems, family's systems running SSD's might as well have HDD's? Hardly. You obviously haven't used a laptop hard drive lately, let alone a 5400RPM one that I have to use on a daily basis at my job. There's a massive improvement overall.

 

I don't know what you read about hdparm - but my client never did that...

 

If this were 2010, you might have a point.

 

Until all SSD controllers are tuned to give more consistency (sustained, over time) without gimmicks like faux SLC, etc., you might have a point too.


Until then; most SSD's respond to OP'ing just like the SSD controller should have been set up: for performance over time, not for high (when new/clean/one time/unused) 'scores' to boast about.