Crucial M4 128GB 2.5-Inch SATA SSD vs Intel 330 2.5 inch 120GB SATA3

Hi everyone,

They are both at ~$110 and are the right amount of storage space I need. Which one is better?

Crucial M4
Amazon.com: Crucial m4 128GB 2.5-Inch Solid State Drive SATA 6Gb/s CT128M4SSD2: Electronics

Intel 330
Intel 330 Series SSDSC2CT120A3K5 2 5 inch 120GB SATA3 Solid State Drive MLC 735858246996 | eBay


Thanks for your time/help

 

Can't say much about the intel 330 because I have never used intel drives, but I've bought around 10 crucial M4 drives so far in the last 1-2 years, 1 for my laptop, 3 for my desktops, and quite a few for my brother and co-workers, never had a single hitch with them, highly recommend it.

 

I had the 330 in my old laptop but will be getting a mSATA version of the M4 for my new laptop. I believe the Crucial scores higher on Crystal Disk Mark.

 

For that capacity (which I do not recommend, btw) the M4 is the easy winner.

See:
Round-Up: 10 mSATA SSDs From Adata, Crucial, Mushkin, And OCZ : mSATA: Solid-State Responsiveness On A Tiny Card

While the above quote is about the Crucial M4 mSATA versions - the 2.5" SATA versions are comparable too, but I'll stress once again that 128GB (nominal) capacity for SSD's are not enough in 2013 - especially if you want the highest performance constantly and consistently.

Good luck.

 

I would go for the 128GB M4, due to the price and performance.

Now if you were lucky like me and got a deal for 60$ for the 330 Intel drive as I did you would purchase the Intel too, but the M4 is definitely the winner when in the same price range, without ANY doubt

 

Does the capacity really make that much of a performance difference?

I usually use under 50 gigs and thus why I decided on a 128 gb one....The 256 gb one is almost twice the price at $190. I have most of my media files on my desktop which has over 1 TB....and I will buy a casing for the 750 gb Harddrive I will be replacing on my laptop with the new SSD

 

128GB M4:
Random Read Performance 40K IOPS
Random Write Performance 35K IOPS
Sequential Read Performance Up to 415MB/s
Sequential Write Performance Up to 175MB/s

256GB:
Random Read Performance 40K IOPS
Random Write Performance 50K IOPS
Sequential Read Performance Up to 415MB/s
Sequential Write Performance Up to 260MB/s

 

I assume loading OS and other programs/games is "Read Performance"? When would one typically use "write"?

 

I would assume those you mentioned are read heavy. Write heavy applications could include video editing/capturing, photoshop etc. in the context of consumers.

 

Capacity makes a difference in performance in three ways:

First, the controllers channels are fully utilized and the nand chips on each channel are optimally interleaved to provide the highest theoretical performance possible. While not all SSD's behave like this (the M4 128GB and 256GB models being the best current examples) most do - and while the smaller ones are sometimes faster in benchmarks while in a new state - the situation changes dramatically when they reach 'steady-state' use and the theoretical performance plummets to less than HDD performance (in some aspects, for eg. sequential writes...).

Keep in mind here that each nand chip might only give 40MB/s or less maximum performance - to get to the point of saturating the current SATA3 interface standard, you need a lot of these chips operating in parallel to do so.


Second, the sheer capacity of the drive helps performance by allowing us to have as much free space as possible - this helps by allowing the controller to defer (not all do this elegantly) Garbage Collection (GC) and TRIM commands until after OUR storage subsystem demands/requests are processed first. This 'deferring' is not an active decision of the controller - the Firmware is programmed so that free nand chips get used first (this is the wear-leveling part of the algorithm) to incur the least Write Amplification (WA) and give the highest performance to the user as they need/request it. While leaving free space is beneficial, it is not as effective as leaving a part of the capacity as 'unallocated' right from a clean install would allow (and 'force' a user to honor).



With 'extra' capacity - along with all the benefits of the two points above - the biggest benefit is the ability to leave 30% or more capacity as 'unallocated'. Real world performance takes a huge (to me) jump when 30% or more capacity is left unused from day one of ownership/use of the SSD.

This performance is not an increase to the baseline performance of the drive - it is a method to ensure more consistent performance from the SSD. Not only is TRIM and GC performed with the least amount of WA - it allows the SSD to service the user's requests first and foremost at the highest possible speed - just like leaving free space, but with the added benefit of not needing to track/check if the 'spare' nand chips have user data on them first.

The consistency that is achieved with some 'unallocated' capacity is not only desirable - it is a requirement in current SSD's to enjoy the fastest storage subsystem possible - no matter how that that storage subsystem is used (or abused). While everyone knows that SSD's are 100's of times faster, on average, than a HDD (in certain metrics) what they fail to educate themselves on is that SSD's can be just as slow (or slower) than a mechanical HDD when the maximum latency is looked at (HDD's have no such issues - their performance is dictated (mostly) by the physics of their spinning platters and the speed/accuracy of their read/write heads.

Over-provisioning via leaving 'unallocated' capacity in SSD's allows the performance drops (maximum latency) to be kept to a minimum while also giving all the benefits of less WA and faster, more efficient TRIM and GC routines.


All of the above taken together in real world use takes an SSD from laughable (in my workloads) to indispensable - yes; from below HDD performance (overall) - to enough above HDD performance to make the $$$ and time involved to change/upgrade to an SSD worth it (and then some).


Look at the following graph (at the bottom of the page/link) to see how much more performance is unleashed by over-provisioning that a higher capacity SSD allows:

See:
AnandTech - Plextor Updates The Firmware on M5 Pro: Promises Increased Performance, We Test It


Click the Samsung 840 PRO and the 25% over-provisioned results to see an almost 6x improvement in one aspect of performance.

(And note how the Plextor M5 PRO is one of the worst SSD's as far as consistency is concerned - most of it's 'scores' are along the Zero axis line).




The following link shows the maximum performance of the M4's at different capacities:

See:
AnandTech - The Crucial m4 (Micron C400) SSD Review



The following link shows the Intel 330 Series drives at different capacities (note the huge difference in performance with incompressible data):

See:
AnandTech - The Intel SSD 330 Review (60GB, 120GB, 180GB)



To complete this post: the Crucial M4 512GB SSD is the only 'large' SSD I would be considering seriously today - it offers slightly higher performance than the 256GB model - but more overall (real world) performance than all other ~1/2TB SSD's we can buy now. In fact, all SandForce based SSD's (including Intel's offerings) take a performance dive when going from 240GB to 480GB capacity (and that just isn't right!).


Hope this answers some of your questions?

 

More than answer, thanks for the very comprehensive answer.

As I have no plans to get the M4 512 gb one (way outside my budget), I will probably get the 128gb one since it is almost as good as the 256 gb one it seems.

2 quick questions:

1. How do I set "Over-provisioning" when my SDD arrives so that I can get peak performance?

2. Just read something interesting - Apparently, switching from "PATA to AHCI" increases SSD speed? How do I do this and how do I check which one my system uses. To be honest, I have no idea what it means.

 

Setting a system to AHCI mode is usually done in the BIOS - if there is a general SATA/storage option for AHCI/'Enhanced' or 'Compatible' you don't want 'Compatible'.

To over-provision your new SSD the best way is to do it will clean installing Windows (after you have confirmed the above 'switch' in the BIOS).

Go to the Advanced Install Disk Options and partition the drive (in MB's) to around 30% less than the maximum capacity.

Note: I do not count the 'nominal' capacity in this calculation - for our specific SSD (M4 128GB model) I calculate it as so:

128 x 1 Billion = 128,000,000,000 (This is how the manufacturer sells it as 128GB).

Now divide that by 1024 and again by 1024 and again by 1024 (we are getting to a 'real world' capacity by converting to GiB's).

and the result is ~119GiB. Now, from this value deduct 30% = ~83GiB total usable capacity.


If you really do use less than 50GB (including Win7/Win8 x64 PLUS Programs installation size) then I agree this is your 'perfectly' sized drive. If however, you use 50GB by yourself and Windows takes it's normal (~25GB) plus whatever programs you install - then you can see how this quickly becomes too small.


The fastest an SSD will be if you over-provision it at 50% or more - but that I usually (but not only) do for my desktop workstations with multiple SSD's (4+) and HDD's (6+) - the most aggressive OP I use is 100GB out of a 250GB (Intel 510 Series) SSD.


It is much better to save a little longer to get the capacity that is most useful for most users: 240/256GB or higher. But getting a 'throw-away' drive (128GB or smaller) and forgetting about all the above will still give higher performance than a HDD for 'normal' use (at least for a short while) - while you save for the 'best/better' option down the road (which may even mean a new drive/technology to consider when that time comes).


Good luck.

 

Will always making sure to leave ~40 gigs free do the same thing as Over-provisioning in terms of keeping performance optimal?

 

No, not even close in my experience - with 'unallocated' space - the SSD controller knows that there is no data on those specific nand chips. With 'free space' the nand chips have to be tracked and processed each time the storage device is accessed (because the controller doesn't know if the data on nand chips is 'valid' or the data is 'throwaway' and if the space is available - data (even temporary data) can be written anywhere).

The way Windows (Vista/7/8/xxx...) works with my workflows - it needs at least 25GB free space at all times to not 'complain' too loudly - while 30%+ over-provisioning via 'unallocated' space is needed for the speed and consistency benefits.


(By 'complain' I mean the little balloon messages that space is too low and also how a system feels for my normal use if the free space is much lower).

 

If I Over-provisioning for say 40 gb...can I use up the remaining 88 gb with no problems or do I need to still leave 5-10 gb empty for optimal performance? Sorry for the barrage of questions.....and THANKS!

 

Yeah, still need at least 10GB free space (in addition to OP'ing).