Crucial RealSSD C300 128GB Review
Single Vs. RAID - W/ Marvell - ICH10R & HPT RR 640
http://i1190.photobucket.com/albums/...ckmainM640.jpg
Special thanks to Kelly & Sam @ Crucial Memory for providing these samples for testing and review.
Crucial has been leading the way during the past year with the worlds fastest SATA SSD, the RealSSD C300. The C300 was the first, and still remains the only SATA 6GB/s solid state disk. These 2.5" Real SSD C300 drives offer blazing speeds for both desktop and mobile users, with many SATA 6GB/s enabled desktops reaching read speeds of 350+ MB/s with a single drive. There are many users who are often on the fence about choosing between a C300 SSD, or a Sandforce based SATAII SSD. Hopefully, with the benchmarks presented in this review your decision will be much easier to finalize. These drives are super fast, sleek looking, and very resilient to abuse by heavy benchmarking and massive file writes. I attempted to slow these drives down multiple times, using several methods, and was unable to make them even blink an eye at me (More on this later).
Crucial is part of the Lexar Media family of products and is owned by Micron Technology, a long time provider and manufacturer of Computer Dram Modules and Ram based products. They manufacture high-quality products in several memory based categories which include USB flash drives, all popular form factors of memory cards and card readers, and DRAM computer memory for PC and Mac systems. They are now also forging a path into solid state disk production, with the M225 series already EOL, and the blazing fast C300 currently being produced. There has also been a few hints rumored around the web in regards to the upcoming C400 series, plans which as of now include a SATA 3GB/s and a SATA 6Gb/s lineup. I'm anxious to see more information on the C400. Once Crucial releases these drives, I am positive they will overtake most any SSD on the market today.
Let's jump in and have a look at the C300 and see how it performs....
The complete lineup of C300 SSD drives natively support SATA 6Gb/s, and SATA 3Gb/s on legacy systems, with each capacity offering an available data transfer kit, making it easier for users to move existing hard drive data to the SSD.
The Crucial C300 line of solid state drives provide the following advanced features:
Crucial RealSSD C300 Product FlyerQuote:
Originally Posted by Crucial
C300 128GB CTFDDAC128MAG-1G1 2.5-inch SATA 6GB/s SSD Specifications:
Features:
- Capacity* (Unformatted) - 128GB
- NAND - Micron® MLC
- Controller - Marvell
- Firmware - Field Upgradeable
- RAID Support - Yes
- Interface - SATA
- Transfer Rate - 6Gb/sec (compatible 3Gb/sec)
- 2.5-inch SSD Power Supply - 5 V
- Certifications - FCC, CE, RoHS, UL
- Command Set - ATA-8 w/ TRIM
- Monitoring - Self-monitoring, analysis, and reporting technology (SMART) command set
Performance:
- Average Access Time - < .1 ms
- Sequential Read (up to) (Note 1) - 355MB/sec (SATA 6Gb/s) 265MB/sec (SATA 3Gb/s)
- Sequential Write (up to) - 140MB/sec (SATA 6Gb/s) 140MB/sec (SATA 3Gb/s)
- Random 4k READ (Note 2) - 60,000 IOPS
- Random 4k WRITE - 30,000 IOPS
Physical Dimensions:
- 2.5-inch SSD dimensions (L x W x H) - 100.45 x 69.85 x 9.50 mm
- 2.5-inch SSD weight - 74g
Environmental Factors:
- 2.5-inch Active Power - 1.7W Read, 3.1W Write
- Idle Power - 0.092W idle
- Operating Temperature - 0°C to +70°C
- Non-Operating Temperature - -55°C to 95°C
- Shock Resistance - 1500G / 1.0ms
- Vibration Resistance - 2-500Hz at 3.1G
Reliability:
- MTBF - 1.2 Million Hours
- Data Reliability - Built-in EDC/ECC
- Warranty - Limited 3 Year Warranty
- Read Endurance - Unlimited
Crucial RealSSD C300 Technical Specifications Datasheet
CTFDDAC128MAG-1G1 - 128GB, 2.5-inch Solid State Drive , Product Web page
This review will be covering the 128GB RealSSD C300 SSD. Testing will be done in several configurations, and several storage controllers will be utilized in order to show users how various setups may perform when using the Crucial C300 SATA 6GB/s SSD in multiple ways.
But first, we shall have a have a look at the product itself...
Product Images:
Packaging:
http://i1190.photobucket.com/albums/...PackFrontS.jpg http://i1190.photobucket.com/albums/...w/ReverseS.jpg
The front of the C300 packaging states the general drive info, size, type and speed specifications. The overall design is very similar to the actual SSD itself, coated in amber and gold waves, using "Vista-like sprites" down the right side of the package and SSD label. On the reverse you will find a general performance statement, in several languages, and a full list of specifications as well as a quick lesson in MB/GB correlation to usable drive size. Also on the reverse is the product sticker, indicating the model name and number, serial number, and UPC codes.
http://i1190.photobucket.com/albums/.../BoxEdgesS.jpg
A quick look at the sides, which match all around, you see the Crucial and the RealSSD logo - what a great name for an SSD.
http://i1190.photobucket.com/albums/...ackageNoBS.jpg http://i1190.photobucket.com/albums/...epackageS2.jpg
The SSD is snugly held inside the packaging with two foam inserts, one has a cut-out that the C300 fits firmly into for safe shipping. Inside you will also find a short installation guide and the warranty papers explaining the 3 year warranty.
A closer Look at the C300 itself:
http://i1190.photobucket.com/albums/...riveFrontS.jpg http://i1190.photobucket.com/albums/...DriveRearS.jpg
As previously mentioned, the SSD logo itself is very similar to the packaging. The label on the front of the C300 shows the full logo, a smaller portion of this image has been used on the outer package. The drive case itself is coated in a very sleek looking gun metal gray, with a very light amount of evenly distributed silver sparkles within it. It is not completely smooth, feeling similar to 1500 grit sandpaper, this is a nice touch that matches the colorization.
On the rear of the drive is the product sticker indicating the drive capacity, the model name and number, serial number, various certification logos, and the native drive speed. The shipped firmware revision is also noted on this label, as you can see this drive was shipped with firmware revision 0001, the initial version.
There are also 4 screw holes on the back, and two on each side of the drive for use when mounting the drive in a laptop caddy, or an adapter using side or bottom style mounts.
http://i1190.photobucket.com/albums/...w/SATATopS.jpg http://i1190.photobucket.com/albums/...ATABottomS.jpg
The drive has normal SATA Connections, a 7 pin data connector, and a 15 pin power connector. The connector has been placed in the proper location as well, so it does line up correctly with other drives when used with a 2.5" to 3.5" adapter.
What's inside you ask? (Please do mind your warranty)
http://i1190.photobucket.com/albums/...view/HawgS.jpg
Ohh no! The Crucial Hawg has reared it's head again, heavily eyeballing the 128GB of NAND memory in there! It is rumored that he really prefers flash memory over any other kind, and this would be a real treat for him.
http://i1190.photobucket.com/albums/...MainFrontS.jpg http://i1190.photobucket.com/albums/...everseSjpg.jpg
So, after wrestling the Hawg into submission, I was able to get a closer look inside the drive. The outer drive cover is secured in place with 4 screws, after removing these we can see the case is made up of three parts. The top and bottom covers are made of a rigid and very sturdy metal. Between them is a small black plastic mid-piece which allows the metal plates to be sealed together without damaging any of the internal parts or the case housing itself.
http://i1190.photobucket.com/albums/.../PCBFrontS.jpg http://i1190.photobucket.com/albums/...w/PCBRearS.jpg
Here you can gain a closer look at the components on both sides of the drive. There are sixteen 8GB ONFI 2.1 MLC Flash NAND chips, 8 on each side. The Marvell 88SS9174-BJP2 controller, which utilizes two ARM9 processor cores. Finally you can see the dram cache located on the reverse, this helps with write tracking, mapping out data tables, and aiding in the processing of data between the controller and the NAND channels.
A closer look at the main components:
Marvell 88SS9174-BJP2 Controller:
http://i1190.photobucket.com/albums/...oller9174S.jpg
IMFT (Intel & Micron) 34nm High-speed ONFI 2.1 MLC Flash NAND (16 x 64Gbit or 8GB for this SSD)
http://i1190.photobucket.com/albums/...view/NandS.jpg
Micron 128MB DDR3 1333Mhz OED12 D9LGQ Cache Buffer
http://i1190.photobucket.com/albums/...28MBCacheS.jpg
Highpoint RocketRAID Controller:
http://i1190.photobucket.com/albums/...R640FrontS.jpg http://i1190.photobucket.com/albums/...40ReverseS.jpg
http://i1190.photobucket.com/albums/...view/9128S.jpg
Finally, here we have the Highpoint RocketRAID 640 controller used for the review. This controller utilizes the Marvell 88SE9128-NAA2 SATA 6GB/s controller chip, and a PLX Bridge chip. These are the same controllers that many motherboards currently use for SATA 6GB/s connections, however, motherboards are only equipped with one Marvell controller thus only being able to offer 2 connection ports.
Testing:
Testing Setup:
Gigabyte GA-X58A-UD7
Intel i7 980X ES EX Hexa-core Processor
6GB Kit Crucial Ballistix Tracer (Blue) DDR3 1600Mhz Memory
Crucial RealSSD C300 SATA 6GB/s SSD 128GB x 2
Western Digital WD1001FALS 1TB SATAII Hard Drive (Short Stroked to 238GB)
Gigabyte 5870 SOC (GV-R587SO-1GD) Graphics Card
Ultra X3 ULT40312 850W Power Supply
Highpoint RocketRAID 640 SATA RAID Controller
Mushkin 8GB Mulholland Drive 8GB USB Drive (32MB/s Read / 17MB/s Write)
AS SSD Benchmark V. 1.5.3784.37609
http://i1190.photobucket.com/albums/...view/ASSSD.jpg
This benchmark has quickly become a favorite benchmark for many SSD users. The program now shows current firmware version for the drive being tested, what disk driver is in use, and what disk offset is being used (And notes if that offset is good or bad for SSD's).
The benchmark tests random data in 1GB file sizes, offering sequential, random 4K, 4K 64-threaded, access times, and also combines those results into an overall total.
http://www.alex-is.de/PHP/fusion/downloads.php?cat_id=4
ATTO V. 2.34
http://i1190.photobucket.com/albums/...eview/ATTO.jpg
ATTO has been a major disk testing utility for many users, for quite some time, offering the ability to test performance using various transfer sizes and test lengths for reads and writes.
It also offers several options to customize your benchmark testing including queue depth, overlapped I/O, a comparison mode and the option to run the test continuously.
http://www.attotech.com/products/pro...Disk_Benchmark
IOmeter V. 2008-06-22 RC2
http://i1190.photobucket.com/albums/...ew/IOMeter.jpg
Iometer is an I/O subsystem measurement and characterization tool, it is often used as a benchmark tool and is easily configured to replicate the behavior of many popular applications. We will be using this tool to measure IOPs. This program was created by Intel, they are no longer updating or offering support for it, and it is now fully open source. You may download my test configuration files Here if you like, if used for comparison you may need to adjust CPU target cores used, and the target test drive letter.
http://sourceforge.net/projects/iometer/
CrystalDiskMark V. 3.0.0f
http://i1190.photobucket.com/albums/...review/CDM.jpg
CrystalDiskMark is another staple in the disk benchmarking world, offering measurement of sequential read and writes speeds, random 512KB, 4KB, 4KB (Queue Depth=32).
Users may also select the method of how data is tested, random, 0Fill, or 1Fill - in the testing done for this review only the default testing method was used - Random Data.
http://crystalmark.info/software/Cry...k/index-e.html
PCMark Vantage Professional x64 V. 1.0.2.0
http://i1190.photobucket.com/albums/...iew/PCMark.jpg
The HDD Suite uses a variety of real world hard drive tests to simulate various tasks, such as virus scanning, gaming, application loading, importing files, and media editing.Quote:
PCMark Vantage is a PC benchmark suite designed for Windows Vista offering one-click simplicity for casual users and detailed, professional grade testing for industry, press and enthusiasts.
A PCMark score is a measure of your computer’s performance across a variety of common tasks such as viewing and editing photos, video, music and other media, gaming, communications, productivity and security.
HDD Test Features:
* 6 HDD Tests total
* HDD Tests use new 4K block size
* All HDD Tests use RankDisk using Intel’s IPEAK Storage Performance Toolkit
* WinTrace is used to record the HDD tracesAnalyzeTrace is used to analyze precisely what the traces are doing
* RankDisk: a Futuremark modified version that it is non-destructive and runs in a temporary sandbox file.
HDD Suite includes the following tests:
HDD 1 - Windows Defender
HDD 2 - HDD: Gaming
HDD 3 - HDD: Importing pictures
HDD 4 - HDD: Windows Vista start-up
HDD 5 - HDD: video editing
HDD 6 - HDD: Media Center
HDD 7 - HDD: Adding music to Windows Media Player
HDD 8 - HDD: Application loading
http://www.futuremark.com/benchmarks.../introduction/
Winbootinfo V. 1.02 x64
http://i1190.photobucket.com/albums/...inbootinfo.jpg
WinBootInfo is an advanced windows boot analyzer, it logs drivers and applications loaded during system boot, measures Windows boot times, and it also records CPU and I/O activity during the booting process. It also logs many other things as well, but for this review I will only be using the overall total boot time result.
http://www.clockmod.com/index.php?/p...nbootinfo_1.0/
Bootracer V. 2.3
http://i1190.photobucket.com/albums/...BootRacer2.jpg
BootRacer allows you to test your Windows booting speed, offering timed results of "Time to Logon", "Time to Desktop", and "Best Desktop Time".
Results used for this review are the "Time to Logon" result times.
http://www.greatis.com/bootracer/
VBS Reboot Script (ReBoot Time) by mrkristins
http://i1190.photobucket.com/albums/...Bootuptime.png
This is a simple visual basic script, written by a Deviantart user.
It tests full rebooting time, which includes shutting down for the reboot, and any POST or Pre-Operating system screens, ending the timer at the desktop after the reboot has taken place.
Boot time in Vista,7 by ~mrkristins on deviantART
Microsoft Office 2007
The testing I completed using Office 2007 is two-fold. One test was opening a large 183MB Excel test file that included many large images and random data.
The second test measured the time it took to install from start to finish, with the program being installed from the desktop.
Adobe Photoshop CS3
Very similar to the Office 2007 tests, two tests were used for real world testing using Photoshop CS3.
The first test was opening a large 1GB image, with many large images within, and several layers of high resolution variances between layers. The second test was install time, from start to finish, from the desktop.
Windows 7 Pro x64 Install Time (Via USB)
A custom created ISO was used for this test using RT Seven Lite 1.4.0, and the installs were completed using a Mushkin 8GB Mulholland USB drive. Nothing affecting the install time was changed, except making the install unattended. No programs, services, or files were removed, and for certain installs controller drivers were loaded. The unattended installer was set to install to the first available partition, which was pre-aligned before the install process or timer began. The install was universally the same, except for controller drivers changed for certain installs, the unattended XML and a sample INF file used for these tests can be downloaded Here for anyone that would like to have a look.
http://www.rt7lite.com/
Results:
AS SSD Benchmark V. 1.5.3784.37609
http://i1190.photobucket.com/albums/...-128x2-R0S.jpg http://i1190.photobucket.com/albums/...ntelRAID0S.jpg http://i1190.photobucket.com/albums/...le-MSAHCIS.jpg http://i1190.photobucket.com/albums/...gleSATAIIS.jpg http://i1190.photobucket.com/albums/.../WD-ASSSDS.jpg
Sequential Maximum
http://i1190.photobucket.com/albums/.../ASSSDSeqW.png
Results here scale at a nice even incline, the HPT 640 can really push out the high speed reads with these drives. It's really too bad that motherboard designers did not use two of these controllers on 6GB/s equipped RAID ready motherboards, if they would have, we would see similar results using on-board Marvell RAID. Hopefully it wont be too much longer until Intel releases a native 6GB/s Intel controller!
4k Read/Write
http://i1190.photobucket.com/albums/...s/ASSSD4KW.png
Notice the SATA 3 with a single drive is right on par with the HPT RAID. 4K operations on the Intel controller are very well tuned, which is even more evident when looking at the 4K writes on the Intel RAID 0 setup.
4k 64-Thread Read Write
http://i1190.photobucket.com/albums/...K64ThreadW.png
Here things are back in line a bit more across the board, but the Intel ICH10R is moving ahead a bit in write operations. Interestingly The 3Gb/s and 6GB/s results are almost the same for this test using a single drive.
Moving ahead the Intel controller is going to quickly become a winning theme for this review, that may be surprising to hear since it's only SATA 3GB/s, but you will soon see it's true in many areas.
Latency
http://i1190.photobucket.com/albums/...DLatency2W.png
It's surprising to see 3Gb/s and 6GB/s latency being so close, the drives are so fast though in reality these times are not going to affect anyone. The WD 1TB results were left out of this graph, as the results did not work out well when graphed with the C300 times.
Again, we have a clear winner when it comes to disk access time.... You guessed it, Intel.
ATTO V. 2.34
http://i1190.photobucket.com/albums/...-128x2-R0S.jpg http://i1190.photobucket.com/albums/...ntelRAID0S.jpg http://i1190.photobucket.com/albums/...le-MSAHCIS.jpg http://i1190.photobucket.com/albums/...gleSATAIIS.jpg http://i1190.photobucket.com/albums/...oWDSingleS.jpg
4K Read/Write
http://i1190.photobucket.com/albums/...hs/ATTO4KW.png
Another test where Intel RAID 0 is leaving everyone else in the dust. All other configurations, WD aside, are pretty much on par here.
Maximum Read/Write
http://i1190.photobucket.com/albums/...s/ATTOMAXW.png
Things are getting back on a more even keel here, with HPT RAID clearly taking the lead at just above 700MB/s.
CrystalDiskMark V. 3.0.0f
http://i1190.photobucket.com/albums/.../CDMHPTR0S.jpg http://i1190.photobucket.com/albums/...ntelRAID0S.jpg http://i1190.photobucket.com/albums/...le-MSAHCIS.jpg http://i1190.photobucket.com/albums/...gleSATAIIS.jpg http://i1190.photobucket.com/albums/...MWDSingleS.jpg
Sequential Maximum
http://i1190.photobucket.com/albums/...phs/CDMSeq.png
HPT RAID takes the lead from Intel here in reads, but only by a little over 100MB/s
4K Read/Write
http://i1190.photobucket.com/albums/...aphs/CDM4K.png
Intel RAID is back at the top of things again, putting a heavy smackdown on everyone else in the 4K department.
4K 32-Thread Read/Write
http://i1190.photobucket.com/albums/...hs/CDM4K32.png
Que-Depth of 32 in these 4K test show things a bit more even, with HPT RAID and Intel RAID very close to one another, and the same applies to single disks on the 3GB/s and 6GB/s controllers.
PCMark Vantage Professional x64 V. 1.0.2.0
http://i1190.photobucket.com/albums/...arkHPTR02S.jpg http://i1190.photobucket.com/albums/...ntelRAID0S.jpg http://i1190.photobucket.com/albums/...ellSingleS.jpg http://i1190.photobucket.com/albums/...ngleIntelS.jpg http://i1190.photobucket.com/albums/.../PCMarkWDS.jpg
HDD Test Total Score
http://i1190.photobucket.com/albums/...MarkTotalW.png
Yet again, the Intel ICH10R controller is tough to beat, but the HPT RAID card doesn't fall far behind.
HDD Test 1-4
http://i1190.photobucket.com/albums/...PCMark1-4W.png
The Highpoint RocketRAID controller is the clear winner of this round, leaping well ahead of the rest. The SATA 3GB/s and 6GB/s results come in just about equal in these tests, except for the HDD test 1.
HDD Test 5-8
http://i1190.photobucket.com/albums/...PCMark5-8W.png
The Intel ICH10R controller takes the cake here, with the HDD test 6 results being high enough to throw the graph out of whack so I had to manually write those results in. The test was re-ran 3 times, on different installs as well, and the result for this test (Media Center) were always massively higher than the others.
Winbootinfo V. 1.02
http://i1190.photobucket.com/albums/...bootinfoSS.jpg http://i1190.photobucket.com/albums/...Intel-R0SS.jpg http://i1190.photobucket.com/albums/...eSATAIIISS.jpg http://i1190.photobucket.com/albums/...leSATAIISS.jpg http://i1190.photobucket.com/albums/...WDSingleSS.jpg
http://i1190.photobucket.com/albums/...nbootinfoW.png
SATA 6GB/s does have it's advantages, one of which is quick shutdown and reboot times. Reboots are super quick when using a single drive on the 6GB/s controller.
Bootracer V. 2.3
http://i1190.photobucket.com/albums/...BootracerW.png
C300 running at 6GB/s using the Marvell controller also topples the competition with Bootracer as well.
VBS Reboot Script (ReBoot Time) by mrkristins
http://i1190.photobucket.com/albums/...ebootTimeW.png
Reboot times are a bit more even between 3GB/s and 6GB/s single drives during these tests, which include the BIOS POST screen times and everything else during the reboot. The Highpoint RocketRAID 640 controller falls behind by a great deal here, due to it's slow drive scan and recognition process at start-up, normally this process is about an additional 30 seconds added to the normal boot time.
C300 SSD IOP/s:
IOmeter V. 2008-06-22 RC2 - 4K Aligned
http://i1190.photobucket.com/albums/...AID0-ReadS.jpg http://i1190.photobucket.com/albums/...AID0-ReadS.jpg http://i1190.photobucket.com/albums/...AIII-ReadS.jpg http://i1190.photobucket.com/albums/...TAII-ReadS.jpg http://i1190.photobucket.com/albums/...OPsWDReadS.jpg
http://i1190.photobucket.com/albums/...ID0-WriteS.jpg http://i1190.photobucket.com/albums/...ID0-WriteS.jpg http://i1190.photobucket.com/albums/...III-writeS.jpg http://i1190.photobucket.com/albums/...AII-writeS.jpg http://i1190.photobucket.com/albums/...PsWDWriteS.jpg
4K Read/Write - IOP/s
http://i1190.photobucket.com/albums/...IOMeter4KW.png
The Intel RAID setup is the clear leader here, with Highpoint coming in a close second. Similar to what's seen in the ATTO 4K results, the single C300 in 3GB/s or 6GB/s are fairly close in performance when it comes to 4K operations.
AS SSD Benchmark V. 1.5.3784.37609 - 4k 64-Thread
http://i1190.photobucket.com/albums/...2-R0-IOPsS.jpg http://i1190.photobucket.com/albums/...RAID0IOPsS.jpg http://i1190.photobucket.com/albums/...leSATAIIIS.jpg http://i1190.photobucket.com/albums/...gleSATAIIS.jpg http://i1190.photobucket.com/albums/...ASSSDIOPsS.jpg
4K 64-Thread Read/Write - IOP/s
http://i1190.photobucket.com/albums/...eadedIOPSW.png
The RocketRAID is able to process a bit more 4K 64-threaded information than the ICH10R, but they do score very close. Again the 3GB/s and 6GB/s single drive results are fairly close here, but the Marvell controller is able to process read operations faster.
Real World Testing:
Microsoft Office 2007
183MB Excel Test File Open
http://i1190.photobucket.com/albums/...7FileOpenW.png
The Intel RAID setup was able to open the 183MB test file the fastest here, but only one second behind a single C300 in SATAII mode.
Install Times:
http://i1190.photobucket.com/albums/...07InstallW.png
A single C300 is the clear winner here, scoring above both RAID configurations in either 3GB/s or 6GB/s modes.
Adobe Photoshop CS3
Install Times
http://i1190.photobucket.com/albums/...S3InstallW.png
RAID setups were able to install the full Photoshop CS3 suite much quicker than single disks, with both arrays finishing the install within one second of each other. Single drives in either mode were not far behind the RAID arrays, and with less than 5 seconds between each of the RAID install times.
1GB Test File Open
http://i1190.photobucket.com/albums/...3FileOpenW.png
Here we see the RAID setups are able to access files and open them much quicker for Photoshop, with both arrays opening the 1GB test file within two seconds of each other. Single drives in either mode were not far behind the RAID arrays, and with only 3 seconds between each of their file opening times.
Windows 7 Pro x64 Install Time (Via USB)
http://i1190.photobucket.com/albums/...stallTimeW.png
These results simply amazed me! This was the first time I had ever installed an operating system from a USB Stick, and I was simply floored by how quickly the install process finished each time as I stared at the clean desktop of windows 7. Crucial's C300 SSD no doubt played a huge part in how fast this happens, but the USB install is clearly much quicker than an average 15 minute install from a DVD.
Install times are very close, with only 40 seconds difference between the longest and shortest installs on the SSD, but again Intel has taken the lead at only 7 minutes and 13 seconds to complete the install.
Additional Tests:
During the extended period I worked on this review I completed several other tests that are not a part of the results previously covered. These tests are a similar comparison to the review results above, testing the C300 across several controllers, but with some minor differences. The results are only quick testing examples, but I felt I should mention them here and provide the results, because they do include the use of Marvell Drivers on the 9128 controller - which the above results do not include and I will mention why shortly.
Here are the results from that testing:
Intel ICH10R
Single (RAID Non-Member)
http://i1190.photobucket.com/albums/...ial-ForumS.jpg
RAID 0
http://i1190.photobucket.com/albums/...ial-ForumS.jpg
Marvell 9128 On-board
Single - Marvell mv91xx Driver
http://i1190.photobucket.com/albums/...ingleC300S.jpg
Single - MSAHCI Driver
http://i1190.photobucket.com/albums/...ial-ForumS.jpg
RAID 0 - Marvell mv91xx Driver
http://i1190.photobucket.com/albums/...9128-RAIDS.png
Highpoint RocketRAID 640 (Marvell 9128 Controller x 2 + PLX)
Single
http://i1190.photobucket.com/albums/...ial-ForumS.png
RAID 0
http://i1190.photobucket.com/albums/...ial-ForumS.png
An ongoing discussion of these tests can be found Here on the Crucial forums
I also ran a quick test of IDE mode Vs. AHCI mode, using the Marvell 9128 Controller. A few Marvell 9128 users were interested in how much of a negative affect this would have, as was I, so I ran a quick test.
IDE Mode
http://i1190.photobucket.com/albums/...9128-IDE-S.jpg
AHCI Mode
http://i1190.photobucket.com/albums/...8-MSAHCI-S.jpg
As you can see in the results, overall sequential read and writes are negatively affected, and more importantly the 4K threaded results show a large drop in performance when using IDE mode. With these results in mind, be sure to inform anyone you see running in IDE mode that it can have a very adverse effect on performance, and it would be wise to use AHCI mode when at all possible. I assume this would probably have a similar effect, possibly larger, on the Intel controller as well as it does outperform the Marvell Controller in the 4k/4K-Threaded areas.
TRIM and Garbage Collection:
I attempted several times to slow the C300 SSD's down, in order to be able to test how quickly TRIM helped the drive speeds recover, but was unable to slow the drives at all. I tried several methods, even disabling TRIM and filling the entire drive with random data, then deleting this data leaving the blocks dirty. This had no effect on speeds, the C300 remained resilient and immune to my torture. This clearly indicates that GC does a great job in keeping the drives clean, and does its work in a very quick and efficient manner.
Here are a few of my results from these tests. Although the end result was not what I had intended to obtain, I believe they should be a part of this review so that you may see how well the C300 performed under these demanding conditions.
Intel ICH10R controller:
Drive Full, no signs of speeds slowing
http://i1190.photobucket.com/albums/...downIntelS.jpg
TRIM Disabled, drive filled then emptied, then tested
http://i1190.photobucket.com/albums/...hAllZonesS.jpg
TRIM Enabled, drive emptied and manually trimmed, then tested
http://i1190.photobucket.com/albums/...es-TRIMEDS.jpg
Marvell 9128 controller:
Drive Full, no signs of speeds slowing
http://i1190.photobucket.com/albums/...lSlowTestS.jpg
TRIM Disabled, drive filled then emptied, then tested
http://i1190.photobucket.com/albums/...dSlowTestS.jpg
TRIM Enabled, drive emptied and manually trimmed, then tested
http://i1190.photobucket.com/albums/...dSlowTestS.jpg
I also tested all controllers for the ability to TRIM, or not, using Bill Jameson's "Force TRIM" Program, and found that ONLY the Intel or MSAHCI drivers passed TRIM along to the drives. The HPT does not in any configuration. The Marvell 9128 controller also did not pass along the TRIM command to the drives when using MSAHCI driver, nor with the 1042 Marvell driver either. To further clarify the above, TRIM is passed to drives on the Intel controller in IDE mode as well as in AHCI/RAID (Non-Member) modes.
A fellow helper at the Crucial forum, targetbsp, has created a simple tool that forces TRIM on SSD's as well, you may download that tool Here. It must be used on a TRIM capable configuration, otherwise it will not function properly. This means your operating system must be Windows 7, your solid state drive must support TRIM, the controller and the disk driver used must also support TRIM, and of course TRIM must be enabled within the operating system.
Crucial has done a great job in programming the latest firmware's garbage collection, I would confidently recommend a RAID setup with the C300 drives anytime. You may also view a discussion on this testing Here, as I discussed this with others in order to see how the C300 had been slowed down in the past on other reviews. I can only come to the conclusion that either the updated firmware has now corrected any past issues, or testing needs to be performed in a different manner in order to slow these drives. I can't see it being possible under normal circumstances, considering that I tried several ways in an atypical setup without being able to produce any slowed speeds at all.
Final Thoughts:
This review would not be complete without my thoughts on the Marvell 9128 6Gb/s controller. You would think the C300 being controlled by a Marvell controller would be a match made in heaven, with an on-board Marvell 9128 SATA 6GB/s controller, but this is simply not the case. This has nothing at all to do with the C300, only the Marvell 9128 controller itself, and it's limitations. The current Marvell 9128 controller used on many motherboards is unable to handle the heavy demands that SSD's can provide, pairing that with sub-par drivers and firmware and things begin to get ugly.
As a follow up, edited into this review after publishing, a few users have asked or messaged me about results RAID on the Marvell Controller. Previously I did not plan to include any results, but with the requests I've received I decided to add a few.
The following results show two tests of a RAID setup on the Marvell controller. As you can see results are dismal with either method results are not up to par with what you would expect from a RAID array, both are well below speeds of what you would get doubling a single drives speeds, and sequential read speeds aside both results are in the SATA 3GB/s range. Also please note, as mentioned above when using the Marvell Controller and Marvell driver in the short comparison results, drives are dropped from the array often. The tests using the Marvell driver had to be ran several times in order to get a finished result with all three benchmarks on the screen at the same time, without the array dropping out causing the system to need rebooted in order to continue further tests. Needless to say, RAID on the Marvell controller is not ideal or suggested with solid state drives.
The following configuration was used for the Marvell Controller in these RAID tests:
Marvell RAID BIOS Version 1.0.0.1027
Marvell Firmware Version 2.1.0.1504
Marvell Driver - 1.0.0.1045
MRU Driver - 4.1.1712
http://i1190.photobucket.com/albums/...91xx1045S2.jpg
MSAHCI Driver - No Marvell driver loaded
http://i1190.photobucket.com/albums/...IDMSAHCIS2.jpg
Gigabyte has recently changed their 6GB/s motherboard specifications to include the following notation for the 9128 controller "SATA3 SSDs are not recommended for use in RAID 0 mode on Marvell SE9128 ports"
GIGABYTE USB 3.0 Motherboards
During my testing and use of these SSD's I did several tests on the Marvell controller, and these issues have spanned across several firmware and driver updates, with no improvement in sight. When using SSD's in a RAID configuration the controller cannot handle the data throughput and drops connection to one or more of the drives. I am also able to cause this with a single solid state disk when putting the controller under mild stress such as a benchmark with AS SSD, but not nearly as easily nor as often. These issues are completely replicable as well, so it's not the drives or motherboards being tested. The same applies to the Highpoint RocketRAID 640 controller, although if using only two SSD's one can span them across both 9128 controllers on the Highpoint card thus effectively cutting the stress level on the controller in half, which cannot be done on current motherboards as they only utilize a single 9128 controller.
So... those are the reasons why you do not see Marvell RAID results as part of this review. While single solid state drives on the Marvell controller are much less problematic, there can be times when a single SSD is able to cause these issues when the controller is put under extended periods of heavy access.
Current 6GB/s solutions are unable to offer the processing power needed to realize the speeds the C300 drives can deliver, while current controllers are still fast in sequential throughput, they lack the ability to deliver what these drives are actually capable of offering. I feel these drives have much more life and surprise left in them, there is absolutely some hidden surprises in the 4K department just waiting for Intel to release it's native 6GB/s on-board controller. Once that happens I think every C300 owner is in for a very big, and very productive surprise.
The Crucial RealSSD C300 is a superb solid state disk, it's the world's first and still only SATA 6GB/s drive. If you are contemplating purchasing a Crucial C300 you can rest assured that it will deliver, without a doubt, now and even more so in the future! So if you are thinking of buying a Crucial C300 SSD, it would be a wise choice as they will only get better with time as drive controllers improve.
Pro's:
• Massive 6GB/s speeds, and great 3GB/s speeds as well
• Great Garbage Collection programming
• Stylish and durable case design
• 3 year warranty
Con's:
• None - Other than lack of Intel 6GB/s controllers as of yet.
The Crucial C300 128GB could be purchased from Newegg.com for $269.99 at the time of this review 11/5/10.
During my testing for this review I also made a few files to help C300 owners, feel free to link these around as needed.
Crucial C300 Firmware Update Guide For Gigabyte Motherboard Users:
http://www.mediafire.com/?nmth3zzel53
Crucial C300 Firmware Update File - Update Via USB (For C300 128GB/256GB Owners):
http://www.mediafire.com/?ux8zfz6ftpmoyy8
Again, I'd like to give a huge thanks to Kelly & Sam @ Crucial Memory for providing these samples, it's been great testing out the Crucial C300!!
A big thanks to GIGABYTE as well, for providing other hardware used in this review!!!
Disclaimer: I am not a professional hard drive or SSD reviewer, this IS my "first day" :D So if you do not agree with any of my testing methods please kindly let me know what would work better for you, or things I might change, and I will keep any and all comments in mind for future reviews. Thank you for leaving only constructive comments, please.
