Jump to content

Corsair ForceGT RAID0 vs Samsung 840 Pro


Pic0o

Recommended Posts

Welcome to another 'nontraditional' look at benchmarking drives. My scope in this testing was to see what SSD configuration benefited the most from running on SATA3 / 6.0Gb/s and SATA2 / 3.0Gb/s ports.

 

I'll put a link in my Signature with PC specs, but I'm still rocking the ASUS P8Z68-V PRO/GEN3. My means for testing between SATA3 and SATA2 speeds is due to my motherboard only having 2x Intel SATA3 ports. Why not use the Marvell PCIe SATA 6Gb/s controller? because it sucks. You are welcome to try it over the Intel ports, but I came to the same result as most, as it is bolded. The Intel SATA2 ports will perform better.

 

As you will see below, really the only benefit the 2x Corsair drives gain on SATA3 ports, is a bit higher throughput. Benching across utilities and trying real-world load tests, the difference was moot, imo. This resulted in my leaving the RAID0 drives on SATA2 ports, while putting the Samsung 840 Pro 256 GB onto a SATA3 port. The Samsung and it's newer tech just got far more benefits from doing so. Please also keep in mind I got the Corsair drives little over a year ago, while the Samsung is about 2 weeks old to me.

 

I'd also like to add this info about the benchmarks to follow, as it explains how each benchmark does, or at least used to work. Thank HW Secrets (Corsair Force GT vs Force 3 review)

The Force GT features synchronous memory rather than asynchronous memory, which performs much better when dealing with incompressible data. It is incompressible data transfer rates that are measured in AS SSD and CrystalDiskMark.

In ATTO disk benchmark, however, the drives exhibited a much more similar level of performance as it tests the drives using compressible data. Some of you may be asking yourselves why Corsair uses asynchronous memory in the Force 3 rather than synchronous memory. Well, synchronous memory is more expensive, accounting for the USD 30 price premium that Corsair demands for the Force GT over the Force 3.

 

Back to the cut of my jib for this thread.

SSD Specs:

 

Corsair Force GT:

SandForce SF-2200 SSD controller

Memory: Micron 29F64G08CBAAB NAND

 

Samsung 840 Pro:

Samsung MDX SSD controller

Memory: 21nm MLC NAND

 

I'm uploading benchmarks for your review, but the slim of it is that the Corsair drives suffer no real loss in performance short of sustained writes. In my experience the IOPS involved with Read and Write operations make far more of a benefit in respect to sustained data rates. Thus why I have decided to keep my RAID0 Corsair drives on SATA2 ports, while keeping the Samsung on a SATA3 port.

 

Namely, because the enhancements of the Samsung 256GB drive are beating the Corsair 90GB x2 setup of my RAID0. SSDs also benefit from the larger memory sizes, so that's part of the scenario here as well. When the Samsung was put onto a SATA2 port, it saw large drops across the board. It won priority for my limited Intel SATA3 ports.

 

Turns out I missed a few 1 to 1 screen caps to match for ATTO, but the drop off should still be apparent in the other programs, as I was seeing 50% drop off on both drives with ATTO.

ASSSD_CorsairRAID0_ForceGT-SATA2.png

ASSSD-Corsair RAID0-SATA3.png

ASSSD-Samsung-SATA3.png

ATTO (Marvel) Samsung SATA3.png

ATTO-CorsairRAID0-SATA6.png

ATTO-Samsung840ProSSD-SATA3.png

HDTune_CorsairRAID0onSATA2.png

HDTune_CorsairRAID0onSATA3.png

HDTune_Samsung840ProSATA2.png

HDTune_Samsung840ProSATA3.png

Link to comment
  • 6 months later...

I ran into a SMART Error: 0 (zero) on my RAID Array. All individual SSD tests were a complete pass, but the controller said the listed RAID issue.

 

After testing continued, I found the cause led to defective RAM. This is the 2nd time I've placed an RMA on this 16GB / 4 sticks of DDR3. CMZ32GX3M4A1600C9 is the model of RAM I am dealing with. I saw a few (maybe 2) hard locks on the machine with the RAID0 array, but I think that may have been due to the listed RAM, as I saw no data loss and each OEM test was good for the SSDs.

 

Hoping for a swift RMA. Ideally with no downtime. I paid for the advanced RMA last run, but I'm not in the means to pay that this round.

Link to comment
  • 4 months later...
  • 5 weeks later...
  • 9 months later...

Ryan are you seeing this issue. im not seeing this and i have yet to install the firmware because i dont want to break my raid. Figured when windows 10 comes out ill break raid flash then raid windows 10.

If you look at the screenshot you can see the performance degraded. This test unit 840 EVO SSD spent more than three months of the shelf, he results are pretty shocking: the months old data on the disk was only recoverable with an average read speed of 35.1MB/s.

After formatting the disk and loading it with new data the disk averaged 430MB/s in HD Tach, there's a abvious issue with long-term degradation of data stored in the triple-bit-per-cell memory.

 

post-30526-0-76246600-1422450497_thumb.jpg

Link to comment

Weird man. Once I had a notification my RAID0 was degraded (due to bad RAM I think), I backed up all my data from those drives and migrated it to a single Samsung 840 Pro SSD. Attached my benchmark with that app below.

 

I did chuckle how old this application is, as I had to turn on Windows 2000 compatibility for it to run. :p According to CrystalDiskInfo, I have written 14.57 TB to this drive. A large part of that was some basic networking services I had enabled, that were constantly hashing out character strings.

HDTach-2015-01-29-SamsungSSD.png

Link to comment

I hear that. I think the reason RAID breaks the ability to flash a device, is because it presents both devices as one storage device to the OS. I'm not saying it would not be nice to apply a firmware flash to a device in RAID, but that is one of the big tradeoffs currently, for sure.

Link to comment
×
×
  • Create New...