Jump to content
Corsair Community

Corsair Force 3 120GB - slow?


Recommended Posts

Hello everyone,

 

Is this a normal Corsair Force 3 120GB performance?

 

http://i286.photobucket.com/albums/ll83/ChoasPhoto/Untitled-1-36.jpg

 

NOTE:

Not in AHCI mode, AHCI setting not seen in BIOS

On SATA 2 and not SATA 3

Fully optimized for SSD (disabled defragmention, indexing etc.)

Link to comment
Share on other sites

Like you said, you're running a SATA III drive on a SATA II port, so you're being limited by that. Your ATTO speeds look like you're approaching the physical limitations of a SATA II port, but it's hard to tell with that small pic (250/235?). Post a pic of just ATTO.
Link to comment
Share on other sites

New MoBo or maybe some external controller, but the latter isn't likely to provide full speed.

 

I can tell you for a fact that Force 3 GT 120GB on a good SATA III controller exceeds it's rated speeds, I''ve had it hooked up to an Asus P8Z68 Deluxe Gen3 to one of two Intel SATA III ports directly, and it was doing more than 560MB/s (ATTO) easily.

Link to comment
Share on other sites

Those CrystalMark scores wont go up by much btw on SATAIII, thats because the drive is an Asynchronous NAND memory, and such memory is just plain crap.

Asynchronous NAND cannot handle uncompressed data so well and therefore will only be slightly faster then a traditional HDD on thatt front (in the 4k range).

ATTO is a superpositive crap benchmark, in my opinion, cause it only uses compressed data, which the controller (sandforce) and memory can handle just fine, but with uncompressed data, it chokes and almost goes to HDD speeds, so I wish the companies would stop using that crap benchmark, cause it only shows HALF the truth!

If you want the speeds they advertise on all data (not just compressed) you'll need an Synchronous NAND drive also, prefferably with a Marvell controller, since they can handle uncompressed better then Sandforce.

 

Edit:

From my Force3 120GB SSD, 2 months old, only Windows 7 with its programs on it (~35GB in total, rest free) on a SATAIII interface (see specs for details).

http://i812.photobucket.com/albums/zz44/DaLoona/ScreenieSSDbench.png

Link to comment
Share on other sites

DaLoona, While I agree with some of what you wrote, which is simply reality, such as the difference between synchronous and asynchronous NAND, we should look at the big picture too. Using asynchronous NAND will (should) lower the price of a SSD, allowing people to buy one when they otherwise could not afford it. It is unrealistic to expect them to perform at the same level all the time, the lower price does not mean we are getting something for free.

 

I cannot agree that a SSD using asynchronous NAND with incompressible data will only be slightly better than a standard HDD with 4k data. To test that, try both types of drives on AS SSD Benchmark, which uses only incompressible data. Testing a HDD with AS SSD is both funny and pathetic, once it finishes the large sequential data test, it will run on and on trying to complete the 4k tests. Any SSD will be finished with the entire test in a few minutes, usually about three minutes.

 

I ran a AS SSD Benchmark on my Seagate Barracuda 1GB 7200RPM HDD. While not as fast as a Velociraptor HDD, it certainly is not a slow HDD. This HDD is not a OS drive, and barely has anything on it, ~23GB used space out of the formatted size of 931GB, so less than 10% used. Meaning that it is not running on the innermost track/cylinders at all, although I would not call it "short stroked".

 

My understanding of AS SSD is that it runs the same sub-test multiple times, and displays an average score, so the time the entire test takes to complete is not necessarily a measure of performance. But I can say that one run of AS SSD Benchmark on this HDD literally took between 1.5 and 2 hours to complete. The results are attached below.

 

If I may compare the results of different tests, usually not a valid comparison, the CrystalDM 4k read result of 25MBs is IMO a typical result for a SSD with incompressible data. My HDDs result of 0.61MBs is less than 1/40 the speed of the SSD. Or the SSD is over 40 times faster than the HDD. The SSDs 4k write speed (by chance) is a bit over 100 times the HDDs speed.

 

The 4k QD 32 SSD results, which is the same thing as the AS SSD 4k 64 Thrd test (QD = Queue Depth, QD 32 means 32 outstanding requests in a line. Thrd = Thread, 64 Thrd means 64 outstanding threads (same as QD), although the maximum Queue Depth of all current SSD is 32) shows a similar result. The SSD here is over 50 times faster at reading, and about 93 times faster at writing. Those familiar with AS SSD know that the final score for my HDD is terrible compared to any SSD, being at least 10 times that score, and scores of 20 times or more are certainly not unheard of, with a single SATA III SSD. If you are familiar with SSD IOP specs, and the IOP test results of AS SSD (they usually match well), you can see the IOP results of my HDD are just terrible from top to bottom. HDD manufactures never quote 4k speeds, or IOPs in their specs. Since NCQ was first invented to improve the performance of HDDs, before SSDs existed, I would think the HDD scores would be better.

 

You can also see that my HDDs sequential write score surpasses the SSD, and the sequential read speed of the SSD is only 1.7 times that of the HDD. The SSD performance has declined over time, and incompressible data is the worst case situation for a SandForce controller SSD (the type of data does not make a difference with HDDs.) The usual average ratio of compressible to incompressible data is said to be 2/3 compressible, 1/3 incompressible. But we can still see the other areas of performance where the SSD is many times faster than the HDD, and the access times of SSDs I have tested are less than 1/100th that of the HDD. That is why older SSDs are still so much faster than HDDs.

 

You are quite correct about asynchronous NAND, which is a spec buyers should be aware of and looking at, just like the speed of a HDD being 7200 RPM, or 5400 RPM. I agree that ATTO is more of a "potential performance" test, rather than a real world test, but few benchmark tests are real world tests. Given the reviews I've been reading lately, which are including more real world type tests, the Corsair Performance Pro SSD, with a Marvell controller, is one of the very top performing SSDs available.

 

ploppoman, The performance of your Force 3 SSD is normal for the PC you are using it in. If you can't run your SSD in AHCI mode, or with a SATA III interface, you'll never get the performance it is capable of. Sorry to say, if you don't have AHCI mode, your PC is not fully optimized for a SSD. If you have an Intel CPU/mother board, and can enable RAID mode and use the Intel IRST driver, you will be in AHCI mode, which is a subset of RAID. I think the same is true for AMD platforms.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...