• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Low write performance on G.Skill SF-1200?

F

falk09

Junior Member
Hi,

I was under the impression that SF-1200-based units did a good job keeping the performance up with use.

Installed a G.Skill Phoinex Pro 60GB, which is SF-1200-based. I run this on a MSI P45-board and Windows 7. AHCI is enabled and Intel Rapid Storage-drivers are installed. I also have a Crucial C300 in the system, which seems to be keeping the performance up very well. Old OCZ Vertex (1) with TRIM-supported firmware worked fine with this system.

I have the latest firmware from G.Skill on the drive.

I tried using hdparm from a Linux-live-cd to secure erase the drive. For a little while, performance is better - after some use, the performance seems to go down quite a bit. I run a few benchmarks, copied some files over and extracted som RAR-archived, and now I see this result:

AS SSD:
http://www.bildedump.no/pics/816f78265f2a9fbb046aea3bca8f5a76.png

(at the start, seq. write was almost 100 MB/sek, which still is a bit slow for a SF-1200?)

HD Tune Pro:
http://www.bildedump.no/pics/beb789898486a535e8f026d809651351.png
 
Last edited:
That looks like some very bad numbers to me, and I'm very interested in the reason for them also.

Seq reads- down by 25 percent
Seq writes-down by 76 percent
Random reads- down by 42 percent
Random writes make me believe the controller is the pre-release firmware, since it is better than the Agility 2 with the authorized 1200 controller by 23 percent. If so it is down by 65 percent compared to the Force 100, and the Vertex 2.

Part of the difference is because the comparison is between a drive with 64Gb flash, and one with 128Gb flash, but that only accounts for a small part of the disparity.

Perhaps there is something inherently wrong with this firmware, which is why it wasn't meant for release? Sorry, I'm just thinking out loud here.
 
Firmware is from G.Skills website. I updated the firmware because of the slowdowns I experienced.

I have a Vertex 2 here now. Will check how that will react to my system and my tests.
 
I will leave the system alone to catch some breath over the weekend and see how it performs on monday ;-)

But seems like the Crucial C300 had no problems with these tests...
 
I could see you use Intel storage drivers; are you sure you updated them to latest 9.6+ release? Otherwise you would not have TRIM support on your SSDs.

Also note that Sandforce models differ in spare capacity:

50GB model = 28% spare capacity
55GB model = 20% spare capacity
60GB model = 14% spare capacity
64GB model = 6.8% spare capacity (difference between GB and GiB)

All these four models have 64GiB (not GB) of physical NAND. The more spare area you have, the better the protection against performance degradation. The Sandforce SSDs have another trick: they create virtual spare space from compression. Assume the compression ratio gives you 10 to 20% of space savings. Those saved space cannot be used by the OS, but can be used by the SSD internally.
 
sub.mesa said:
I could see you use Intel storage drivers; are you sure you updated them to latest 9.6+ release? Otherwise you would not have TRIM support on your SSDs.
Click to expand...

Doesn't Windows 7's built in driver support TRIM? Do you NEED to install the 9.6 RST drivers if you are not using RAID or drives on a RAID controller in a non-RAID setup?
 
Windows 7 built-in drivers (msahci.sys and pciide.sys) both support TRIM; these work on normal SSDs on AHCI-controllers, including those running in "IDE emulation" mode.

But the original poster is not using the Microsoft drivers; he is using the Intel driver instead. This you can see in the AS SSD screenshot output (iastor.sys instead of msahci.sys). Now the Intel driver should also support TRIM, UNLESS you have an older version before Intel 9.6 drivers; 9.6 was the first Intel driver that supports TRIM.
 
Some say it is faster. And you have to use the Intel drivers if you want to use RAID, for example consider this setup:

Intel X25-M 80GB boot SSD
2x 7200rpm disks in RAID0 on Intel controller

Pretty straight forward. But assume both the SSD and the two HDDs are on the Intel controller. Now if you use Microsoft AHCI driver; the SSD has TRIM but the two HDDs won't be in RAID (though you can use software RAID in windows; but not bootable).

If you want to setup the two disks in onboard RAID, you would need to install the Intel RAID drivers. When you do, they will replace the Microsoft drivers. This could mean you lose TRIM support if the Intel driver does not support it.

Intel RST 9.6 supports TRIM, but only on SSDs not part of a RAID array; those will run in AHCI mode and have TRIM via the Intel driver.

Some people also say the Intel driver is a tad faster, particularly it does low-level optimizations which benefit some poor media players which do not buffer properly; with the Microsoft AHCI driver they might stutter frequently.

Unless you have a good reason, i would prefer the Microsoft drivers instead.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top