Super Slow SSD r/w speeds

Nathen313

Reputable
Aug 17, 2014
49
0
4,530
So recently I realized that one of my drives was super slow. I did a r/w test using crystalmark and userbenchtest nad they both concluded to Read: 220 Write: 73.4 Mixed: 23.4 I have another SSD of the same type and it was significantly faster. Yes my ports are Sata 3 and it is set to ACHI mode.

Bench Test Scores
http://imgur.com/a/jQNyS
 
Solution
I read the same review Geo. If you look at the random and sequential 2MB read/write tests they both show the drive ~500MBs on the read. He gets that with the larger drive, but the smaller one doesn't get anywhere near that. He only gets 220MB read.

I'm still at a bit of loss as to why it's doing so bad. My problem was solved by moving my 840pro from SATA 5 too SATA1. (My 470 is on SATA0.) I see in that pic that both drives show how full they are, the smaller one is "nearly full" at only 25GB free. As a percentage you are looking at only ~22% free, that might be enough to slow it down? Is the 120GB your OS drive? Maybe OS things are slowing down it's scores? Finally I see they are using different firmware numbers, though the...

4745454b

Titan
Moderator
Is the drive full? Full SSDs will slow down.

http://www.storagereview.com/corsair_force_series_ls_ssd_review

According to this review both drives are performing worse than they should. You should be looking at 560/535, and even your faster 240GB drive is writing at only ~300MBps. I have this happen with I got my Samsung 840pro. Don't use SATA3 ports that are attached to secondary chips. Make sure the SATA3 port you use is on the main chipset. I suspect you have them hooked up to the wrong ports. Once I did that I went from ~350MB write to 525MB.
 

Nathen313

Reputable
Aug 17, 2014
49
0
4,530



The drive is not full,

I tried
achi mode
changing sata ports
changing sata cables
swapping mobos (Same model)
reinstalled windows
optimization tweaks
 

Nathen313

Reputable
Aug 17, 2014
49
0
4,530


The drive is not full,

I tried
achi mode
changing sata ports
changing sata cables
swapping mobos (Same model)
reinstalled windows
optimization tweaks
 

4745454b

Titan
Moderator
System specs? How full is "not full"?

You said you swapped ports, but if both ports are on the second controller they will both be slower. You need to make sure you have them plugged into the "main" ports controlled by the chipset.

You need native chipset port, AHCI turned on. Again, you should be seeing near or above 500MBps on both drives. You aren't seeing that on either.
 

Nathen313

Reputable
Aug 17, 2014
49
0
4,530


It's got like ~75gb free out of a 120gb, already checked they are on Intel controllers.

4790k
Asus vii hero
Gtx 1070
16gb corsair ram
 

4745454b

Titan
Moderator
I read the same review Geo. If you look at the random and sequential 2MB read/write tests they both show the drive ~500MBs on the read. He gets that with the larger drive, but the smaller one doesn't get anywhere near that. He only gets 220MB read.

I'm still at a bit of loss as to why it's doing so bad. My problem was solved by moving my 840pro from SATA 5 too SATA1. (My 470 is on SATA0.) I see in that pic that both drives show how full they are, the smaller one is "nearly full" at only 25GB free. As a percentage you are looking at only ~22% free, that might be enough to slow it down? Is the 120GB your OS drive? Maybe OS things are slowing down it's scores? Finally I see they are using different firmware numbers, though the 120GB looks to be on a newer one. Try looking online to see if there is a newer firmware for that drive. I assume if you swap cables between the 120 and 240GB drive the performance for both stays the same?

I take back what I said about the 240GB drive having issues. While the start of that review did say it should hit ~535MB in write tests none of the tests showed it can hit that. You are lucky to hit 300MBs in write tests when they test it. Shame on Corsair for listing "up to 535MB" write, but give us a drive that can only do ~300MBs. Looks like your 240GB drive is fine as measured by real tests.
 
Solution
SSD benchmarks seem to focus on data transfer rates.
To me, the real test is the read/write latency at a queue depth of 1 or 2.
That is what we do mostly.
Latency is a number which is generally missing.

The OP only recently realized that his drive was slow, presumably because it did not meet expectations from a synthetic stress test at unrealistic queue depth conditions.

I say, do not worry.
 

4745454b

Titan
Moderator
I'm not disagreeing with those points. It should still meet the listed/measured transfer rates though. Listed isn't going to happen because the review site didn't get anywhere near that. But he's not at the lower numbers they measured either. They were able to get ~300MBs write, and he's not even at 100MBs. Something is off. Even if he didn't realize that until he ran the tests.