• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Is this normal?

Whach

Whach

[H]ard|Gawd
2FA
Joined
Dec 22, 2011
Messages
1,293
I just re-installed Win7 x64 on my system and decide to check the SSD speeds after noticing some sluggishness when installing windows and other programs (compared to when I first built the rig). I am using the latest drivers from ASUS - basically the same as the first time.



Compared to review benchmarks using the same tools, my write speeds seem to be lower (as much as 30%). I am using the standard cables that came with my motherboard. (I assume they are SATA 3 as I had no issues before - maybe deteriorating?).

Do SSD's become slower after a reinstallation as I wasn't expecting by this much.

Thanks for any help as I am baffled =/
 
Review benchmarks are for the most part straight out of the box. You have the Windows installation on the SSD -- most of the time (especially with SF) this tends to reduce benchmark performance somewhat.

I have a 120GB Mushkin Chronos Deluxe that I used for a couple weeks as a system drive and basically see similar numbers to what you have (in terms of percentage reductions). If you were to SE the drive, then put a partition on it and run CDM again, you'd see very similar numbers (to benchmark result you've referenced). This isn't a knock on SF -- that's just what happens after some usage (Xbit labs did a recent SSD roundup looking at more steady state performance). Your drive is fine, though AV software can reduce performance by a healthy chunk.

Most modern drives are quite good in this regard (maintaining performance over time). One thing to consider is this: windows is always reading or writing something. A lot of small requests, mostly, but this also reduces benchmarking numbers. But that doesn't matter, since you bought the drive to use and not benchmark.
 
Everything ckryan said, but if you really want to see what your SSD can do, disable EIST, C states, turn processor power management minimum to 100% and run the benchmark again.
 
Last edited:
MissJ84 said:
Everything ckryan said, but if you really want to see what your SSD can do, disable EIST, C states, turn processor power management minimum to 100% and run the benchmark again.
Click to expand...

Thx guys. is there anyway to regain the initial performance out of the box (ie first use situation)?
 
Whach said:
Thx guys. is there anyway to regain the initial performance out of the box (ie first use situation)?
Click to expand...

Yes. Buy a new one and open the box. You'll have initial performance out of the box restored completely :).

More seriously. The Force GT is a Sandforce 2281 based drive. I do not have direct experience with Corsair's drive, but all Sandforce based drives have a concept of "secure erase" which returns each MLC block to a completely initialized state. This will be as close to "out of the box" as you can get a used SSD. The exact mechanics of achieving secure erase depend on the OEMs firmware - which is another way of saying visit Corsair's support website and look for something along the lines of "Force GT secure erase tool".
 
PigLover said:
Yes. Buy a new one and open the box. You'll have initial performance out of the box restored completely :).

More seriously. The Force GT is a Sandforce 2281 based drive. I do not have direct experience with Corsair's drive, but all Sandforce based drives have a concept of "secure erase" which returns each MLC block to a completely initialized state. This will be as close to "out of the box" as you can get a used SSD. The exact mechanics of achieving secure erase depend on the OEMs firmware - which is another way of saying visit Corsair's support website and look for something along the lines of "Force GT secure erase tool".
Click to expand...

All of this lol. Contrary to popular belief, SSDs do lose performance as they fill up.

http://www.hardocp.com/article/2011/08/07/nand_flash_faces_off_synchronous_vs_asynchronous/4
 
You could get out of the box performance with a secure erase. But then you're gonna put the Win7 install back on it, defeating whatever performance increase you'd get from the SE. That's what happens to SF2281s used as system drives after a short time.

Yes, most drives do not perform the same as they are filled with data. The real point is that while you were benching the drive, Windows was still reading and writing to the drive. Not only that, but internally, the drive is wear leveling and performing background garbage collection, and a host of other stuff. So you could SE the drive with a variety of tools, but you can't just do that every couple days. And as soon as you put your data back on the drive and boot from it, you're gonna be in a similar performance state.

Most other SATA III SSDs do tend to manage this better, but I seriously doubt your perceived sluggishness is the drive. Based on the SMART data, your drive has written 1879GB to NAND, and 1473GB host writes. Normally, the raw NAND writes should be less than host writes on a SF due to compression. My guess is, you've been benching the shit out of the drive, resulting in a WA of 1.27.

So if you want, SE the drive, reimage your system back on to it, and quit benching it. SF doesn't handle random data gracefully. Ideally, your WA should be less than 1. 1.27 is still really good for an SSD, but either your workload is fierce or you've been benching it a lot. So chill on the benching.
 
Latest CDM uses random data by default, showing worst case perf for Sandforce drives.

Choose 0 Fill from File -> Test data and re-test.


Performance will likely be the same as when you tested new.
 
msitpro said:
Latest CDM uses random data by default, showing worst case perf for Sandforce drives.

Choose 0 Fill from File -> Test data and re-test.


Performance will likely be the same as when you tested new.
Click to expand...

I'm guessing you didn't read the OP.

He knows CDM is using 0 fill. He was experiencing up to 30% lower performance over new OOB numbers. If he was using 0-fill when the drive was new, current benches would look a hell of a lot worse. Random data may be worst case for SF, but 0-fill is just an unrealistic scenario for every desktop workload.

Besides, since his WA is over 1, it's safe to say that the drive almost never sees anything but mostly incompressible data.
 
Wha? Show me where OP says it's using 0 Fill on that bench?

When using 0 Fill CDM shows it on the Window Title as <ALL 0x00, 0Fill>


From what I have read OP likely tested using an old version if CDM when first build now reloaded and latest version of CDM now uses random by default ,therefore showing lower figures.

This explains where that perf has 'gone'


I cant see anywhere where the OP mentions he is aware of 0 fill vs random testing.
 
Most SF2281's end up with exactly that sort of performance drop after being used as a boot drive for a couple weeks, so I assumed he was wondering why it is that his SF isn't exactly performing as it was during fresh benches.

This has come up in 520 discussions, as they're identical in this respect. I have a couple SF2281s that look exactly like the OPs after two weeks of system drive duties (including some 30% drops in certain performance parameters). So I didn't want to insult his intelligence by telling him that he was benching it wrong.

Most people understand SF and compressible data by this point. Most people don't know that used state performance is not spectacular.
 
ckryan said:
Most SF2281's end up with exactly that sort of performance drop after being used as a boot drive for a couple weeks, so I assumed he was wondering why it is that his SF isn't exactly performing as it was during fresh benches.

This has come up in 520 discussions, as they're identical in this respect. I have a couple SF2281s that look exactly like the OPs after two weeks of system drive duties (including some 30% drops in certain performance parameters). So I didn't want to insult his intelligence by telling him that he was benching it wrong.

Most people understand SF and compressible data by this point. Most people don't know that used state performance is not spectacular.
Click to expand...

Bullshit.

I've been using a Corsair Force 60GB @ 90% full in this laptop for 18 months without 'secure erasing' once. Yes, boot drive, apps, pagefile, everything.

Benches exactly the same performance as out of the box new.


You may want to check this out: http://www.storagereview.com/ssd_performance_review_270tb_written



You REALLY have to murder drives with random writes (intentionally) for modern MLC SSD designs to show performance degradation doing desktop/laptop duties.

Enterprise is another matter entirely, however.
 
The crystaldiskmark I used upon re install was v3.0.1.c & I had tested using the default settings (it may have been a earlier version last time I tested). I don't usually benchmark at all. The last time had had done this (apart from this time) was when I had initially built my rig back in December '11.

I did try using the <ALL 0x00, 0Fill> setting as suggested and received significantly higher scores, but I suspect this is not a real-use representation (is that right?) Anyway, its all probably a storm in a teacup. Its still faster than my mechanical drive!
 
You must log in or register to reply here.
Back
Top