• 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.

HDD RAID and HDD Variation

I

icor1031

[H]ard|Gawd
Joined
May 29, 2010
Messages
1,295
Tests were never ran simultaneously.
Drives were zeroed before being tested.

They seem to all be using the same firmware. And what's more, it does not appear there are other versions for these drives.

I know my controller is terrible, and thus you will see that more drives is not always better.
All 6 ports on my board are SATA 3. They all use the same controller.

Comments?

Zero

zerofill.png


Four drives, tested one at a time.

separate.png



-------------

Two drives in RAID 0

Both pairs
201v.png
-
202g.png


Three drives in RAID 0

Changed one drive
305o.png
-
306k.png



Four drives in RAID 0

401ti.png
 
strange there would be that much variation, maybe one of the sata cables is bad, try swapping the cables and testing, that or like you pointed out the controller itself is not liking it, try to see if there is updated drivers for the controller.

However looking at the bottom couple, the variation is to be expected as well, nothing is perfect after all and for them to be like what within 4% of each other in speed, thats actually not that bad. I had 2 caviar black 640 that sometimes would test at 90mb/s and other times 122mb/s so it can kind of be expected, I suppose it may be the way the drive is being read or something?

I believe another test utility is Crystal Diskmark, as well as HDtach, maybe try those out as well?
 
Are these OS or NON-OS drives/arrays? If they are OS, there could be alot going on in the background at different times leading to the discrepancies. Also, what motherboard do you have and which SATA ports are they being plugged into. If you plugged 2 into the onchip SATA and 2 connected to an addon controller, that could cause issues as well. Give us some more info please. Also, did you use the same cables for all the tests?

With the exception of the horrible write speed of test 1 drive 2, they are all within about 5% or so of each other, which for the Samsungs is about right and could account for everything from thermal variations, reseeks, vibration adjustment, etc. Do they all have the same settings with the Samsung software, acoustic management modes, etc? Check them all with the Samsung ES Tool software to verify that.
 
Last edited:
mwroobel said:
Are these OS or NON-OS drives/arrays? If they are OS, there could be alot going on in the background at different times leading to the discrepancies. Also, what motherboard do you have and which SATA ports are they being plugged into. If you plugged 2 into the onchip SATA and 2 connected to an addon controller, that could cause issues as well. Give us some more info please. Also, did you use the same cables for all the tests?

With the exception of the horrible write speed of test 1 drive 2, they are all within about 5% or so of each other, which for the Samsungs is about right and could account for everything from thermal variations, reseeks, vibration adjustment, etc. Do they all have the same settings with the Samsung software, acoustic management modes, etc? Check them all with the Samsung ES Tool software to verify that.
Click to expand...

"All 6 ports on my board are SATA 3. They all use the same controller."

They drives were zeroed. My OS is an SSD. Thus, non-OS. In fact, no boot record or file system.

AsRock 880G Extreme3
 
I tried looking into it, all 6 sata ports are as is, it does not list a different controller, as in most boards provide 4 via x chip and 2 vs y chip, in this case it does not list it as such.

I think setting it into AHCI mode, or making sure the bios is set accordingly. The only thing I can think is something is maybe not allowing it to react the way it should, although, like mw pointed out, they are within each others specs/variablity, nothing is perfect, if in all test, multiple times, one of the drives showed say 70mb avg and all others showed 180mb average then I would be concerned, but they are close enough that I would call it just bench discrepancies more then anything else.

Wont hurt swapping to different ports, or even swapping cables around, you may figure it out better this way?
 
Thanks, but I'm not trying to figure it out.
As you see, I tested each 'pair' in 2 drive raid, and swapped a drive in 3 drive test.

Quite simply this onboard AMD RAID controller appears to suck. My old Intel did better, up to at least 3 drives.
 
well, if the so called performance from the AMD raid controller sux, why are you not still using the older Intel one then????

Also accordingly http://www.xtremesystems.org/forums/showthread.php?237319-Samsung-Spinpoint-F3-Raid0-Benchmarks this shows on average you might be a tad slower then you should be, but really not by much, to me it seems like something else is amiss.

There is not a single thing I have ever seen that will get its out of box, max rated speed, everytime. Someitmes its drivers, firmware, accoustic management yadda yadda yadda set improperly.

If you came here to ask opinion, help etc, but are not willing to figure out what the real issue is, then why did you ask for comments which I am assuming also meant suggestions?

Just one more thing to point out, you did say you know the controller is not a faster one, but you are getting overall very close to what you should be seeing from what I can tell, that and do realize, it is an 880G based motherboard, not thier best but then again not the worst, Intel and AMD both do this, you want the best, you have to pay for it, be it the controlled used or whatever.

I think 1 or 2 of the drives are below where thye should be when all is running, however, I believe there is something else going on, I do not think the controller itself is "faulty"
 
If you came here to ask opinion, help etc, but are not willing to figure out what the real issue is, then why did you ask for comments which I am assuming also meant suggestions?
Click to expand...

Thanks, but I'm not trying to figure it out.
Click to expand...

I posted to give information. People can see a normal variation between drives, RAID, and on poor controllers - lack of performance increase.

And I'm not on the Intel one because I can't use my 1055T on it... My E8400 would be a drag.
 
Yes, AMD motherboard aka FakeRAID controllers are not very good beyond RAID1.

Seriously consider using software RAID if in Linux or UNIX, or a different FakeRAID or hardware RAID controller if under Windows.
 
Yes, AMD motherboard aka FakeRAID controllers are not very good beyond RAID1.
Click to expand...
Even so, Intel beats AMD by at least 10%.

AMD can't even do AHCI correctly.

AFAIK the only reason to buy AMD is the lower cost.
 
Old Hippie said:
Even so, Intel beats AMD by at least 10%.

AMD can't even do AHCI correctly.

AFAIK the only reason to buy AMD is the lower cost.
Click to expand...

Yep, they win on the low-cost front.

But hey, at least both are better than NVIDIA chipsets! :p
Oh wait, NVIDIA doesn't make chipsets any more, snap! :eek:
 
Last edited:
Well, I am using 3 single drives on my M5A99X EVO, 1 Agility 3, and 2 caviar black 640, they are actually getting better speeds then I did with my P5Q-E motherboard, so I do not know about raid, but overall, it seems to operate just fine on my end, although, I suppose that also has something to do with the market segment, I would assume the higher end boards do better simply based on components being used on the motherboard.

Example the 970EVO just does not compete well with 990X EVO which doesnt compete with Sabertooth 990FX, move up the ladder and the components used are slightly better, as is the bench/testing of hard drive/ssd access, speed, latency etc. Maybe I am wrong, and I know for a fact that AMD side is a "little" slower for drive speeds, and Intel does do better in this regard, however getting a lower end board or at least a more "mainstream" board will have something to do with this, and I suppose the older gen boards compared to the newest and best I can assume there is differences as well.

That and I am running in AHCI had zero issues thus far, to me without some kind of direct proof it is just fud, whenI see my caviar black only doing 2mb/s faster be it Intel or AMD I do not call that enough speed difference to say 1 is crap and the other isnt, course I have zilch experience with Raid so possibly more goes into the picture then I as a user am seeing.
 
Yeah, AMD chipsets and SATA controllers are just fine, but the FakeRAID implementation is outdone by Intel chipsets however.

Your issue is well beyond the chipset however and is definitely the drive having a problem.
 
Well I suppose that they now are able to do software based SLI or put a NF200 chip on the gpu itself, and current chipsets generally offer enoug lanes and such there is no real "need" for Nvidia to do chipsets anymore, though it would be cool if they had a nf200 chip on thier gpus not just the top end ones, cause generally speaking, though the NF200 chip is crap in many ways, it does offer the best SLI scaling/interface as it stands(to my knowledge)

The last Nvidia chipset was the 980i if I recall?
 
Old Hippie said:
Even so, Intel beats AMD by at least 10%.

AMD can't even do AHCI correctly.

AFAIK the only reason to buy AMD is the lower cost.
Click to expand...

Well, AHCI is an Intel technology. For a time, Intel couldn't do AMD's x86-64. At all.
 
That was because it was AMD that made it all work properly :p

Intel DID have an x64 set way back when, but they never could figure out the kinks.

Either way, like it or not, it is a dmn good thing we have Intel, AMD, Nvidia and others, without them all, we would not be enjoying what we do, this is something all the "haters" and "fanboys" should keep in mind, and try to show relevant descisions/recomendations accordingly, this way here maybe we all get and continue to get the best we can for as long as we can :)
 
dragonstongue said:
Well I suppose that they now are able to do software based SLI or put a NF200 chip on the gpu itself, and current chipsets generally offer enoug lanes and such there is no real "need" for Nvidia to do chipsets anymore, though it would be cool if they had a nf200 chip on thier gpus not just the top end ones, cause generally speaking, though the NF200 chip is crap in many ways, it does offer the best SLI scaling/interface as it stands(to my knowledge)

The last Nvidia chipset was the 980i if I recall?
Click to expand...

It was the 980a and 790i for each platform.
Both were incredibly power hungry and had more heat output than the CPUs themselves.
 
You must log in or register to reply here.
Back
Top