Speed difference between Sata II and Sata III cables

[ZenZimZaliben]

According to SerialATA.org there is no difference between cables but just quality of the cable. To achieve 6gbs a high quality cable is required, but any cable will work as long as it has enough copper and low resistance.

http://www.serialata.org/documents/SATA-Revision-3.0-FAQ-FINAL.pdf

See Section 9.

I believe this is part of JEDEC set standards.

One thing that was mentioned was NCQ needs to be enabled.

 

[Steevo]

Update your intel storage drivers.

 

[1freedude]

Update your intel storage drivers.

Right, and my sabertooth x58 needed a bios update to get the marvel right.
something else...I'm confused as to what you mean by install, as in the drivers?

 

[MT Alex]

I have a Corsair Force 3 120GB, and an AsRock Extreme 4 Gen 3 P67, also using a SATA II cable. I just ran an ATTO for you to compare with. I also assumed it would be faster.

 

[PaulieG]

Well, I was probably worried for nothing, based on the ATTO bench. Looks like I'm hitting well above 500 read. I wonder why HD Tune reports much slower speeds?

 

[MT Alex]

Dang, I thought something was jacked up. Yours is quite a bit faster.

 

[PaulieG]

Dang, I thought something was jacked up. Yours is quite a bit faster.

Yeah, it's supposed to be one of the fastest 2nd gen SF drives. Yours should be a little faster, from what I'm seeing from your ATTO bench. In my case, I think it's an issue with HD Tune. Maybe it needs an update for the newer drives?

 

[MT Alex]

Silly me, I could have sworn I changed IDE to AHCI I did, but it's when I was using the Marvell controller.

If I wouldn't have seen your bench, I would still have things screwy.

 

[PaulieG]

Silly me, I could have sworn I changed IDE to AHCI I did, but it's when I was using the Marvell controller.

If I wouldn't have seen your bench, I would still have things screwy.

http://img.techpowerup.org/120118/Untitled251.jpg

Funny how that works sometimes. I'm still wondering why HD tune is reporting screwy.

 

[cadaveca]

try using a newer HDTune. Like 4.61.

I have no comments about test validity...it reports exactly what I expect, and I've never seen results like yours.

Contrary to EarthDog's post, I wanna jsut make it obvious that that comment about HDTune refers to the IOP test only, which you are not using.

 

[PaulieG]

try using a newer HDTune. Like 4.61.

I have no comments about test validity...it reports exactly what I expect, and I've never seen results like yours.

Contrary to EarthDog's post, I wanna jsut make it obvious that that comment about HDTune refers to the IOP test only, which you are not using.

Good catch Dave. Didn't realize the version on my flash drive was that old. Damn!

 

[cadaveca]

I play with HDTune a lot. 3rd favorite game, next to the 3DMarks and SuperPi.

 

[PaulieG]

I play with HDTune a lot. 3rd favorite game, next to the 3DMarks and SuperPi.

Well, it was a good theory, but it gave me close to the same results as 2.55 did. Seems ATTO reads it correct, and HD Tune does not. Hmm.

 

[Steevo]

HD Tune is crap. Try the trial Pro version to get accurate results. According to it my RAID array of SSD's only does 200ish average, copying files or otherwise I know it to be much more.

 

[PaulieG]

HD Tune is crap. Try the trial Pro version to get accurate results. According to it my RAID array of SSD's only does 200ish average, copying files or otherwise I know it to be much more.

Yeah, my last run was with the Pro trial. The benchmark test still reports wrong, but the IOPS seems to report correctly.

 

[EarthDog]

Well, I was probably worried for nothing, based on the ATTO bench. Looks like I'm hitting well above 500 read. I wonder why HD Tune reports much slower speeds?

http://img.techpowerup.org/120118/HDD_bench.png

Because of the type of test it is. I believe that, aside from all all of its other weaknesess in benching SSDs, that it uses data that is not compressible. OCZ SF2281(?) drives compress files and that is why it shows such great performance in ATTO but not in AS SSD or CDM which also uses data that is not very compressible.

IOMeter
AS SSD
CDM
ATTO

... all valid SSD benchmarks. Others, because of the way they test (buffered spot samples) the accuracy comes in to question (according to that article).

@ Cad - You are right, but took that out of context really. It was just another example of how poor that program can be when testing SSDs. The important thing to note is in bold at the end of the first paragraph . Nothing is perfect (including ATTO), but there are better tools that HDTune/Tach for the job is the point.

 

[Mussels]

most SSD's are slow with non compressible data, since they use compression to boost their speed.

its the key difference between say, a vertex III and an agility III from OCZ.


this is why ATTO is used, because its how the SSD makers test their drives - with compressible data.

 

[cadaveca]

Weird. HDTune works perfectly fine with my Crucial M4 and Corsair F60. I must assume that neither uses much compression then, as both report exactly the numbers the OEMs say they should.

Interesting info though.

Screens(both drives have 49.8 GB of data):

 

[EarthDog]

Notice those are not SF drives (which are terrible with incompressible data)?

EDIT: The F60 is SF... hmmm...

EDIT2: Its the 12xx SF controller...I dont know if it was better at handling imcompressible data...

EDIT3: Good god that performance is all over the map with that drive!!!

 

[cadaveca]

EDIT3: Good god that performance is all over the map with that drive!!!

Yep. That's wha happens when it's near full. Empty(tested two days ago), it's just a stright solid line. The access time pattern is interesting in SSD's, period, too.

I also see expected performance from mechanical drives, so I'm not sure why noone likes it much. OEMs like to use other programs...of course, becuase they portray the product in the best light possible. HDTune seems, to me, to portray accurate real-world results rather than optimized results.

 

[Steevo]

For me the Pro version worked perfect and the results were almost exactly the same as ATTO provided.


I know for a fact my drives can sustain over 200MBps of write as my mechanical array can supply 200+ of read, and copying large already compressed files and timing it, or using windows to tell me how many bytes per second it comes out to 209MBps or so, and even my large home movies, or some ziped folders with my pictures in them perform the same.


The standard HD Tune when performing write tests shows only about 170ish write on my SSD array though. So it is NOT indicative of real world performance for me.

 

[EarthDog]

Its made for mechanical drives (HD = HARD DRIVE) that isnt in question here as HDD results SHOULD be spot on with those applications (HDTune/Tach).

As far as HDTune giving real world results... well. In essence you are correct. A lot of files a person uses are not compressible (I think music is one that is difficult to compress?) so that does show the differences in the controllers and its compression abilities. That said, and as said before by me and subsequently others, that manufacturers use ATTO to gauge performance on the drive. So in order to not have people new to SSD's running applications that dont remotely show their specifications, ATTO should be used when trying to see if the drive is matching its specs (to bring this back on topic to the thread! ).

Our SSD reviews, after a discussion a year ago or so, do not use HDTune or HDTach because of that article. I see other major sites, like Anand are still using it so that makes us all scratch our heads a bit since they are one of the more respected review sites out there. I talked to Anand and he is aware of the article and the programs shortcomings. But since they used it in the past and is a comparable metric across drives, its still used regardless.