Sandforce and HD Tune "roller coaster" issue"

[G.SKILL]

Greetings,
More info:
These snapshots were taken today. The GSkill drives are only a few weeks old. The Intel drive is two years old and was my old boot drive.

I wonder why the SSDs' life span has dropped 10% already.

I have a RMA'd 60 GB drive coming back today, and I'm going with RAID0 on two 60 GB drives. This will be my second go.

grits

Hi grits

FW_3.1 has already fixed this bug.
It is bug issue.

[G.SKILL]

sorry guys

I didn't notice I tick this option but actually it won't affect the result.



It get better

[mauro]

GSKILL,

Thanks for your reply.
I have a question about your last result where you got a straight line after parted magic secure erase. You got that result after OS installation or you got that with ZERO data on the ssd (benched as secundary drive)?

[G.SKILL]

Hi mauro

I got with ZERO data.

if you have OS or any data on the storage, you can't get the straight line.
That's why our website point out new drive

[mauro]

Hi mauro

I got with ZERO data.

if you have OS or any data on the storage, you can't get the straight line.
That's why our website point out new drive

Thanks for you fast reply.
I have changed the thread title to "Sandforce and HD Tune "roller coaster" issue"

Also, I have found that most users of sandforce based ssds, like ocz vertex 2 for example, have the same roller coaster issue.

Here is an example of a vertex 2 I have found from random user on the web.



Looks like the problem is not related to the nands and I was wrong.
Looks like that is the way sandforce controller works after all. Kinda sad because the advertised performance of "an empty drive" is not real. An empty ssd has no use at all. We are all going to put data in it, so the performance that matters, is the one with data on it, and it is this "roller coaster" performance.

Still, I find confusing that THERE ARE some users with sandforce drives getting a straight line result in HD Tune even after the OS install. Because of that, I still think there is something to discuss here.

Please give us more info about it if you can.
Thanks.

[mauro]

I have changed the thread title to "Sandforce and HD Tune "roller coaster" issue", but it doesn't seen to work. Looks like the thread still have the old title.

After the Gskill explanations here, I don't think the thread deserves the old title anymore.

So please, some admin from gskill, change the thread title for me.

[G.SKILL]

Hi mauro

Thank you for supporting G.SKILL

I know for user it is quite difficult to accept this result.
so we try to do our best service for all of you.
if you have any suggestion, please don't hesitate.

by the way, I change title


Thanks

[eva2000]

Hi mauro

I got with ZERO data.

if you have OS or any data on the storage, you can't get the straight line.
That's why our website point out new drive

yeah mine above was with zero data too

the more data on ssd, the less performance it will get.

mauro, SSD performance depends on amount of free space on the SSD - more free space the better the performance.

So if your OS install has say 30GB of data, then a 120GB ssd will have better performance than a 80GB ssd = 240GB >120GB > 80GB >60GB. All ssds regardless of controller work the same way in terms of free disk space's relationship to performance. Same with traditional sata/pata disks too

[G.SKILL]

Hi eva2000

you got it.


thanks

[mauro]

Thanks eva2000 and Gskill for the explanation.
I can accept that larger SSDs will perform better, but that makes room for this question:

Why 60gb phoenix pro has +200MB/s seq. read speed in CrystalDiskMark, and a larger version of it, the 80gb, has only +150MB/s ? As per your arguments, a larger version should perform better.

Gskill advertised performance for 60GB phoenix pro:
http://gskill.com/products.php?index=286


Gskill advertised performance for 80GB phoenix pro:
http://gskill.com/products.php?index=311


I see that the write speeds improved, but read speeds are just much more important in a SSD.

So, why a larger phoenix pro (80gb) has lower read performance than a smaller phoenix pro (60gb)?

Is the 80gb limited by firmware or something? You think a future firmware update could give the 80gb version the same read performance as the 60gb version?

[rlb9682]

Hi mauro

I got with ZERO data.

if you have OS or any data on the storage, you can't get the straight line.
That's why our website point out new drive

I have the same fluctuations as others with zero data on the drive, testing as a secondary unit-nothing installed on it yet. I'll post pics later if it helps, but I had similar results to what I've seen with others 139mb minimun/230mb maximum read speeds with HD Tune Pro.

So, I guess my question is why would a new drive, unwritten, be getting fluctuations with zero data written on it? I guess this goes back to the Sandforce controller but it's not very encouraging with a new unwritten drive.

Edit:My drive came with 3.1 installed

[mauro]

I have the same fluctuations as others with zero data on the drive, testing as a secondary unit-nothing installed on it yet. I'll post pics later if it helps, but I had similar results to what I've seen with others 139mb minimun/230mb maximum read speeds with HD Tune Pro.

So, I guess my question is why would a new drive, unwritten, be getting fluctuations with zero data written on it? I guess this goes back to the Sandforce controller but it's not very encouraging with a new unwritten drive.

Me too. I have the fluctuations with the clean drive (zero data on it, after secure erase) too.

AND, I know people that have a straight line bench in HD Tune even when the drive has the OS installed on it.

AND, larger sdd like phoenix pro 80gb / 160gb gets lowers read performance compared to the phoenix pro 60gb.

All that makes everything VERY confusing.

[Betao]

I was coming to ask the same question...how to avoid burn neurons?

[G.SKILL]

Hi all

80GB is lower than 60GB on CrystaldiskMark because their layout and hardware design are different.
but they have same score on ATTO or IOMETER.

[mauro]

Hi all

80GB is lower than 60GB on CrystaldiskMark because their layout and hardware design are different.
but they have same score on ATTO or IOMETER.

Oh my god... If I knew about this, I would have got the 60gb or 120gb phoenix pro. If they have different hardware and design, I think it would have been better to name them something else, like "phoenix pro-E" for example.

Its very confusing for the end user. When we see a line of "phoenix pro" ssds (60gb, 80gb, 120gb, etc) we kinda assume they are the same product, just with larger capacity. And since larger capacity drives should perform better, we would all assume 60gb < 80gb < 120gb etc...

I feel kinda sad now that I ended up buying the 80gb phoenix pro.

Anyway, It's not just CrystalDiskMark... Here in ASSSD, you can see that the 80gb is also slower on the reads.

My 80gb phoenix pro:


Can you give me more information about the differences between the 80gb and the 60gb?
Looking on the benchmarks, it seens that the 80gb is stronger on the writes, and weaker on the reads. But read speed is the most important thing on the ssd.

My 80gb phoenix pro:


Overall, is the 80gb a slower drive then the 60gb?
Do you think a future firmware could improve the read speeds of the 80gb phoenix pro?