Show us your SSD performance 2

[Hopalong X]

The one good thing besides the confusing nums is that in real life it probably makes very little difference. A little more or a little less in R/W times is hardly measurable during daily operations.

I totally agree.

There is some questions I have about the alignment though. Is there an alignment number that is better than the rest?

If so, how would I change the alignment number?

I plan on imaging the Vertex4 tonight and start from scratch with both the Vertex4 array and the Vertex3 array. I need to be consistant with the tests.
That is why I am asking about the alignment.

As you can see on AS SSD yours is 1024 and that is perfect.

 

[whs]

The one good thing besides the confusing nums is that in real life it probably makes very little difference. A little more or a little less in R/W times is hardly measurable during daily operations.

I totally agree.

There are some questions I have about the alignment though. Is there an alignment number that is better than the rest?

If so, how would I change the alignment number?

I plan on imaging the Vertex4 tonight and start from scratch with both the Vertex4 array and the Vertex3 array. I need to be consistant with the tests.
That is why I am asking about the alignment.

I understand that 1024 is a good alignment. But I am not sure whether that really performs better than any other valid number.

If you want to change it, there are several ways. I have described one method here: http://www.sevenforums.com/tutorials/205905-ssd-fix-misalignment.html

 

[BarryBGB]

Hopalong X,

Yes, those are but I have a couple with 103424 as well. Not sure what happened there. I think that is when the Vertex4 was with firmware 1.3 and I did a clean install instead of imaging.

 

[Hopalong X]

If they have System Reserve they often end up at 103424.

1024 is often the offset when there is no System Reserve partition.
As WHS stated above it does not really matter.
The only important thing is the number showing divisible by four (4)?
In both case yours are so both are perfectly fine.

Mike


This is one you previously posted below 32K. It is divisible by 4 so you are good to go.

 

[BarryBGB]

The one good thing besides the confusing nums is that in real life it probably makes very little difference. A little more or a little less in R/W times is hardly measurable during daily operations.

I totally agree.

There are some questions I have about the alignment though. Is there an alignment number that is better than the rest?

If so, how would I change the alignment number?

I plan on imaging the Vertex4 tonight and start from scratch with both the Vertex4 array and the Vertex3 array. I need to be consistant with the tests.
That is why I am asking about the alignment.

I understand that 1024 is a good alignment. But I am not sure whether that really performs better than any other valid number.

If you want to change it, there are several ways. I have described one method here: http://www.sevenforums.com/tutorials/205905-ssd-fix-misalignment.html

Now I remember! I used Diskpart when I first got my Crucial in 2010 and used that method. Thanks for the tip.

If they have System Reserve they often end up at 103424.

1024 is often the offset when there is no System Reserve partition.
As WHS stated above it does not really matter.
The only important thing is the number showing divisible by four (4)?
In both case yours are so both are perfectly fine.

Mike


This is one you previously posted below 32K. It is divisible by 4 so you are good to go.

Ah yes. That was the single Vertex3 but not sure how it got to 32K.

You are right about the 103424, I remember I did a fresh install with the DVD and of course Win decided to add the 100MB partition. I can't tell you how much I hate that. I remember now that when I saw that, I started over and partitioned the drive from another OS just to keep Windows from making another one.

As far as the alignment goes, someone on another site told me that to do a proper comparison of the single, Vertex3 Raid, and Vertex4 raid, they all need to have the same alignment.

I never thought that mattered, only that they were aligned.

Am I wrong or are they wrong?
As it pertains to benchmarks that is.

 

[Hopalong X]

They are wrong.

Benchmarks can vary as much as 2-5% from run to run from what I have seen.
TRIM, and garbage collection (GC). How close to needing TRIM run? Did it just run GC or is it almost due? No way of knowing after you run the Bench the first time when the SSD is new.
Antivirus running, indexing by Windows all can have an effect on the benchmark. These shut down if running in the background but did they have time to shut down before the benchmark started running or slightly after? That can effect the results.

Benches can be good to make sure everything is good when they are new. After they are used a while if it seems to be slow and you run the bench again and it is slow then???
Say your 4K drops from 24mb/s to 10Mb/s. Now you know something is off.

Sequential drop from 500Mb/s to 490Mb/s is not a big deal on the other hand.

Otherwise enjoy.
Mike

Your question. As far as I know alignment has no effect on benchmarks unless alignment is wrong- not divisible by four.

 

[BarryBGB]

They are wrong.

Benchmarks can vary as much as 2-5% from run to run from what I have seen.
TRIM, and garbage collection (GC). How close to needing TRIM run? Did it just run GC or is it almost due? No way of knowing after you run the Bench the first time when the SSD is new.
Antivirus running, indexing by Windows all can have an effect on the benchmark. These shut down if running in the background but did they have time to shut down before the benchmark started running or slightly after? That can effect the results.

Benches can be good to make sure everything is good when they are new. After they are used a while if it seems to be slow and you run the bench again and it is slow then???
Say your 4K drops from 24mb/s to 10Mb/s. Now you know something is off.

Sequential drop from 500Mb/s to 490Mb/s is not a big deal on the other hand.

Otherwise enjoy.
Mike

Your question. As far as I know alignment has no effect on benchmarks unless alignment is wrong- not divisible by four.

Thank you Mike for the very clear description.

 

[Hopalong X]

Thank you Mike for the very clear description.

Rats!!! I was trying to confuse you. :devil:

 

[smarteyeball]

128GB Vertex 2


128GB Crucial C300


128 Crucial M4


240GB Intel 520



I was going to Raid the intel drives, but figured they're fast enough without doing so.

 

[Dave76]

Nice upgrade.

Any noticeable difference between the 128GB M4 and the 520?
The 4k read and access times are slightly better on the M4 but, my guess is it's not a noticeable difference.

 

[smarteyeball]

Yeah, nothing noticeable.

 

[Terronium 12]

My Agility 3 is running numbers far below what's advertised, and while it's not a big issue at all (things still load incredibly fast!) I'm wondering if I'm doing something wrong.

I can post results from CDMark, if necessary.

 

[A Guy]

Download and run AS SSD

Alex Intelligent Software - Downloads

Post a screenshot of the outcome. A Guy

 

[Terronium 12]

I've no idea where these scores came from - first time I've seen numbers that high.

 

[Dave76]

I think you have a world record there, that SSD is a keeper

Your alignment and driver looks good.

 

[Golden]

My 2 year old Vertex2 60GB still chugging away reliably. Upgrade in Decemeber....thinking Samsung.

 

[Hopalong X]

Golden

Good thinking.
I like my Samsung 830 128GB I got last week.

Intelligent minds think alike.

 

[Golden]

The Samsung 830 128GB is exactly the model I was thinking of.

 

[Hopalong X]

The Samsung 830 128GB is exactly the model I was thinking of.

Genius at work!

 

[mjf]

I didn't realize this before

"Basically, even though you have Windows and your drive nicely aligned - there's nothing stopping developers from reading and writing whatever they want. Which is a perfectly valid thing for software to do but will show as a non 4k access."
SSD SMART "Non-4k Aligned Access" - Crucial Community
So I wonder if software developers don't have SSDs in mind then real world software may not reflect blistering benchmarks?