Show us your SSD performance 2

[A Guy]

AnandTech liked it

A Guy

 

[kbrady1979]

They also said Intel had ironed out the Sandforce issues in their 520 series of SSD's. I went through 2 before I switched to Samsung lol. I do respect their reviews though.

 

[essenbe]

Well, I had to try it. Here's what 2 M4's in Raid look like. The score looks great, but real world it is no faster than a single M4.

 

[mjf]

... but real world it is no faster than a single M4.
...

Interesting. Your "Real world experience" is what it's about in my view. I could return my M4 because of Benchmarks but I doubt I will.

 

[Snakeyeskm]

A thank you to effective garbage collection.

My Revodrive3 X2, due to its internal RAID configuration, is unable to take advantage of trim. However, after nearly one year of intensive use (and I mean intensive write operations). It's garbage collection has helped it maintain its performance (aided by 25% OP). The drive achieved a fully mapped status by February of this year and garbage collection has managed to cope with fairly heavy. I/O activity since then. Attached are two ATTO runs from September 2011, and August 2012.
The dates of each ATTO run are at the bottom-left of each attachment.

 

[wachu13]

Hello everyone
This is my first ssd. Installed last weekend and did clean install. My Inspiron 1520 have only sata 3 so will not run on full speed but still better then my old hdd. Here are the results, I hope they are in good range

 

[A Guy]

Welcome to Seven Forums wachu13. Looks good

A Guy

 

[Britton30]

Hello everyone
This is my first ssd. Installed last weekend and did clean install. My Inspiron 1520 have only sata 3 so will not run on full speed but still better then my old hdd. Here are the results, I hope they are in good range

One of the best total scores and killer access times, very good.

 

[wachu13]

Thx guys, so looks like everything is working like supposed to.
Is there are some kind of guidelines (range) for speeds? My friend will be switching to ssd this weekend and will like to know if his will work in good speed too.

 

[mjf]

Welcome to Seven Forums wachu13. Looks good

A Guy

Just for my curiosity - why do you think the sequential read score is that low?
and the 4k-64 read so high?

 

[bigmck]

I need some analyzing. What does this mean and is it bad, average, good or great?

 

[Britton30]

I need some analyzing. What does this mean and is it bad, average, good or great?

It could be a little better Jim, here's mine.


Do you leave you computer on and idling for a few hours a week? That's when TRIM, wear leveling, garbage collection etc goes to work and cleans up SSDs.

 

[bigmck]

I need some analyzing. What does this mean and is it bad, average, good or great?

It could be a little better Jim, here's mine.
View attachment 225843

Do you leave you computer on and idling for a few hours a week? That's when TRIM, wear leveling, garbage collection etc goes to work and cleans up SSDs.

Yes, it says on most of the time. Is TRIM automatically on or do I need to check and make sure? If so, how do I do that? Thanks,

 

[A Guy]

At a command prompt (start/run/cmd), type the following: fsutil.exe behavior query DisableDeleteNotify

It should respond back with DisableDeleteNotify=0 if trim support is ready and active. If it is not, type fsutil.exe behavior set DisableDeleteNotify 0

This will set Windows 7 to use TRIM when the drive and drivers are ready to do so.

http://www.sevenforums.com/tutorials/70822-ssd-tweaks-optimizations-windows-7-a.html

A Guy

 

[mjf]

I need some analyzing. What does this mean and is it bad, average, good or great?

AS-SSD apparently places a high weighting on the 4K-64 metric. From what I read apart from servers this is a pretty useless metric for most single user PCs. A queue depth of 64 would be unusual. I would have thought the 3 other more meaningful metrics looked reasonable.

I was puzzled by a recent poster who had good AS-SSD metrics except a poor Sequential read score. This would cause me more concern and I am beginning to wonder if AS-SSD is flawed.

 

[whs]

Thx guys, so looks like everything is working like supposed to.
Is there are some kind of guidelines (range) for speeds? My friend will be switching to ssd this weekend and will like to know if his will work in good speed too.

What exactly do you mean by "guidelines (range) for speeds"?

 

[chrysalis]

here is mine right now.

Notes.

I have done various tests probably already half filled drive

Initially was on intel port as everyone seems to reccomend intel for SSD. Was dissapointed in 4k performance, poor threaded performance and high cpu usage during threaded (maxed out all 4 cores on my i5).

Results from original as ssd test.

seq read 262.35
seq write 247.06
4k read 15.29
4k write 47.09
4k-64thrd read 179.22
4k-64thrd write 78.82
acc-time read - error cant find drive
acc-time write - 0.064ms

also the 4k-64thrd tests saturated my cpu, these sucked cpu hard, is this normal? I got a 4 core I5 that was otherwise idle during tests. So the cpu bottlenecked the 4k-64thrd benchmark.

Current which is in screenshot has following changes.

Using marvell 6G controller, which has 5G of pcie bandwidth enabled.
C1E off - this slows down ssd's funny enough.
CSTATES off - this also slows down ssd's however since it also disables turbo mode on the cpu it also slowed down threaded speeds on the intel port which were cpu bottlenecked.

So people may notice my marvell is in general outperforming my intel port. Both using msahci drivers. Some question marks if the msahci driver allows marvell to use trim tho. Review sites are generally anti marvell so I take their comments with a pinch of salt regarding marvell vs intel performance.

Even with my new results tho some will notice my threaded writes are slower than my non threaded writes, abd ny threaded reads dont have as large jump as other have got.

Single threaded 4k is better on my marvell.
Threaded on the marvell is worse however the cpu usage is reasonable whilst on the intel it uses all 4 cores to the max.
Sequential reads way better on marvell but writes very slightly worse.
access times better on marvell.

 

[chrysalis]

I need some analyzing. What does this mean and is it bad, average, good or great?

AS-SSD apparently places a high weighting on the 4K-64 metric. From what I read apart from servers this is a pretty useless metric for most single user PCs. A queue depth of 64 would be unusual. I would have thought the 3 other more meaningful metrics looked reasonable.

I was puzzled by a recent poster who had good AS-SSD metrics except a poor Sequential read score. This would cause me more concern and I am beginning to wonder if AS-SSD is flawed.

I dont think its the best app out there.

crystal diskmark in my view is better, Qd of 32 instead of 64 used, still very high but more realistic.

Also a guy called anvil made a great freeware ssd benchmark app called anvils storage benchmark, it allows you to customise queue depth etc.

samsung also told me in a support reply the drive will decline in performance above 32 queue depth as its optimised for home not server usage.

 

[Hopalong X]

here is mine right now.

So people may notice my marvell is in general outperforming my intel port. Both using msahci drivers. Some question marks if the msahci driver allows marvell to use trim tho. Review sites are generally anti marvell so I take their comments with a pinch of salt regarding marvell vs intel performance.

Even with my new results tho some will notice my threaded writes are slower than my non threaded writes, abd ny threaded reads dont have as large jump as other have got.

Single threaded 4k is better on my marvell.
Threaded on the marvell is worse however the cpu usage is reasonable whilst on the intel it uses all 4 cores to the max.
Sequential reads way better on marvell but writes very slightly worse.
access times better on marvell.

I have the exact same motherboard. Click my specs.
Your Intel controller is perfect for SATA 3.0Gb/s SSD's like my Intel 320 or the Samsung 470.

My Samsung 830 is a SATA 6.0Gb/s SSD just as yours and works much better on the Marvell 6.0Gb/s controller.

The Marvell controller on our motherboards is not as good on benchmarks as Intel 6.0Gb/s controllers but is much better than any SATA 3.0Gb/s controller.

So you have to match the SSD to the proper SATA controller which you have now done.

The point is you are now set up as I am with my Samsung 830 using the Marvell controller and that is the best.

 

[chrysalis]

yes so it seems, do you know if trim is working on the marvell with msahci?

to rest my theory if cpu was holding back threads on the intel, I moved the ssd back and overclocked my cpu (this be temporary as dont want to run this 24/7) and I got an extra 20meg/sec on the threaded 4k reads.

So I want to use marvell but the crunch point is if trim will work on it. Also slightly annoyed that I will have to have my graphics card pcie at 8x instead of 16x as well, but if I am right my geforce 460GT should still run at full speed with that spec.