Finally joining Club SSD

[shiven]

From what I heard the higher capacity drives are a lot faster than the slower ones.

Isn't that a bit counter-intuitive? One would think that the larger the drive, the more the drive has to read through, so the slower it would be. I guess I don't know enough about the actual process...

[kiddingguy]

Isn't that a bit counter-intuitive? One would think that the larger the drive, the more the drive has to read through, so the slower it would be. I guess I don't know enough about the actual process...

.... and are also more expensive :(

[Odom Member]

Isn't that a bit counter-intuitive? One would think that the larger the drive, the more the drive has to read through, so the slower it would be. I guess I don't know enough about the actual process...

Often larger drives are faster due to the higher data density on the platters, or the amount of platters themselves, the amount of heads per platter, or a combination of all these, etc... There are many variables in effect to explain why a bigger drive is often faster than a smaller one, but the specifics are different from drive to drive.

[Astra.Xtreme]

Often larger drives are faster due to the higher data density on the platters, or the amount of platters themselves, the amount of heads per platter, or a combination of all these, etc... There are many variables in effect to explain why a bigger drive is often faster than a smaller one, but the specifics are different from drive to drive.

Exactly. Same amount of platters, same size (physical) platters, same rpm, and higher density = faster performance

[Jason S. Global Moderator]

The Force GT is definitely faster but not as instant what everyone claims SSDs are.

Installing Windows 7 takes 20 min (starting from setup screen next to drive partition) on Seagate while it takes 18 min on Corsair Force GT.

Speeds have settled to 140 mBps write and 230 mBps read which is a far cry compared to 550/550 write/read printed on the packaging.

The 550MBps ratings that you see on the specs are for sequential reads and writes. this would be like copying a giant iso or movie from a source to your SSD, or vice versa.

Just about every other I/O on your computer is going to be small-block random I/O. this would include installing and booting an OS, loading programs, etc.

when buying an SSD, the biggest thing you want to consider is the random performance, not the sequential.

[TheLegendOfMart]

Speeds have settled to 140 mBps write and 230 mBps read which is a far cry compared to 550/550 write/read printed on the packaging.

You are using SATA2 which bottlenecks your SSD, you should only get 230-270MB/s on SATA2.

You only see the speeds printed on the packaging if you are using SATA3 and only then if it is reading/writing sequential data, random reads and incompressible files (videos, music, games) are the Sandforce achilles heel.

The 550MBps ratings that you see on the specs are for sequential reads and writes. this would be like copying a giant iso or movie from a source to your SSD, or vice versa.

No, the SF-2281 is awful at reading/writing incompressible data like ISOs and Movies.

[TEX4S]

One thing I havent seen mentioned in here - with his controller/chipset - is it even SATA3 ?

Also, is he getting more than 1 PCIE lane ?

Reason I ask is, all of this ridiculous bickering about speeds would be moot if he is limited by something like that.

For instance, since I have a x58 chipset (original Core i7) I am limited to 400MB/sec due to PCI-Express bandwidth issues.

I have a Vertex 3 in my desktop & cant get better than 400MB/sec - but it matters not since I boot up in 12-13 seconds & everything is practically instantaneous anyway...

Just a thought

UPDATE:

Legend of Mart just beat me to it -

OP - All of the crap in this thread is only good for 2 things - learning a little bit about SSD & now you know you need to upgrade everything if you want to see the speeds everyone is talking about -

Kinda funny IMHO that nobody checked this earlier.

[Jason S. Global Moderator]

No, the SF-2281 is awful at reading/writing incompressible data like ISOs and Movies.

ok, but my point is still valid though - the 550MBps claims are sequential data transfers, regardless of controller. i agree that the SF controller is bad at incompressible data, but there are many other controllers...

[Jason S. Global Moderator]

One thing I havent seen mentioned in here - with his controller/chipset - is it even SATA3 ?

Also, is he getting more than 1 PCIE lane ?

Reason I ask is, all of this ridiculous bickering about speeds would be moot if he is limited by something like that.

For instance, since I have a x58 chipset (original Core i7) I am limited to 400MB/sec due to PCI-Express bandwidth issues.

I have a Vertex 3 in my desktop & cant get better than 400MB/sec - but it matters not since I boot up in 12-13 seconds & everything is practically instantaneous anyway...

Just a thought

you can get better than 400MBps b/c of the SATA2 controller... you're limited to <300MBps

[TEX4S]

you can get better than 400MBps b/c of the SATA2 controller... you're limited to <300MBps

When you say "you" are you referring to me or people in general ?

Im on a SATA3 controller, but limited to 400MB/sec for other reasons. (PCI-E)

SATA 2 limit is 375MB/sec, but any benchmark I run caps right @ 400MB/sec -

Doesnt bother me a whole lot because computer is so fast

[shiven]

When you say "you" are you referring to me or people in general ?

Im on a SATA3 controller, but limited to 400MB/sec for other reasons. (PCI-E)

SATA 2 limit is 375MB/sec, but any benchmark I run caps right @ 400MB/sec -

Doesnt bother me a whole lot because computer is so fast

Wait, I'm confused. I didn't think PCI-E had anything to do with it?

[TEX4S]

Wait, I'm confused. I didn't think PCI-E had anything to do with it?

Yep, I dont know if it is this way with AMD, and I dont know if Ivy Bridge is this way either

but I

the SATA3 Controller uses the PCI Express lanes (probably due to them both stemming from the Northbridge.

n the case of x58, this is surely the case.

It is very easy to get caught up in too many details when trying to learn the basics.

[Astra.Xtreme]

Yep, I dont know if it is this way with AMD, and I dont know if Ivy Bridge is this way either

but I

the SATA3 Controller uses the PCI Express lanes (probably due to them both stemming from the Northbridge.

n the case of x58, this is surely the case.

It is very easy to get caught up in too many details when trying to learn the basics.

I had the same problem when I had a first gen i7. I put in a Vertex 3 and then was disappointed by the benchmark. When I upgraded to a 2600K, the benchmarks were spot on with the rating of the drive. Not that it made a huge difference though. :p

[shiven]

Yep, I dont know if it is this way with AMD, and I dont know if Ivy Bridge is this way either

but I

the SATA3 Controller uses the PCI Express lanes (probably due to them both stemming from the Northbridge.

n the case of x58, this is surely the case.

It is very easy to get caught up in too many details when trying to learn the basics.

Does the video card that's using the PCIe slot make a difference? For example, newer cards use a much higher bandwidth than older ones, so older ones should have less PCIe bandwidth used, right? Or does the PCIe allocate a certain amount regardless?

[TEX4S]

shiven, good question.

It is a pre-determined amount of bandwidth that goes to the SATA controller (I believe it is simply a single, 1x lane)

Newer chips have so many things built into the CPU and these issues dont exist anymore - as Astra mentioned, SB & IB chips have use of a lot more bandwidth.

But like I said, 400MB/sec is pretty darned fast as it is

[Noir Angel]

Isn't that a bit counter-intuitive? One would think that the larger the drive, the more the drive has to read through, so the slower it would be. I guess I don't know enough about the actual process...

Nope. With SSD's larger capacity drives have both more memory chips on their PCB's and usually higher density ones as well. And the benefits that brings are twofold.

Firstly, the controllers of some (most high performance SSD's have this feature) disks have a feature that implements a sort of raid on the memory chips, files are spanned across the drive's memory chips and the more chips it has to span files across the faster reads are likely to be,

The benefit of higher density chips should be pretty obvious really.

And in fact, even with mechanical hard disks, the size of the disk isn't relevant. A 1MB file still takes up 1MB of space whether you store it on a 32GB drive or a 320GB drive. A disk's file allocation table points the operating system to the physical location of the file on the disk so the argument that it would have to search the disk is false.

[TEX4S]

Nope. With SSD's larger capacity drives have both more memory chips on their PCB's and usually higher density ones as well. And the benefits that brings are twofold.

Firstly, the controllers of some (most high performance SSD's have this feature) disks have a feature that implements a sort of raid on the memory chips, files are spanned across the drive's memory chips and the more chips it has to span files across the faster reads are likely to be,

The benefit of higher density chips should be pretty obvious really.

And in fact, even with mechanical hard disks, the size of the disk isn't relevant. A 1MB file still takes up 1MB of space whether you store it on a 32GB drive or a 320GB drive. A disk's file allocation table points the operating system to the physical location of the file on the disk so the argument that it would have to search the disk is false.

I think what he was saying was not that it has to search for the spot where data is held, its that since there is more space/data on the drive, it takes longer to go to it when directed by the OS...I think.

[Noir Angel]

I think what he was saying was not that it has to search for the spot where data is held, its that since there is more space/data on the drive, it takes longer to go to it when directed by the OS...I think.

It's still a false argument. The physical size of hard drives hasn't changed, and accessing the file allocation table does not significantly differ depending on the size of drive, the factor that impacts that the most is how full the drive is. And even then we're only looking at a difference of little more than a few ms

[shiven]

I think what he was saying was not that it has to search for the spot where data is held, its that since there is more space/data on the drive, it takes longer to go to it when directed by the OS...I think.

Yeah that's basically what I was saying

It's still a false argument. The physical size of hard drives hasn't changed, and accessing the file allocation table does not significantly differ depending on the size of drive, the factor that impacts that the most is how full the drive is. And even then we're only looking at a difference of little more than a few ms

And yeah, I figured there was something flawed with my logic. I need to learn how it actually works...

[Noir Angel]

And yeah, I figured there was something flawed with my logic. I need to learn how it actually works...

Although I'm underplaying the complexity of this a bit, the best analogy I can think of is this.

Think of the hard disk of your computer as like being one massive filing cabinet. The surface of a disk is not solid but is in fact divided up into sectors on which a small amount of data can be stored, if you consider each sector to be a seperate file in your cabinet, the whole of the disk would be equal to a cabinet with loads of files in it. And in the same way you would identify a file in your cabinet by the shelf it's on and the number on the file, the same is true of each cluster on the computer's hard drive. And as your disk fills up the clusters get more and more used.

Now think of the file allocation as like being a massive index, as the index will tell you what paperwork is stored in what file and in what drawer, the file allocation table's function is very similar, it is essentially a database of what files are stored on a disk. So when, for example a program writes a file to sectors 1101, 1102, and 1103 an entry will be created in the file allocation table to tell the operating system what sectors this file is on. As you can imagine if the file allocation table weren't there the OS would have to search the disk sector by sector every time it wanted to access the file which would be extremely cumbersome. The file allocation table allows the OS to simply spin the disk up, move the head directly to the sectors that the file allocation table says the file was written to, and read the data back in the same way the index in your file would direct you to the shelf and number of folder you wanted. And that is why performance degrades as a disk fills up, as the amount of entries in the file allocation table increases, it obviously takes marginally longer to read them all (although again the differences are very small)

Although SSD's don't have any moving parts so they don't have to spin up, the principle is the same except the data is stored in a one of millions of flash cells rather than one of millions of physical sectors.

I hope that helps you somewhat

[shiven]

Well, I got my SSD installed and running, and I'm sad to report that I'm less impressed with the benchmark than I expected. It is noticeably faster than my WD Raptor though, so definitely an improvement. I realized that since I transferred over my existing imagine (as opposed to a fresh install), the drive wouldn't have as good as results in CDM as a completely empty drive, but nonetheless, I thought they'd be better than this (see attached).

Few PC Specs:

X4 955 @ 3.4

4GB 1600MHz RAM

(SSD, obviously), which is connected through SATA III. It's the only drive connected ATM.

Thoughts on what could be going wrong here?

[Noir Angel]

Well, I got my SSD installed and running, and I'm sad to report that I'm less impressed with the benchmark than I expected. It is noticeably faster than my WD Raptor though, so definitely an improvement. I realized that since I transferred over my existing imagine (as opposed to a fresh install), the drive wouldn't have as good as results in CDM as a completely empty drive, but nonetheless, I thought they'd be better than this (see attached).

Few PC Specs:

X4 955 @ 3.4

4GB 1600MHz RAM

(SSD, obviously), which is connected through SATA III. It's the only drive connected ATM.

Thoughts on what could be going wrong here?

Don't worry about synthetic benchmarks. Even low end SSD's should still provide much better system responsiveness than mechanical hard drives so I'd just ignore what the benchmarks are saying and focus on the real world performance.

[shiven]

Don't worry about synthetic benchmarks. Even low end SSD's should still provide much better system responsiveness than mechanical hard drives so I'd just ignore what the benchmarks are saying and focus on the real world performance.

While quite true, shouldn't I at least get something near the advertised results in benchmarks? While the drive is still much faster than my mechanical one, it bugs me that it isn't even close what I thought it would be.

** And yes, most importantly: I've updated to firmware v1.4.

[shiven]

I thought I'd go ahead and post this on both threads of mine, since there are some people following this one. Installed SSD, enabled AHCI, and fixed my random keyboard problem. Here are some benchmark screenshots! :D

[NextGen_Gamer]

On Monday, I'll run through those same benchmarks on my Vertex 4 and post the results. I'll also attempt to run them both in Win7 and Win8 Release Preview, which I just installed on Friday night. It should be interesting to see any differences between the two operating systems. Right now I can say that Win8 definitely boots faster than Win7, which is pretty impressive considering how fast Win7 already boots-up.