500Mb\s?

dek8

n00b
Joined
Mar 17, 2011
Messages
32
EDIT: i think i will do it like this to start with, cause my initial impression of SSD's are from running it on a bad setup, i cant find out what the problem is and the disks are fine in Windows 7. Everyone who posts now doesnt understand that i have issues with the drives on XP, and rather than trying to help me out sorting my problem they are giving me explanations as to why SSD's doesnt obtain the speeds by what they are advertized as. They are basically outputting 1/4 to 1/3 by swapping them back and forth in windows and the vertex 3 has given me the same results in benchmarks since it was brand new 4 days ago. Before you start with that its trim thats kicking in think a little about that the drive would not bench much higher in windows 7 if it was that cluttered and choked up like that. There are issues on my XP installation with them which im trying to sort out, and if you can help, or have some tips or ideas then please. Here is my uber photo shops skills and thanks... lol

I will leave the rest of the thread as it is, its still fun to read, but i think people read the first post and think that im crying about not getting 500MB/s and people start to post without realizing that i actually have a bad setup and got my impressions from that. Windows 7 is fine and the disks reflects much of the performance from the benchmarks on both of them respectively working with standard reads and writes on files through the operating systems.


The disk is unused and just benched after the windows 7 bench.

xpv7.png





Heres is where the thread started;
------------------------------------------------


What i don't understand with these SSD drives are their advertized read/write performance. When you try to do a simple task like copy a file in windows it ends up with 1/4 of the actual advertized performance and a mere 2-2.5 increase over the old mechanical HDD's. I'm not talking super small file copies.

Where is the bottleneck anyways and whats up with those numbers? 500Mb/s? Where? How?
There is probably an explanation for this with smart-ass calculus, Ive heard a few, doesn't make sense to me, somebody willing to explain?
I even run ATTO and i get numbers in the 400-550Mb/s region, both Read and Write. When do i actually get this speed when working in windows? My system can read and write it but my drive is not able to write at those speeds when im actually copying a file? whats the deal? What about actual file transfers from and to the disk? How can i see this when i work on a system? The highest i have seen in real world performance is about 78Mb/s, trying to copy a simple file from 1 drive to another and end up with speeds around 60-75Mb/s.... whats up with all these numbers? file copy is basically just doubled, tripled in some cases compared to the old HDD, that's good and all, but all the fuss and numbers about they are so super fast. I don't see it.

EDIT: lol.

The benchmarks and advertizement numbers are like 10x higher than their actual performance when u get to work on them... If i actually had the advertized speed working in windows i would have copied 720p rips in a few seconds, i end up with the 60mb/s and i sit there watching wondering what those numbers are about. Only reason I'm reasonable happy with ssd's are cause they run cold, silent and low power, and last about the speeds. And every single site i go to are comparing these drives with those benchmarks of those half a gig numbers up to each other. Whats up with that? When you actually see that the drive is on the average just performing 2-2.5 times of the old mechanical drives when you copy a file it makes one wonder what all these ads and benchmarks are about. am i bottlenecked? lol
 
Last edited:
If ATTO is telling you 500MB/sec, then that's what the drive is capable of, so you can't be bottlenecked. Why you're not seeing this in real world situations I've no idea. SSDs are not affected by the physical location of the data on the drive, in the way that HDDs are, so it can't be that. They're not so affected by OS accesses either, I'd imagine.

What's the size of the file you're trying to copy? Is it big enough to enable a sustained transfer?
 
Copying a file involves creating a new entry in the filesystem for the new file, finding the entry in the filesystem for the file to be copied, locating and translating the blocks occupied by that file, copying the data from each block to memory, finding a free spot on the drive for the copied data, writing the data to that free spot, verifying the data was written correctly, and updating the filesystem. Rinse and repeat for as many clusters as file occupies.

I don't believe benchmarks have that kind of overhead since they read/write straight from/to the drive without regard to what it's actually being transferred. Maybe I'm wrong.
 
the question am i bottlenecked is irony, i know ssd's are performing close to each other and i dont see any of the ssd's on the market are actually outputting the numbers they are advertized as. they are basically close to each other without trying to make another comparison that some drives are actually faster i fail to see that any of them performs to the numbers on their box, when it says 500mb/s, why dont i get 500mb/s from them? and whats up with those numbers?
doesnt matter what file size, anything from 2mb to 4gb, any file size, max approx. 78mb/s.

and that was the smartass calculus i was talking about ryan :) so its not really 500mb/s then? :p but it says 500mb/s on the box ryan? :)
 
Last edited:
the question am i bottlenecked is irony, i know ssd's are performing close to each other and i dont see any of the ssd's on the market are actually outputting the numbers they are advertized as. they are basically close to each other without trying to make another comparison that some drives are actually faster i fail to see that any of them performs to the numbers on their box, when it says 500mb/s, why dont i get 500mb/s from them? and whats up with those numbers?

Because 500MB/s is the max possible transfer rate for the drive. It is a measure of a single stream of data coming from the drive without interruption. When you copy a file, you're actually doing several different operations that conflict with each other and start and stop the stream of data.

and that was the smartass calculus i was talking about ryan :) so its not really 500mb/s then? :p but it says 500mb/s on the box ryan? :)

that wasn't calculus, that was just showing that copying a file isn't just reading data, so you're not going to get maximum read speeds.

EDIT: missed that you're seeing 78MB/s... that's seems kinda slow for drive to drive transfers even with filesystem overhead..
 
I see that Ferrari's are capable of 200MPH, but every time I see one driving through down town New York, it never goes that fast! Ferrari must be lying!
 
EDIT: missed that you're seeing 78MB/s... that's seems kinda slow for drive to drive transfers even with filesystem overhead..


what should it be between a vertex 2 and a vertex 3 drive approx.?
at first i had half the output in atto until i aligned the drives.. still wonder if somethings not right.
 
Last edited:
I see that Ferrari's are capable of 200MPH, but every time I see one driving through down town New York, it never goes that fast! Ferrari must be lying!

my p67 is down town new york?
what system are able to achieve the specified performance of those drives then? cause i know there are roads that proves that that car is able to go 200mph.
 
what should it be between a vertex 2 and a vertex 3 drive approx.?

I honestly don't know. But 70-80mbps is about what I was getting from my 6400AAKS drives. How old is the Vertex 2? Don't SSD's performance diminish as they're used without proper maintenance?
 
I honestly don't know. But 70-80mbps is about what I was getting from my 6400AAKS drives. How old is the Vertex 2? Don't SSD's performance diminish as they're used without proper maintenance?

i actually dont see any performance drop since the drive was new, its about 8 months i think, the vertex 3 is brand new, atto outputs twice the score on the 3 compared to the 2 but i dont see much of performance diff between them. 6gb/s or 3gb/s, doesnt matter, just that the vertex3 is bottlenecked on sata2 in atto.
 
Last edited:
my p67 is down town new york?
what system are able to achieve the specified performance of those drives then? cause i know there are roads that proves that that car is able to go 200mph.

Nope, Windows is down town New York..

And benchmarks can prove a SSD can get 500Mbp/s
 
when benchmarks tell you your drive can reach the 500MB/s mark then thats the case. Its like the ferrari comparison when its being put to the test. You can reach up to ...mph that does not mean that you can drive your car that fast in real live.
Start by measuring transfer from a medium significant faster than your ssd to the raw ssd, something like dd if=/dev/ramdisk of=/dev/ssd_without_partition_or_filesystem bs=4096 count=10000
Your numbers look like your system is fucking up your desired speeds. Is YOUR file coming from a faster drive than the ssd or is it from hd? Is your system hosted on one of these drives? Are you running eMule, torrent, antivir or other fuckup software in the background?
As long as you are not measuring with the least amount of variations possible quit whining about not reaching the THEORETICAL max of your ssd.
One last thing. a ssd does not feel that fast because it can reach up to 500MB/s, it is fast because the access times are 1/10000 to 1/1000s of that of a harddrive
 
what should it be between a vertex 2 and a vertex 3 drive approx.?

Looking around the web at large uncompressable data copy benchmarks:

Vertex 3 60GB -> Vertex 2 50GB is probably ~65MB/s.
Vertex 2 50GB -> Vertex 3 60GB is probably ~85MB/s.
 
Last edited:
when benchmarks tell you your drive can reach the 500MB/s mark then thats the case.

Markus, would you say if atto runs up to its specs then the drive is setup correctly?

and nop, running xp barebone on 17 services, boots on 140mb ram. 8gigs of cl7 ram, [email protected], intel onboard sata.

3 benchmark scenarios, vertex 2 & 3 and a ramdisk. obtainable transfers on big uncompressable files 60MB/s average to max 78. I have seen 78, but it averages around 65.
 
heres my benchmarks,

v34h.png


those copy benchmarks are basically what im talking about, they are HDD speed?
 
i think you are bottle necked. i can copy 80-100 meg/second over my gigabit network.

When i go from my raid 0 wd blacks to my agility, it is much faster then the HD. I'm not sure, but i think the sustained transfer speed is depending on how much cache your ssd has. I short stroke my SSD's by 20% to get smoother file transfers.

Also, the sandforce based ssd's dont shine unless the data is compressible. uncompressable data on sandforce doesn't really get the high speeds.
 
i saw another guy with the same drvie as me and his benchmarks on the copy was like 250MB/s on the iso test, maybe something is wrong.
 
what are your system specs?

make sure you have the ssd's plugged in the native sata ports. also put them into AHCI mode in the bios. In addition, do not install any drivers for your sata ports, use the default microsoft sata drivers.

also in windows, right click your drive properties, go into the hardware tab and disk policies and enable: "enable write caching", "turn off windows write-cache buffer". play with those 2 settings to see if things improve.

also, virus scanning could be hindering performance. check to see if your cpu is getting maxed out.
 
p67 deluxe, 2600k, running off intel sata, ahci... but i did install the intel drivers... i slipstreamed the intel rst onto my xp install to get ahci going?... where do i turn off write cache buffer?

system is running on 20 processes, basically barebone without anything in the background.

EDIT: Doesnt seem like i have a write cache buffer option on these drives.
 
Last edited:
are you using windows xp? if so that would be your problem. windows xp doesn't support trim, when you are testing data transfers and running benchmarks, no TRIM is taking place to "clean" up the drive. Your drive is just gonna get slower and slower. You should check out the OCZ forums to see how you can scrub your drives clean again that is sandforce friendly. I recall back on the barefoot days, there was an manual trim app (but dont know if it supports xp). You really should be using windows 7 for TRIM support. You can verify if this is the problem by doing a sanity erase on one of the drives and check if the performance returns to normal.

the intel rst drivers should be okay, you can delete them from your device list and it will revert back to the Microsoft default.
 
Last edited:
Your benchmarks show that you can read / write from the drives at 202 / 64MB/s and 453 / 74MB/s respectively. What would lead you to believe that you can write to a drive faster than it is capable of performing that action? The benchmarks show that your sequential speeds are to spec...
 
What OldSchool said. Since dek8 has low-capacity 50GB/60GB Sandforce-based SSDs, what he seeing is normal for uncompressed data.
 
first of all i have slipstreamed irst into the xp install cd so even if i remove the rst drivers/app from the xp install it will not revert to anything else than the rst i slipstreamed. i have tried locating MSAHCI drivers for the sata chipset without any luck, the performance i am comparing it to are using msahci which makes me believe that the rst drivers on xp are no good.

i wish somebody with vertex3's on xp could compare my benchmarks, basically seeing a score 1/5 of the 120gb version of the drive makes me think something is not right.

3rd option for me is that i have a couple of days to return the drive and get something else, i dont think the 120gb will yield that much more of a performance on xp, i think the culprit is the slipstreamed irst drivers are the reason and not the drive. i have seen benchmarks on this drive on 7 and my drive on xp is far from it in any of the benchmarks, we are talking 3-5 times slower on the copy tests.

Maybe theres a better drive for me since im sticking with xp? like the C300? not sure what to do. all i know im not happy with the performance im getting from this drive.



heres the xp vs 7, is it possible? i mean its a 60 vs 120, but 5 times the speed on the bigger drive?

v344.png
 
Last edited:
just remove your intel rst device in the device manager, reboot and it will revert back to windows.

regardless of what ssd you get, you really need to be using it on windows 7 to maximize the benefits. The c300 / m4 is the worst possible drive you can get for windows, and that drive really needs TRIM working. The sandforce drives are probably one of the better choices.
 
its just reverting back to irst.

and i thought about getting the 60gb version to raid 2 of them up and beat the 120, i wont swap to 7, if i could get msahci running and nothing would change then i will just let it be. guys over at ocz are basically saying the same things, well, 1 guy.
i even believe if i raid them on this setup i wont notice much more improvement, its like something is stuck on this speed.
 
Last edited:
its just reverting back to irst.

and i thought about getting the 60gb version to raid 2 of them up and beat the 120, i wont swap to 7, if i could get msahci running and nothing would change then i will just let it be. guys over at ocz are basically saying the same things, well, 1 guy.
i even believe if i raid them on this setup i wont notice much more improvement, its like something is stuck on this speed.


Why don't you want to switch to 7?
 
Question for you, you say you are only getting 78 Mb/s which I assume you mean 78 MB/s (megabytes), where are you copying that file from / to?

Let me put this out there in layman's terms:

-If you are copying a file from a normal HDD to your SSD, it is only going to copy as fast as your HDD can read it.... (an average consumer HDD ~ 50-100 MB/s)

-If you are copying from your SSD to your HDD, it is only going to copy as fast as your HDD can write the data... (an average consumer HDD ~ 40-80 MB/s)

-If you are copying within your own file system (your SSD to another location on your SSD) your drive will be both writing and reading at the same time, which will limit your speed (I'm assuming this decreases the copy roughly in half at least as well.. ~250MB/s, most likely lower)

-The only way you are going to see some really fast transfers is by copying data from your really quick SSD to another really quick SSD or RAID Array.


I've seen my SSD copy a file to a really fast RAID array as quick as 150 MB/s (limited by SATA controller...) and I can copy between two different RAID 10 arrays at 250 MB/s under ideal conditions.

Also, forget copying anything faster than 125 MB/s over a gigabit network anyways as it is theoretically impossible.

The point is, you are always limited by your slowest device. Think about where from and to the data is actually moving, and your speeds will most likely make sense.

The benefits of the speed of the SSD is when it loads data to memory (i.e. OS loading, game and program loading, etc.... Your quick 550 MB/s SSD may actually be loading data into memory at close to that rate, since memory is very quick, at about 6400 MB/s.. but a file copy is not actually going to happen at that speed.
 
Why don't you want to switch to 7?

Every time i get asked that i understand that the person is not able to understand whats the deal with why a user opt for xp instead of a fresh and new install of 7 with all its fancy ding dongs and gadgets which is just in the way of people who are perfectly happy with how xp is, an no matter how hard i try to explain it they dont understand it cause the preference are different and people look at computers differently. I can name a few thing, i never asked for windows 7, sure some of the features are nice but i can live without 100% of them, 7 has alot of overhead on things i will never use, background services which i know i can disable most of them, but even tweaking 7 to extreme it will not even get close to the performance of a tweak xp install. Theres like a million clicks to achieve the same type of job with a few hundred clicks in xp, when you use a computer as much as i do you want speed and you want everything to be a few clicks away with the least amount fo guidance around the gui. 7 is for the most part half guidlines and half xp, everything is hooked up in stupidity and its much like vista in that its a clumsy and alot of stuff unneded for a guy who knows how to use windows. popups and annoyances with flashy circus gui, and for the fact that the simplest things of having windows explorer windows to remember its position, size and preference is too much for the super intelligent windows 7. apart from that there is a ton of things that annoys with it, just the simple fact that you are dragging alot of dead weight just browsing windows explorer is a reason alone for me not to want to use it. I used windows 7 for over a year, tweaking and tweaking and tweaking, i thought it was good until i told myself, lets see how xp is... after that i dont ever want to go back. performance freak or whatever i am, its preference. i dont need it.
 
well. i should just have said speed, and we would get into a discussion about that 7 is really faster than xp.
 
yes butters, i took that into account, the slowest link, my as ssd benches kinda approves the max speed of my drives which at first i thought would be higher than their 50-70MB/s copy transfers. With that i understand that i wont be able to copy a file from one drive to the other at much higher rate, i thought it was higher. How i look at it now is that these drives are approx. 1.5 times of my old HDD's when it comes to big uncompressable file read/writes and im stuck with that transfer unless i go to windows 7. Seeing other peoples benches with these drives people are outputting 3-5 times my numbers, im asking myself whats the deal with this drives on xp. If i had the benchmarks of their runs i would have seen far higher real copies on my system.

I have tried a huge amount of bmp copies and it has delivered by the specifications, somewhere around 250MB/s maxing out my slowest drive, its just that the benches on the drives as they are in xp now is stuck on 1/5 of windows 7 when i look at other peoples scores.
Got every reason to suspect that its the IRST driver on xp with these drives, even read a few people having higher initial performance on xp compared to 7 with ssd's too and i dont see any other "things" that is wrong with my setup than the IRST.
 
I just skimmed through this thread, and I have to say it is hilarious.

But to be serious for a minute, there is nothing odd here. All the Sandforce-controlled SSDs I have seen always advertise much higher write speeds than are actually achieved. For a 60GB Sandforce drive, 70-80MB/s write speed for incompressible data is par for the course.

If you want an honestly advertised SSD, go for a Samsung 470, Crucial m4, or Intel 510.
 
I just skimmed through this thread, and I have to say it is hilarious.

But to be serious for a minute, there is nothing odd here. All the Sandforce-controlled SSDs I have seen always advertise much higher write speeds than are actually achieved. For a 60GB Sandforce drive, 70-80MB/s write speed for incompressible data is par for the course.

If you want an honestly advertised SSD, go for a Samsung 470, Crucial m4, or Intel 510.

john, the 120gb vertex3 is doing 250MB/s writes on incompressible data so i dont know what youre talking about. the discussion is really that xp has 1\4 perfomance to 7 tho.
 
Last edited:
Umm look for the Thread I made under Data Storage Systems (this forum) and look for Corsair released 1.3 firmware...

I have screenshots of my Corsair Force 3 SSD speeds. I also notice that SSD slow down when they have data on them as opposed to benching them while new and empty.

I think the more data that fills the nand the slower they get due to having to sort through data to get what pieces of information you need therefore making benchmarking a little slower than when they are empty. This is just opinion really.
 
When SSDs have data on them they slow down. Please see the last two SSD articles on the main site. I made a test that shows performance at 25%, 50% and 75% of capacity.

Also, the SSD might be able to write 500MB/s with data that is no compressed already but SandForce drives slow down when writing data that is already compressed, MP3, most videos and such.

The last thing to look at is your source. If you are only reading at 100MB/s from a HDD then you can't write at 500MB/s to the SSD.
 
dek8, you do not have TRIM support on windows xp. The more you test, copy files, and benchmark you ssd, the slower it will get. It will eventually crawl and the performance will not return until you do a full erase and format the drive. You can get away with a sandforce drive on linux/mac without TRIM support, but on windows it just has too much disk usage.
 
john, the 120gb vertex3 is doing 250MB/s writes on incompressible data so i dont know what youre talking about. the discussion is really that xp has 1\4 perfomance to 7 tho.

First, no, it does not.

And second, you showed a benchmark for a 60GB vertex 3, not a 120GB.
 
Back
Top