Forum Vs Naplam - Fasted real world storage solution

[Rhys]

In a previous post Napalm has challenged people here to see if their storage setup can surpass his in terms of performance.
He has posted fast boot times and fast benchmarks but I'm sceptical about the benchmarks due to the cache.

I was thinking if Napalm would be kind enough to participate we could set a bunch of real world tests that test (Time) the drives in the following areas:

Application/Game Loading
File Copy/Creation
Software Installation

Note:
I think it would be a good suggestion to only include the 1st loading times after the OS has booted due to data being stored in controller cache.

Some people here don't believe raid 0 helps speed up loading times, but others
do.
I'm not interested in who is right or wrong but just the performance truth.

Perhaps if we can devise a set of standardised tests we can slowly but surely establish the best performing setup, that will help others so they don't waste £'s.

[NapalmV5]

ohh bring it! please bring it! slap me silly bring it!!!

im @ 4 sec bootup and you guys cant even get under 10 seconds!!

see where its headed @ ?

[Rhys]

Lol!
Just curious, have you made any OS tweaks to achieve such a boot time?

Perhaps some x25 or Acard owners have a chance?

[Gogeta]

I'm looking for F.E.A.R. with his dual ACARD setup to show up on the starting line.

[justin.kerr]

I say we use this program. http://alex-is.de/PHP/fusion/downloads.php?cat_id=4
Use the copy-benchmark program in the tools tabs.

[Rhys]

I say we use this program. http://alex-is.de/PHP/fusion/downloads.php?cat_id=4
Use the copy-benchmark program in the tools tabs.

Yeh that looks good to test copying times.

[justin.kerr]

ok, so lets see some numbers! lol

[Rhys]

Napalm perhaps you could record a series of times loading app/games etc etc and others with the same apps can run the same test and post their findings.

The tasks can then be split into file types, and we can then analyse the results.
however this experiment is reliant on people being honest and also not timing the second attempt at loading an app as it will load from cache etc.

I can't run any tests yet as I'm waiting on an RMA'd x25-m and only have 2.5gb free on my vertex and performance is now suffering but it's not noticeable.

[XS Janus]

Napalm could you please state what storage goodies are you using

Maybe there should be a rule to first state the goodies and then the scores on the benches...

[Rhys]

Napalm could you please state what storage goodies are you using

Maybe there should be a rule to first state the goodies and then the scores on the benches...

I agree.

[justin.kerr]

Here is my wifes rig, slowest set up in the house, but at least it will start some numbers. lol
2 Intel x-25 m raid 0 on ich10r

[SteveRo]

This should be fun - napalm is so engaging!
I don't think AS SSD is the best all around bench - but it is easy.
AS SSD seems to give a lot of credit to cache and gives MFT huge scores.
Running iometer just for 1 minute shows what the drives (w/o cache can do).
Below is AS SSD with acard 4xR0 on 1680ix-12 (no MFT) and then Napalm's favorite - iometer - again acard 4xR0 on ich10r -

[SteveRo]

AS SSD for acard 4xR0 on ich10r -

[antiacid]

SteveRo, your results are just retardedly fast compared to what's in the average user's machine :P

I'm wondering how much it costs to get that performance though, care to share the final price?

[road-runner]

Here is mine, 4-30gb vertex onboard controller.

[NapalmV5]

lol wow.. get the whole flock take everything youve got and bring napalm down!!


hardware: same 1231/4x jmicron ssd raid0 here: http://www.xtremesystems.org/forums/...d.php?t=207557


you guys wanna do this or just post useless benches ?

as far as.. as ssd benchmark goes the results are mostly based on iops..

let me put it this way.. 1x x25 has more iops than my 4x ssd combined and 4x x25 combined has less access times than my 1x ssd

just one x25 should wipe the floor with my setup but @ real world apps its whole diff ball game.. you guys wanna play?!?

with thousands of apps out there.. many have built in timers.. so postem up.. post up apps you guys use day in day out..

[NapalmV5]

Lol!
Just curious, have you made any OS tweaks to achieve such a boot time?

Perhaps some x25 or Acard owners have a chance?

no os tweaks just slimmed down fully functioning xp

i could further slim it down to barely just a bootup and get it down to 3 sec maybe less.. whats the point of that.. unless you guys poise threat to the 4 sec

for those interested/want to compare.. boottimer.exe @ http://www.planetsoft.org/

[NapalmV5]

Here is my wifes rig, slowest set up in the house, but at least it will start some numbers. lol
2 Intel x-25 m raid 0 on ich10r

ive looked at all the as ssd benchmarks/benchmark it myself and just like stevero said earlier.. basically the benchmark benches your cache (ram) 0.033ms write @ 2x x25m ?? come on guys!

[Rhys]

AS SSD for acard 4xR0 on ich10r -

notice how your iso, program and game were faster on ICH10R?

[Gogeta]

This should be fun - napalm is so engaging!
I don't think AS SSD is the best all around bench - but it is easy.
AS SSD seems to give a lot of credit to cache and gives MFT huge scores.
Running iometer just for 1 minute shows what the drives (w/o cache can do).
Below is AS SSD with acard 4xR0 on 1680ix-12 (no MFT) and then Napalm's favorite - iometer - again acard 4xR0 on ich10r -

<snip>

65000 IOPS

[NapalmV5]

notice how your iso, program and game were faster on ICH10R?

all his results are acard+1680 (SAS) controller

and for goodness sake if as ssd benchmark is de facto measurement for you guys.. pity on you

[SteveRo]

notice how your iso, program and game were faster on ICH10R?

Yes, I did notice that but I also should have stated that my 1680ix benches are with the acards as sata I speeds - so that is probably affecting the timed iso, program and game benches significantly.

[SteveRo]

65000 IOPS

I wonder if 3xacard 6xR0 on ich10r would be faster than 4x or slower - more latency due to more drives in R0?

[SteveRo]

We need to have some folks with large arrays of vertex and x25-e chime in!
I am particularly interested in seeing ich10r, areca 12XX or 1680ix with 6 or more vertex or 4 or more x25-e run the AS SSD iso, program and game benches!
Nizzen - come-on come clean!

[Rhys]

all his results are acard+1680 (SAS) controller

and for goodness sake if as ssd benchmark is de facto measurement for you guys.. pity on you

Perhaps he bought the wrong card, but it is STILL interesting how the 1680 added more mb/s in the benchmark (cache) but slowed it down to about half speed in the simulated real-world tests (ISO Game App)!
Obviously this is not definitive but if this is the start of a trend, then it could mean there are allot of people spending big £ to ruin the performance of their drives.

My current 'suspicion' is that if he used the same controller as you he would get better results than the 1680, but not as good as ICH10R.
I'm kinda thinking that you are seeing better results on Areca simply due to J-micron lack of cache. i.e. I don't think it will benefit other SSD's that now come with plenty of cache.

BUT as I have said before I'm more than happy to be proved wrong, so perhaps you can time some App / game level loads.

Does your array load COD MW levels as fast as this?
http://www.youtube.com/watch?v=DWCDch4YDPk