First IOXTREME Review

[lowfat]

cmon lowfat, we are waiting here....

sorry installing new mainboard and reinstalling windows takes time

[Computurd]

its fine, i am just impatient

[Jor3lBR]

sorry installing new mainboard and reinstalling windows takes time

Hey Low, it only works on X64 bit OSes, right?

[lowfat]

Hey Low, it only works on X64 bit OSes, right?

No idea. I haven't used a 32-bit in ~ 4 years.

[Tiltevros]

Quick run of the fulltest.icf from the OCZ Forums. 64 outstanding IOs.

Damn near identical to 4 x X25-Ms on ICH10R. However the above bench was w/ a stock CPU. The run I did w/ the X25-M's was @ 4.2GHz. I can't run higher clocked ones until I have the time to throw a new motherboard in (sitting on my desk).

I don't have a whole lot of time to bench today so far. As I am just on my lunch break and likely will be working for the remainder of the day.

ok im going to forget the maths that 16 years im learning.....
1 IO/s is 4kb/s
20094,53 is 257,60Mb/s????????????

lol ok just do the equation right..... if 1 OI/s is 4kb/s then 20094,53 IO/s
how many Mb/s are???
simple..... x= 20094,53 * 4 = 80378,12.... kb/s then?????????????????????????????????????????????? ?????????????????????????????????????? how the heck got 257Mb/s?

[lowfat]

ok im going to forget the maths that 16 years im learning.....
1 IO/s is 4kb/s
20094,53 is 257,60Mb/s????????????

lol ok just do the equation right..... if 1 OI/s is 4kb/s then 20094,53 IO/s
how many Mb/s are???
simple..... x= 20094,53 * 4 = 80378,12.... kb/s then?????????????????????????????????????????????? ?????????????????????????????????????? how the heck got 257Mb/s?

take it up w/ the iometer programmers

[Computurd]

LOL i think he is trying to say that the numbers dont add up...that was a 4k random write test? because that profile 'fulltest.icf' that was posted is 4k random write test. i was wondering the same thing, is that the test you ran?

[lowfat]

LOL i think he is trying to say that the numbers dont add up...that was a 4k random write test? because that profile 'fulltest.icf' that was posted is 4k random write test. i was wondering the same thing, is that the test you ran?

Ya it was the test I ran. Downloaded it this thread.
http://www.ocztechnologyforum.com/fo...hlight=iometer

[Computurd]

wow impressive results! cant wait to see a 4k random read or just sequential of anything lol...any feedback on access times, could ya tease us with that? a pcie device it should be spectacular

[alfaunits]

The default test for that iCF is "All-in-one" - all possible 4-64KB, 0-100% random, read/write 0-100% dispersiion. Check it.. it is not 4K random write test.

[Computurd]

@alfa-thanks! i had a feeling i was missing something, that is why i kept asking., i was running the 4k random write file on there!

[lowfat]

Got everything installed. Will run some gaming/photoshop tonight if work isn't too busy.

[Anvil]

lowfat

Could you please run CDM 3.0, it includes 4K QD32.
(CDM 3 it's still in beta but you should be able to find it)

Link to post on forum where you can safely download the 3.0 Technical Preview.

[One_Hertz]

Got everything installed. Will run some gaming/photoshop tonight if work isn't too busy.

holy crap why are those 4ks so high? This thing has no sizeable cache AFAIK. Does it's driver use RAM for caching?

[F.E.A.R.]

Hm, why is seq. write so bad?

Now IOmeter 4k 100% random 100% r/w 64 threads
and FC-Test (create, read, copy) with my big-pattern.
Please not less than 3 runs per create/read/copy (for exactly results)

example: (the first 9 results 2x MTRON Mobi with ARC-1210 - the rest Solidata K5)




CrystalDiskMark 3.0 Beta 2 http://release.crystaldew.info/CrystalDiskMarkSetupBeta @ Low - please only 1000MB or 2000MB filesize

How much did you pay?
And how much are X25-Ms in Can or USA?

[lowfat]

holy crap why are those 4ks so high? This thing has no sizeable cache AFAIK. Does it's driver use RAM for caching?

Yes it uses ram. The amount used depends on the formatting block size.

[lowfat]

Hm, why is seq. write so bad?

The card is only capable of 300 MB/s writes

[F.E.A.R.]

Ok, more to the point is read

Please read #190 once again

[lowfat]

Hm, why is seq. write so bad?

Now IOmeter 4k 100% random 100% r/w 64 threads

CrystalDiskMark 3.0 Beta 2 http://release.crystaldew.info/CrystalDiskMarkSetupBeta @ Low - please only 1000MB or 2000MB filesize

How much did you pay?
And how much are X25-Ms in Can or USA?

how do i setup an Iometer run for 4k 100% random 100% r/w 64 but not use the entire drive? I just want to create like a 5GB test file or something.

After shipping and conversion it was around $1050 CAD. An X25-M G2 sells around $280 or so.

[F.E.A.R.]

I create a pattern with 5GB testfile for you
Wait a moment...


edit:

2 patterns (4k - 100% random - 100% write/read - 10 minutes - 64 threads - 5GB filesize) http://filestore.to/?d=74677AED6

Please click your username at "Topology" and choose your favoured partition in "Targets". Don´t change other parameters!!!

[Tiltevros]

take it up w/ the iometer programmers

lowfat u r testing 8k file...

[lowfat]

I create a pattern with 5GB testfile for you
Wait a moment...


edit:

2 patterns (4k - 100% random - 100% write/read - 10 minutes - 64 threads - 5GB filesize) http://filestore.to/?d=74677AED6

Please click your username at "Topology" and choose your favoured partition in "Targets". Don´t change other parameters!!!

Here is the read test. Will do write tonight

[GullLars]

May I suggest an IOmeter setup to test this card?

100% read 100% random 4KB, 5 sec ramp-up to avoid latency spike at start, 1-2GB test area, 1 min runtime. 1 worker, and following queue depths:
1, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 24, 28, 32, 64, 128, 256.
This will provide data for making a depth analysis of how the card handles random read at different queue depths. I have done simelar setups to analyze other SSDs, and it gives good graphs.

If it's not too much work, the same setup with 100% write would also give great info.

If you still feel up to more benching after that, a run with 66% read will how how the card handles mixed read/write .

I quote myself from the thread Storage > LSI 9211-8i: (partial quote)

As an example, here is Test 1 with almost the same parameters done on an x25-M connected with eSATA to a laptop. I know this is a bit different from what we will be testing, but the analysis method is the same and should bring usefull data and graphs. Benching data provided by Anvil, I've crunched the numbers.
Link to benchmark screenshots. (click spoilers to see screenshots)
Links to graphs generated from data:
IOPS by QD
Average accesstime by QD
Max accesstime by QD (this one went bad because of eSATA and craptop)
Snapshot of the spreadsheet used for first 3 graphs
And then the complicated and even more interresting stuff:
IOPS/average accesstime by QD
IOPS vs IOPS/accesstime by QD (2 competing graphs)
IOPS + IOPS/accesstime by QD (2 stacked graphs)
And finally link to excell 2007 spreadsheet with all raw data typed in, calculations (a few notes) and graphs. (XS wouldn't allow me to upload .xlsx, so it put it on a fileshare service)
Personally i love the last graph here with IOPS+IOPS/accesstime, as it depends on both high IOPS with simultaneously low accesstime for high scores. I bet ioDrive and ioXtreme will own at this particular graph, as they are designed for low latency and with massive parallell design.

So like i said in the other thread, I think ioXtreme will do great at the IOPS+IOPS/[average accesstime]. The reason for making a graph with those parameters is it weighs heavy on accesstime, wich is the great strenght of SSDs, and also level of parallell architecture and controller technology to allow scaling IOPS whitout gaining too much latency. But it also rewards high IOPS at high queues even with higher latency, wich the IOPS/[average accesstime] doesn't, and I think most of us agree that high IOPS throughput at high QD is valued despite higher accesstime, though lower acesstime is better.

[Computurd]

@gullars youshould make a profile to do the testing that you need then we could just show you the results on the spreadsheet i.o meter creates!
oh wait i guess you cant put that all in one profile!

[alfaunits]

Yes it uses ram. The amount used depends on the formatting block size.

That's not caching, that's for MFT-like record keeping. (MFT as in Easy Co's MFT for SSDs not NTFS MFT)