Meausuring QD in win7

[Tiltevros]

ok now for the record and for not be missunderstod Everything in my bios that it has to do with cache is disabled im reffering to this in order not to question the results

[Tiltevros]

i am up to 2-3k more I/Os @ 512B more details in a while

[Anvil]

I added 3 x X25-M's and surprise surprise

iops_qd_single_vs_raid0_9211_x2_qd32_3.png

QD 64
iops
104968,95
mbps
410,03
avg io
0,6095
max io
2,0926
%cpu
28,31

[Tiltevros]

the 9211 is caped in IO/s mate u will find out when it comes to u

[Tiltevros]

btw make charts if u have the time for 8k 16k and 32k files

[Anvil]

the 9211 is caped in IO/s mate u will find out when it comes to u

Looking forward to it anyways.

It looks like there is a flaw with everything.
9260 is great for bandwidth but not for iops.
ICH is sortof great for iops but not for bandwidth.

Well, one can't win them all.

[Tiltevros]

imagine that with the new firmware of 9211 the maxedup the cap limit from 72k to 75k
plz make the chart with 8k 16k 32k and 64 :S

[Anvil]

btw make charts if u have the time for 8k 16k and 32k files

Will do, but not tonight.

Could you zip your Excel files and make an attachment.
(I would save some time by copy/paste)

[Tiltevros]

so if i do that we shall see them tonight?? fix my night man think that i am sick with fever and hopping is not h1n1 lol

[Anvil]

so if i do that we shall see them tonight?? fix my night man think that i am sick with fever and hopping is not h1n1 lol

I'll create one for your setup the others will have to wait.

[Tiltevros]

im waiting the chart Mr Anvil :P

[Anvil]

im waiting the chart Mr Anvil :P

I'm waiting for the Excel spreadsheet
Just make it an attachment.

[Anvil]

Updated 4K random read chart.

4 x Intel X25-M 80GB G1 - raid-0, 128KB stripe size

iops_qd_single_vs_raid0_9211_x2_qd32_4.png

x25m_x4.PNG

[Ao1]

Here is a comparison of 4k random & sequential writes between a single drive and a raid array. Thanks to Anvil for the raid array figures.

[Anvil]

I need to point out that the x3 raid tests are performed without WBC. (WriteBackCache)
(WBC On would result in increased write iops, at the same time reducing the read iops)

What I found interesting is the fact that random writes performed so well compared to the single drive while sequential writes suffered at low QD's.

[Tiltevros]

OK here we go with writing speeds something is going on with my array and that is making me sad

[Tiltevros]

im so confused right now i dont know what is happening .... my random speed are higher that my sequential now ..... im hiting 1,6GB/s in random @ 64k files with 32 queue and 1,3GB/s in sequential...

[Anvil]

Tiltevros,

Reboot and take a break
At first glance your QD1-8 scores looks just fine.

[Tiltevros]

ive reboot every 1 hour but still nothing.... i dont know man till the time that i hit that record and made that sheet table everything gone wild here... dont know the smart monitor told me that i have written 101TB on my G1s do u know what is the maximum writing point??

[Anvil]

quote from Anandtech
"How Long Will Intel's SSDs Last?
...
Intel went one step further and delivered 5x what the OEMs requested. Thus Intel will guarantee that you can write 100GB of data to one of its MLC SSDs every day, for the next five years, and your data will remain intact. The drives only ship with a 3 year warranty but I suspect that there'd be some recourse if you could prove that Intel's 100GB/day promise was false."

Link to Anandtech

[Tiltevros]

wtf ?????

[Ao1]

im so confused right now i dont know what is happening .... my random speed are higher that my sequential now ..... im hiting 1,6GB/s in random @ 64k files with 32 queue and 1,3GB/s in sequential...

Tilt, storage is always a compromise so it’s a matter of finding a solution that works best for you. If you are working at high queue depths your array is great. If you are working at low queue depths it is not so great, but it can’t be all things.

Bottom line is that at low queue depths you are going to be much better off with a single ssd and therefore you would want the fastest ssd you can get hold of. Stevecs explains low qd options here

[Tiltevros]

can we see some charts with write performance?

[Ao1]

quote from Anandtech
"How Long Will Intel's SSDs Last?
...
Intel went one step further and delivered 5x what the OEMs requested. Thus Intel will guarantee that you can write 100GB of data to one of its MLC SSDs every day, for the next five years, and your data will remain intact. The drives only ship with a 3 year warranty but I suspect that there'd be some recourse if you could prove that Intel's 100GB/day promise was false."

Link to Anandtech

It's also a case of what you write as opposed to how much you write

X25-M 160GB - 100% random host data 4k writes = 15TB
X25-M 160GB - 100% sequential writes = 370TB

[Anvil]

I know audienceofone,

Hard to tell from the SMART values as they only tell host writes.
Last time I checked, one of my older M's had only written 3.5TB.