Anvil's Storage Utilities

[Anvil]

Anvil - any luck changing the read tests to test actual data that is on the SSD instead of the test file?

Yepp, it's been working for quite some time.
What is left is making it presentable.

I'll PM you when it's ready for testing.


@B Gates

I've updated my V4's but haven't got time to play just yet, I've been close to that result though

[One_Hertz]

Excellent! Now that is the first large advantage over IOMeter

[tiro_uspsss]

could someone show me how I can use ASU to do something along the lines of what I wanted to do in this thread:

http://www.xtremesystems.org/forums/...ing&highlight=

read: I want to thrash a HDD non-stop, 24x7 for about a week straight. Basically fill the HDD up, delete, rinse & repeat so-to-speak!

[Christopher]

could someone show me how I can use ASU to do something along the lines of what I wanted to do in this thread:

http://www.xtremesystems.org/forums/...ing&highlight=

read: I want to thrash a HDD non-stop, 24x7 for about a week straight. Basically fill the HDD up, delete, rinse & repeat so-to-speak!

just use the endurance test in ASU. Go into settings and check 'keep running totals'. And change the 4K write duration to 0ms. You may need to experiment with adding time per 1000 files deleted. Lastly, you don't need the 10 second break between loops on a mechanical HDD either.

[One_Hertz]

could someone show me how I can use ASU to do something along the lines of what I wanted to do in this thread:

http://www.xtremesystems.org/forums/...ing&highlight=

read: I want to thrash a HDD non-stop, 24x7 for about a week straight. Basically fill the HDD up, delete, rinse & repeat so-to-speak!

Stress testing HDDs like this is not the best idea tbh... A prominent failure point is during spinup/spindown when the heads have to unpark/park and the drive has to do a bunch of its firmware initialization procedures. Also if you are just using it 24/7 then its temperature will be static. Changes in temperature cause a lot of internal changes in the drive (the drive automatically changes some firmware parameters based on its temperature) and can be the cause of failure as well. Also, drives do internal clean-up procedures (to look for and reallocate weak/bad sectors mostly) if given time to idle. If you don't let them idle ever then it is not the best... Just reading/writing is a part of it of course but I question how useful it is for what you are trying to do. It is more or less impossible to predict failure accurately... this is why backup systems and RAIDs exist.

[Anvil]

When running the ASU test, the pause between loops is needed (and then some)

I ran a few tests on a 2TB drive (a bad drive starting to accumulate bad sectors) and it is painfully slow, with 500,000 files per loop it takes hours? (will check) deleting the files.

It might be due to that all files are in a single folder, it is extremely slow and a few days of writing did not help accumulate more bad sectors, it looks like I have to let it die the natural way or try to get it replaced.
(it's a secondary backup drive and I've already cleaned the drive)

Temperature during ASU testing was more or less fixed in the low 40's (C). (42-43)

[Zaxx]

could someone show me how I can use ASU to do something along the lines of what I wanted to do in this thread:

http://www.xtremesystems.org/forums/...ing&highlight=

read: I want to thrash a HDD non-stop, 24x7 for about a week straight. Basically fill the HDD up, delete, rinse & repeat so-to-speak!

The most practical purpose to put a drive through 'hell week', imo, would be to check initial drive quality before commiting it a new HDD to a raid0 array (or any array I guess). Most drives fail in the first few months or after ~3 yrs or more from a story I read (no I can't remember where...lol). I've always made sure to 'break-in' my HDDs for a couple of months before putting them in raid...never tried the 'hammer test' so to speak.

[Christopher]

I've never tried running ASU endurance on a HDD, and I don't much see the point.

[B Gates]

HyperX 3K 240 RAID 0 8kb stripes:



first Ive seen over 10,000

[felix_w]

@Anvil,

Is there any case that AHCI/RAID OROM and driver version can be displayed on AMD Boards?

Thank you for the time, effort and all the updates.

[Anvil]

I'll have a look at it!

To find out, I'll probably have to re-install my AMD rig so I can't say when.

[B Gates]

Pair Force GS 240GB 64kb stripe:



the review will be up soon at http://www.rwlabs.com/

[Anvil]

Looks like I need to try those "GS" drives!

"Soon" as in today or this week

[B Gates]

Looks like I need to try those "GS" drives!

"Soon" as in today or this week

as in next week

[B Gates]

How come I cant thank anyone?

[Computurd]

pretty stanky Bill

[B Gates]

pretty stanky Bill

thanks comp

[Gavric]

Anvil, could you please to add possibility of copy all results to clipboard as text? The screenshot doesn't always comfortable, and "copy results to clipboard" function grabs only summarized scores, not read/write ratings in MB/s.

[tived]

Hi guys,
couldn't sleep so I ran these test just to see what was how my IBM M1015 was performing together with 8 Intel 520s in RAID-0 from 2-8 disks and JBOD

I will be replacing this controller, or at least moving it down the food chain, with an Areca 1882ix 24channel 4GB cache and battery backup unit, which hopefully will give me better performance.

Anyway, i thought i'd share this with you all.



Probably should do a graph ;-) naaahh just kiddin

OHH, it didn't help that Anvils program RC 2 expired in the middle of these tests, so i reset the date to a date later and was able to finish

Henrik

[Nizzen]

If you are using software raid, disable "write cache buffer flushing" on each drive in "disk properties" that you have connected to the M1015. That helped me with the performance, if I remember correctly :p

Finally got my last M3p @ LSI 9207
8xr0 software raid srv08r2

It suffers by no cache, but it is still pretty fast


ATTO: for compare

[tived]

I know I was joking but here are the results tabulated up in a google spreadsheet - haven't got MS office installed yet. figurs marked in red are the fastest, but its clear that there is not much linear about the scaling on this controller



:-)

Henrik

[tived]

If you are using software raid, disable "write cache buffer flushing" on each drive in "disk properties" that you have connected to the M1015. That helped me with the performance, if I remember correctly :p

thanks Nizzen, I only wanted to use it as a reference, so just to note, i did not do that, so bare that in mind when reading the results

I would be very interested in how I can improve the performance on this controller

thanks

Henrik

[tived]

Holy crap!!! :-)

What drives are you using... (OK, its the Plextors) I better have another play with this.... did I just spend on an Areca because I have not educated myself well enough? that is some excellent numbers

thanks - I am going to have a try tomorrow night ;-)

Henrik

If you are using software raid, disable "write cache buffer flushing" on each drive in "disk properties" that you have connected to the M1015. That helped me with the performance, if I remember correctly :p

Finally got my last M3p @ LSI 9207
8xr0 software raid srv08r2

It suffers by no cache, but it is still pretty fast


ATTO: for compare

[tived]

Hi Nizzan,

this is what I get when looking at both my M1015's ??? its disabled, and thats the settings that was used on my tests yesterday
????

What am I doing wrong, is it when I set it up in MegaRAID Storage Manager? I am using the default settings

Henrik

[mobilenvidia]

Hi Nizzan,

this is what I get when looking at both my M1015's ??? its disabled, and thats the settings that was used on my tests yesterday
????

What am I doing wrong, is it when I set it up in MegaRAID Storage Manager? I am using the default settings

Henrik

The drive cache can only be changed on the LSI controller or via MSM

Did you reboot after setting each RAID test up ?

What's your ATTO speed with R0x8 ?

Remember the M1015 is a cheap 8 port 6Gbps controller the Areca controller you are looking at will cost how many times more ?
Will you get the same performance boost.
If you want RAID0 and the M1015 let windows do the striping and the M1015 do the controlling (best in LSI9211 IT mode)