2010 mac pro 3.2Ghz results and 40 gig OWC SSD

[Honumaui]

OK the 3.2Ghz arrived today put in 24 gigs from transintl
stock 5770 graphics card and for PS test Open Gl turned off since its quicker that way ! sadly
kept the OS on the 1TB drive that was in it (WD black)

I also picked up 3 of the 40 gig SSD to go along with 2 of the 100 RE I also have ?
I only had one 40 gig in as PS scratch though installed off the extra connector in the optical bay ! this was a quick burning me up looking at that box sitting their test so here are my findings so far !

first the SSD the 40 gig using aja basic settings on 128 MB and 4 gig test files
I am getting about 225 MB/s both ways
my 100s tend to get about 250 MB/s both ways
so this is good news cause I had heard they were slower I am happy again

really quick I noticed www.macperformanceguide.com has the 3.2 up now also so good to go check that out
my PS times were the same as his using his testing ! I have not tested LR yet only PS !
the ones that count
speed1 - 23 seconds
small - 9 seconds
Medium 16.8

the retouch artist was 13 seconds avg

playing around with some 300 meg files which are not totally uncommon for us ? the memory usage was around 19 on one meaning their was 5 free !
and around 16.5 on another meaning 7.5 free
so for me working with 300 meg files having 24 gigs is going to be the way to go

I have not done geekbench yet ? but will later if anyone cares ?
me I care more about real world what I work in but I will do it just low on priority

 

[Honumaui]

more later but quickly did a test of 3 40 gig in raid 0

W is Write
R is read
test done with AJA basic settings used 2 GB file
did each test 3 times thats the numbers across with the - in between
so 1st run is 1st column 2nd run 2nd column 3rd is 3rd

single:
W 227.7 - 232.1 - 232
R 230.5 - 239.8 - 234.8

2 dosc raid 0
W 388.2 - 386 - 383.7
R 406.5 - 400.2 - 401.2

3 disc raid 0
W 466.7 - 466.6 - 471.9
R 565.7 - 563.6 - 563.5

 

[2contagious]

more later but quickly did a test of 3 40 gig in raid 0

W is Write
R is read
test done with AJA basic settings used 2 GB file
did each test 3 times thats the numbers across with the - in between
so 1st run is 1st column 2nd run 2nd column 3rd is 3rd

single:
W 227.7 - 232.1 - 232
R 230.5 - 239.8 - 234.8

2 dosc raid 0
W 388.2 - 386 - 383.7
R 406.5 - 400.2 - 401.2

3 disc raid 0
W 466.7 - 466.6 - 471.9
R 565.7 - 563.6 - 563.5

Hi,
so you bought 3x 40GB SSDs to use them as scratch drives in raid 0?

 

[Honumaui]

Hi,
so you bought 3x 40GB SSDs to use them as scratch drives in raid 0?

yeah bought 3 and have 2 100 RE also
not sure yet what I am going to do with them ?
my original thought was to test also with LR ?
going to also test them out on the areca and decide best setup

original thought was one for the layout computer and 2 for the PS computer wife and I work together she does layout I do color
we both get a 100 RE for boot

but for sure going to run some tests first and will decide and of course will be posting info here when I do

 

[2contagious]

yeah bought 3 and have 2 100 RE also
not sure yet what I am going to do with them ?
my original thought was to test also with LR ?
going to also test them out on the areca and decide best setup

original thought was one for the layout computer and 2 for the PS computer wife and I work together she does layout I do color
we both get a 100 RE for boot

but for sure going to run some tests first and will decide and of course will be posting info here when I do

The 40GB models were not made for raid use, no? I thought only the RE versions of OWCs SSDs were supposed to be used for Raid setups because they have more error correction? Will using the 40GB in raid-0 hurt the SSD?

 

[nanofrog]

The 40GB models were not made for raid use, no? I thought only the RE versions of OWCs SSDs were supposed to be used for Raid setups because they have more error correction? Will using the 40GB in raid-0 hurt the SSD?

No, they're not designed for it.

But the real problem is the Flash used in SSD's (applies to the RE versions as well). MLC typically only has a 10,000 write cycle limit by the manufacturers (there's some out with less, some more; as low as 4k, and up to 30k writes from currently available chips). SLC is better suited (100,000 typically, and some is capable of 300k). But it's also much more expensive. Wear leveling helps the overall lifespan of the drive, but it doesn't actually extend the write cycle limit of the cells (it's a rotation scheme that spreads the writes out evenly amongst available cells to prolong the lifespan of the disk).

For example, the Intel X25-E is SLC based, and it's by no means cheap ($700 for 64GB from newegg; they're out of stock, but it's an idea of what it can run).

Honumaui is aware of this, and is willing to be a guinea pig (accept a Mean Time Between Replacement frequency of a year or so) to obtain better workflow (earns a living with it).

Assuming the MTBR is right at a year, that $300 a year can be a worthy investment if it allows for more jobs to be taken on in that year which results in more profit than the cost of the drives.

 

[Honumaui]

The 40GB models were not made for raid use, no? I thought only the RE versions of OWCs SSDs were supposed to be used for Raid setups because they have more error correction? Will using the 40GB in raid-0 hurt the SSD?

nanofrog hit it with my situation

the other are more optimal maybe a lot of yes no discussions I am not saying they are just saying many say yes they are OK and others say no ?
also I think I am just going to use two and one will go in the wifes
my thoughts (as stated above) for $200 on my end I am willing to test it out

my cache which is mostly write the files once on load and then read them off as need so its really like moving a photo over and reading it then when done its gone ?
the scratch side ? thats a who knows but willing to try a lot of discussion of the RE are worth it ? figure with both I can have first hand experience over the next year ?

 

[jwestpro]

No, they're not designed for it.

Assuming the MTBR is right at a year, that $300 a year can be a worthy investment if it allows for more jobs to be taken on in that year which results in more profit than the cost of the drives.

to remedy this is it not possible to "refresh" the ssd?

as I understand it, which may be wrong, a hdd can be refreshed and completely erased if you want to start over. i guess it's only limitation is any moving part in terms of failure.

 

[Honumaui]

to remedy this is it not possible to "refresh" the ssd?

as I understand it, which may be wrong, a hdd can be refreshed and completely erased if you want to start over. i guess it's only limitation is any moving part in terms of failure.

you can refresh it but you can not take away how many times it can write to

a engine can only go so long before a rebuild ? your refresh is like a tune up ? it helps bring it back but eventually it wont ?

think of it as a light bulb ? only so many times it can go on and off

I am dreading that day with the SSD but we shall see will be interesting


now part of me does not buy it heeheheheh and thinks they are not as fragile as they are made out to be ? but again thats the skeptic in me and I am wrong on stuff all the time ask the wife

 

[nanofrog]

as I understand it, which may be wrong, a hdd can be refreshed and completely erased if you want to start over. i guess it's only limitation is any moving part in terms of failure.

Flash NAND has a limited number of write cycles specified by the manufacturer (thier stated numbers are worst case, but even those that will be able to exceed that number, are still finite).

What this means is, you have a limited number of writes (physics, so you can't get around this). A refresh actually causes a little bit of wear (not to be concerned with, unless it's done often), but it cannot "wind back" the write cycle limit. What it can do, is get performance back up to what it was when new.

How the number of write cycles will affect you, depends on your usage pattern. For say OS/applications usage, mosts of the disk operation will be spent performing reads, which don't have a published/fixed limit. Just writes.

But for say scratch space, this is an issue in terms of how long the disk will last. To me, $100 or so for disk = disposable. Now that OWC has their 40GB disk in that range ($110 yesterday, maybe day before when I last checked), you can get a disk that's 2x or so faster than mechanical for the same money.

Granted, it may need to be tossed in as little as a year under high write conditions, but if you're making a living with the system, that's acceptable IMO. The increase in production should more than make up for it = increase in net profit (earns you more profit as you can complete more jobs in a fixed period of time).

 

[johnnymg]

Is a two drive RAID 0 (SW enabled) for boot/OS/apps generally considered reliable for OSX 10.6.4?
edit: I'm considering going with the 40 or 60 OWC's in a RAID 0 boot/app drive.

I've run a RAID 0 data drive for years but I have had some problems (Win) using a RAID 0 boot drive.

thanks
JohnG