Well I just tried the ignore ownership in the get info and the aja system test worked. What I learned is that when I want to partition a drive for multiple startup drives I should put my preferred startup configuration on the first, outermost tracks partition, because in this case the performance on the 1st outer partition is approx 60% faster than the eighth partition. Of course when I get my dream configuration leaned out I'll find a quality SSD just so Mr. Greg won't have to Zelonate me. Did I say Greg I meant Mr. Sans OKI. ;-)
--- In Avid-L2@yahoogroups.com, "johnrobmoore" <bigfish@...> wrote:
>
> Okay I broke up the raid and did individual tests. Two drives are approx 163MB/sec and one is 155MB/sec. Both write and read speeds are pretty much the same. I swapped the slower drive from bay 3 to bay 4 and it performed virtually the same at 155MB/sec. I then played around with partitioning a single drive and comparing the different partitions for disk speed. Partition 8 which I assume is the inner most tracks is the slowest at around 100MB/sec and partition 1, I assume the outer most tracks is more like 150ish MB/sec. When I put everything back to normal I repartitioned all the strip drives as individual drives HFS+ no journaling. I then dragged them into the raid box and remade the raid with 32K block size and I got read write in the 460MB/sec range, quite a difference. I then went back and did the same process but this time set the raid to 256K block size, as I had originally. This time that configuration also gave me approx 450MB/sec write and read. Why my test last night were so much slower is beyond me. Perhaps I had not set the drives to never sleep but I would think that once the diskwhack test is running all the drives would have been front and center so the energy saving settings wouldn't have an effect. Well now I feel better. These tests also explain why when I tested my 4TB startup drive it was only showing 100MB/sec write read because it was the eighth partition being the slowest. I did find that the aja system test doesn't seem to want to work on the 10.8 startup partition. I did a little research and apparently the AJA test doesn't work with 10.7 or 10.8 boot drives. There was a suggestion get info on the drive you are trying to test and check the ignore ownership on this volume box and that might get it to work. I've read the default permissions in Lion and maybe Mountain Lion can create issues for some software etc.... Well I will sleep better knowing I can safely capture virtually anything that might come my way onto this raid. Of course when one drive goes I'm screwed but one step at a time.
>
> --- In Avid-L2@yahoogroups.com, Bogdan Grigorescu <bogdan_grigorescu@> wrote:
> >
> > break the stripe and check disk speeds individually. Also, open the side panel and touch the drives, one by one, strong vibration means lower speed.
> > Hitachi drives have, in my experience, the highest failure rate - except when they are enterprise grade, so I would not be surprised if you've got a lemon out of 4.
> >
> > And you are right, an empty, freshly formatted 3way stripe should sustain 300-350MBps. Both read and write.
> >
> > cheers,
> > BG
> > www.finale.tv
> >
> > --- On Fri, 12/28/12, John Moore <bigfish@> wrote:
> >
> > From: John Moore <bigfish@>
> > Subject: [Avid-L2] 3 way internal stripped raid on MacPro Speed?
> > To: "Avid L2" <Avid-L2@yahoogroups.com>
> > Date: Friday, December 28, 2012, 1:33 AM
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Â
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > I've put 4 4TB Hitachi Deskstar 7200 RPM drives in my new 12 core MacPro. Using one drive as startup and three for the raid strip made with disk utility. AJA system test yields 153.2MB/s Write and 240.4MB/s Read. AJA data calc shows 1920x1080 1080i 10 bit YUV data rate at 165.72 MBytes/sec. If both these are accurate wouldn't that mean I would have a potential bottleneck trying to capture uncompressed 1:1 in Avid? I was also surprised to see the AJA disk whack speed test came out around 100 MB/s write and 150MB/s read on the single startup drive. I would think the three way strip should be ball park 3 times faster. Is this lower than expected performance a product of the mac disk utility software raid? Am I oversimplifying the process thinking internal drives should be faster than some external connected storage solutions? I've been told internally the macpro is sata 2 not 3 but in googleing sata 2 can yield 300MB/s. Clearly I'm
> > not
> >
> > fully clear on the data balistics of the MacPro internal drives. I have had decent performance with a 2 drive strip with DNX 220 on a MacPro SNDX 4.x. Never tried uncompressed on that config. Curious what others have done. I set the raid block size to 256K after reading a tutorial on how to set up the raid. Does that seem like an optimal block size? TIA
> >
> >
> >
> > John Moore
> >
> >
> >
> > Barking Trout Productions
> >
> >
> >
> > Studio City, CA
> >
> >
> >
> > bigfish@
> >
> >
> >
> > [Non-text portions of this message have been removed]
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > [Non-text portions of this message have been removed]
> >
>
| Reply via web post | Reply to sender | Reply to group | Start a New Topic | Messages in this topic (14) |
No comments:
Post a Comment