Pegasus2 R4 vs R6 speed

[krell100]

Shockingly slow speeds. I wrote promise support and they told me it would reach around 800mb/s..

Stripe size of 128 responsible?

 

[chfilm]

Hm it might be a problem but I can't imagine it having this much of an impact? Maybe 100mb/s difference..

 

[joema2]

Shockingly slow speeds. I wrote promise support and they told me it would reach around 800mb/s..

This more likely reflects the characteristics of AJA System Test, not the P2 R4 capability.

The fact that each different benchmark program often produces radically different numbers on the same hardware indicates the source of variability. They can vary in number of threads used, I/O size, number of overlapped (async) I/Os, what caching request flags the API submits, etc.

This makes it difficult for users to compare disk benchmarks -- you have to run the exact same test, configured the same way. Oftentimes tests (such as AJA) have many different configuration parameters. So besides the parameters you typically cannot change (such as # of async I/Os and I/O size) there may be user-adjustable parameters you *can* change, such as file size. Sometimes it's unclear what the user-adjustable parameters even do.

Even running the same test on different machines means nothing unless that test is configured identically on both machines. You cannot compare the results of one test like AJA to a different test like BlackMagic or QuickBench.

E.g, on my 8TB P1 R4 with 4x2TB Toshiba drives in RAID 5, QuickBench Large gives a read rate of 830MB/sec, yet the AJA "Read Existing File" test on a 5GB file only gives about 430MB/sec -- on the exact same hardware, run within 60 sec of each other. The P4 didn't suddenly get slow or malfunction -- the difference is in the benchmark programs.

 

[mikepj]

Shockingly slow speeds. I wrote promise support and they told me it would reach around 800mb/s..

You're not going to get 800MB/sec out of the R4 unless you are hitting data in the cache. Even in a RAID 0 setup, each drive would need to be able to transfer 200MB/sec to reach that high of a rate, and most 7200 rpm hard drives I've seen are lucky if they can sustain 150MB/sec.

 

[Ifti]

Hey,

sure, good question!

Just take a quick look at the actual speeds of both systems. As you can see, the drobo is a hell of a lot slower than the Pegasus!

In real world, it's just impossible to playback a 1080 exr filesequence from the drobo stutterfree. Editing from the pegasus is suuuper smooth. I can even play back red files at 1/2 debayer setting in real time on my late 2012 imac!

Lightroom and Aperture lib are both on the pegasus as well, just so much faster scrolling through the images than on drobo.

The drobo is just hosting my itunes library and old projects and stuff. It's basically my archive. It's like a huge, expandable USB 3 disk with redundancy to me. I wish I could sell both and just get a Pegasus2 R8 for the new mac pro. But no one wants the drobo. And the pegasus is fast enough for most cases.

Just look at the table in the speedtest. You can clearly see what the drobo will be able to playback and what not.

Great - now I have to start saving all over again!! lol

I use my Drobo 5D like a large storage device, like yourself. I just need somewhere to archive all my completed FCPX projects, and as a general backup disk for my server in additional to my Time Capsule Backups.

I currently use a 1TB Buffalo MiniStation ThunderBolt drive as a scratch disk for FCPX.

I was thinking of keeping the Drobo and replacing the Buffalo drive with a Lacie 1TB ThunderBolt 2 drive (the PCIe ones just announced) since it would make a fantastic scratch/render disk for FCPX.

Although the sensible side of me says sell the lot and just get a Pegasus!!

 

[spaz8]

How much of a performance hit is the Write Thru.. setting ? vs. Write back.

I'm curious if its like 150 mb/s penalty?

I just setup my Pegasus2 R4 lastnight night in RAID5 .. 4 x 2TB - Hitachi 7K4000 - HGST HUS724020ALE640

Stripe Size: 256KB
Sector Size: 512 bytes
Read Cache Mode: Read Ahead
Write Cache Mode: Write Thru

Updated the firmware to 5.04.000.18

Plugged into a nMP 6 core over TB2. Only one monitor hooked up atm, on TB bus 1, P2 r4 is on TB bus 2 AFAIK.

Tested with Blackmagic and AJA..

I'm getting about 195 MB/s write, and 440 MB/s read.

 

[pprior]

Any updates,

I'm ordering a nMP and still not much out there for TB2 raid enclosures.

I've got an 8 disk external mini-SAS raid box for my current MP.

Thinking the 4 bay empty and putting 4TB drives raid6

 

[joema2]

How much of a performance hit is the Write Thru.. setting ? vs. Write back....Tested with Blackmagic and AJA..

I'm getting about 195 MB/s write, and 440 MB/s read.

On my R4 Thunderbolt 1 with 4 x 2TB Toshiba DT01ACA2 drives in RAID 5 (512KB stripe size) connected to a 2013 iMac 3.5Ghz, I get about 170 MB/sec write, 480 MB/sec read with Write Thru on BlackMagic. With Write Back, I get about 550 MB/sec write, 480 MB/sec read.

RAID 5 has a significant write penalty unless write back cache is used.

 

[pprior]

I ended up ordering the Pegasus2 R6 12 TB unit (6 x 2TB drives).

I configured it as RAID6 (8TB useable).

Getting 577mb/sec writes and 685mb/sec reads via Blackmagicdesign test.

unit is almost the same height of the nMP and similar color - looks very nice sitting next to it. About the same noise level as my old mac pro.

Wish I could have gotten it empty and put 3 or 4TB drives in it, but needed it quickly and as a shout out to Beach Audio via Amazon - it came next day!

Overall at this point I'm very happy.

 

[Alan Mac]

Just received my R6 (12TB Raid 5 with 1MB Stripe straight out of the box)
After synchronisation ( 6hrs) I ran a BM disk speed and Tested with Blackmagic .

I'm getting about 744 MB/s write, and 709 MB/s read.
So, happy with that - I read on here earlier that (512MB) MY BAD! 512kb stripe was best but will give this a go for a few days and see how it goes.

 

[Alan Mac]

Well I have been editing in FCPX today and it was great "to start with" - no spinning beach balls when I jumped around the edit and rendering was basically non existent.

This with basic HDV 35Mbps footage - but this afternoon I began to get the beach balls back and it was sluggish. Created proxy and turned it on and no difference.

---Other factors
late 2013MBP 2.3/16GB RAM/2GBVRAM/500GB SSD (300GB free)

Any ideas?

----------------------
Update:

Well been editing all day today without problems