Thunderbolt storage real-world comparison

[VirtualRain]

Hey, all:
Got my TB2 upgrade card for the Sonnet Echo Express III-D enclosure and spent all of 3 minutes installing it.

Attached are the before-and-after results with a 2x840 EVO 1TB RAID0 on a Tempo Pro SSD card.

In brief (via QuickBench):
* small sequential reads/writes went from 362/319 to 433/349
* small randoms: 324/340 to 368/358
* large: 796/735 to 992/750
* extended: 840/768 to 1033/785

Blackmagic:
From 814/737 to 973/747

Attached are the detailed QB results (TB1 in this post, TB2 in the next).

That's awesome... 1GB/s!

----------

i thought the same but have been unable to access them in any way now.

That sucks... 7 dead HDs?!

I would contact Promise and see what they have to say for themselves.

 

[analog guy]

didi some testing with the pegasus2 on a tb2 connection. the results were a little puzzling. there must be some sort of caching going on with the HW RAID.

attached are tests results for a 4-drive RAID0 (4xWD Black drives, nothing special) -- hardware RAID vs (slower) software RAID for comparison.

1200MB/s? seems odd.

 

[analog guy]

last 2 quick bench results for the HW vs SW RAID.

 

[analog guy]

anyone know anything about caching on the pegasus2 series?

those results seem crazy...way more than 4x the max speed of one WD Black HDD. seems unbelievable.

 

[iBug2]

I have the Caldigit. Speeds are great, but the noise is not that low. I want a totally silent desk so I put the T3 on the bookshelf behind me, it's still out in the open but I don't hear it this way. If it sits right in front of you near the monitor, then you can hear the fan easily.

 

[Mr. Retrofire]

anyone know anything about caching on the pegasus2 series?

those results seem crazy...way more than 4x the max speed of one WD Black HDD. seems unbelievable.

Has nothing to do with the Pegasus.

Please disable the (file system) caching and the async IO options. Async IO uses at least 2 RAM buffers (or 4, 8, 16 buffers and so on).

The number of cycles is also too low. 10 or more is probably better. The file size should be larger than the built-in RAM (if possible). The read/write delays should be very low (2 ms, for example), so that the HDD has no chance to read/write data from/to the HDD cache (64 MB for WD Black HDDs).

 

[FredT2]

I have the Caldigit. Speeds are great, but the noise is not that low. I want a totally silent desk so I put the T3 on the bookshelf behind me, it's still out in the open but I don't hear it this way. If it sits right in front of you near the monitor, then you can hear the fan easily.

Ah man, that's too bad. I really wanted it to be super quiet. How far away from you on the bookshelf is it that you can't hear it? Just how loud is it when on the desk, and how would you characterize the sound that it makes?

 

[propower]

I know the BM Multidock is not the form factor of choice here (though perfect for me)... I find the results from the BM surprisingly close to the others represented here. One of the only truly silent options ....

from the only RAID testing I did... and never with matched drives!
I did a couple of quick RAID tests before loading up the drives....
Not same size but it worked for testing
(EVO 1TB + 840 Pro 512GB)

RAID 0 - striped was symmetrical 700/700 MB/s W/R
RAID 1 - mirror was 350/700 MB/s W/R

I also have QuckBench results for single drive performance... For what I do - single drive speed is plenty - no need for RAID.

 

[danny_w]

I have the Caldigit. Speeds are great, but the noise is not that low. I want a totally silent desk so I put the T3 on the bookshelf behind me, it's still out in the open but I don't hear it this way. If it sits right in front of you near the monitor, then you can hear the fan easily.

You are correct it is not silent but it is much less annoying (to me) than the higher rpm and pitch of the OWC Mercury Dual. I am keeping the T3 for sure.

 

[analog guy]

I know the BM Multidock is not the form factor of choice here (though perfect for me)... I find the results from the BM surprisingly close to the others represented here. One of the only truly silent options ....

from the only RAID testing I did... and never with matched drives!
I did a couple of quick RAID tests before loading up the drives....
Not same size but it worked for testing
(EVO 1TB + 840 Pro 512GB)

RAID 0 - striped was symmetrical 700/700 MB/s W/R
RAID 1 - mirror was 350/700 MB/s W/R

I also have QuckBench results for single drive performance... For what I do - single drive speed is plenty - no need for RAID.

2 drives in your RAIDs?

how is RAID1, a mirror, getting 700 read?

something is going on, or i'm missing something.

QB reported speeds of 1200MB/s for a 4xHDD (WD Black) RAID0 recently, which just really isn't possible.

 

[Mr. Retrofire]

something is going on, or i'm missing something.

QB reported speeds of 1200MB/s for a 4xHDD (WD Black) RAID0 recently, which just really isn't possible.

Read post #31! Thanks!

 

[analog guy]

Read post #31! Thanks!

thanks--see below.

Has nothing to do with the Pegasus.

Please disable the (file system) caching and the async IO options. Async IO uses at least 2 RAM buffers (or 4, 8, 16 buffers and so on).

this can't be it. i ran it with and without those options checked and there were no large differences. i posted the results with the options checked, but tests were run both ways.

The number of cycles is also too low. 10 or more is probably better. The file size should be larger than the built-in RAM (if possible). The read/write delays should be very low (2 ms, for example), so that the HDD has no chance to read/write data from/to the HDD cache (64 MB for WD Black HDDs).

i did run tests where the test file exceeded RAM (used QB & blackmagic).

something else seems to be going on. 90MB file written at 1306 MB/s?

does not compute.

SW RAID numbers make sense (e.g., 90MB file @ 540MB/s for 4 drives in RAID0).

???