Thursday, November 26, 2020

Re: [Avid-L2] Disk Speed limited to 300 Mbs on dock that supports Sata III?

By our experience the Allegro 3.0 USB cards are helpful in only some limited ways.

We found the card only reliably working in 10.8.5. We tried the card/s in 10.10.x(latest), 10.11.6 and 10.15.6 but were less than impressed, even figured some system freezes due to this card. However, we haven't found a liable replacement yet.

So, were are transferring materials from USB 3.0 through a MacMini attached to our FC network (via Atto TB1 — 8GB/FC) with a much better speed. That is: not eth but FC since the mini is limited to 1G eth (err: was limited until the latest refurb).

Topography is MacMini (2018), several MacPro4.1 / 5.1 up to MacPro2019. Cards are Atto FC 16GB and MacMini runs on Atto 8GB FC, one additional MacPro 4.1 runs on 10.8.5 with the Allegro.

By our experience Allegro in MacPro OS > 10.8.5 and even in the current MacPro 2019: is a disaster.

@Sonnet: you read me?


my 2¢



Jo


  

On 26. Nov 2020, at 16:10, John Moore <bigfish@pacbell.net> wrote:

I was thinking that but thought I was bypassing that by going through the expansion chassis.  Now you point out what I hadn't considered.  The expansion chassis is connected through the MacPro PCI 2.0 16 lane slot so I thought it would have tons of bandwidth potential but I hadn't thought about if my expansion chassis being PCI 2.0.  Even though the backplane has 16 lanes the Allegro USB 3 card is probably sitting in a PCI 2.0 slot.

Here's specs I found:
Summary of PCI Express Interface Parameters:
Data Rate: PCIe 3.0 = 1000MB/s, PCIe 2.0 = 500MB/s, PCIe 1.1 = 250MB/s. Total Bandwidth: (x16 link): PCIe 3.0 = 32GB/s, PCIe 2.0 = 16GB/s, PCIe 1.1 = 8GB/s.

From this it looks like with the 16 lane and 2.0 I should have 16GB/sec and PCIe 2.0 is 500MB/sec.  So it would seem the food chain feeding up to the dock should handle at least something closer to 500 MB/sec.  From this it still  makes me think the dock itself may support 6G but could be built with SATA II chipset.  Given I got it off the shelf at Fry's I wouldn't be surprised.  The spec said it supports 6G but does that really mean it has 6G technology components or is it just limiting a 6G drive to 3G SATA II?


On Wed, Nov 25, 2020 at 08:37 PM, <pale.edit@gmail.com> wrote:
I think you are maxing out the PCI 2.0 slots in the old Mac Pro.   In a PCI 3.0 system you will get better performance.

On Wed, Nov 25, 2020 at 11:17 PM John Moore <bigfish@pacbell.net> wrote:
I've got a pair of Seagate IronWolf SSDs that are Sata III and spec says they can get approx 550 Mbps in Sata III.  The dock(s) are connected to a Sonnet Allegro PCIe card with 4 USB 3.0 connections.  I can only seem to get around 300 MB/sec per individual drive but when I raid 0 them I still only get around 300 MB/sec.  I had thought given the vintage of my docks they were all SATA II.  I went to Fry's and got the dual dock because it said it supports 1.5, 3.0 & 6.0G Sata drives.  I took that to mean it the dock slots are 6G Sata III.  When I test the individual slots on the dual dock I only get the 300MB/sec.  This leads me to believe that the dock is only Sata II slots.  Even given that if I raid 0 the two slots I still only get just a little over 300 MB/sec.  I figured it wouldn't double but may get 1.5 times the individual drive speed.  I then split the drive set across two docks both USB 3 to the Sonnet Allegro card.  I get a little better like 333 Mbps.  The Allegro is in my cyclone microsystems expansion chassis hooked up to my mid 2012 MacPro using the 16 lane slot to connect the two.  I know the MacPro is Sata II in the tower but I would have thought if the dual dock is truly 6G Sata III then I should be able to get close to 550 Mbps for an individual SSD . 
 
So what in the food chain is limiting things to around 300 Mbps regardless of whether I raid or not.  The conversion of USB 3 5 Gbps to Mega Bytes per second is 625MB/sec.  So in theory I should be able to get close to the IronWolf spec of 550 MB/sec shouldn't I.  I know there is only one controller in the Allegro card I have.  I've ordered another to get more USB 3 connections and the new one is the dual controller version.  Maybe when I get that in I'll see if splitting across two USB 3 cards helps.  Am I missing some overall architecture limitation under the hood?  I know sometimes I think in theoretical terms without the experience of practical application.
 
Someone please bit slap me into a better understanding of what's going on.  ;-)
 
John Moore Barking Trout Productions Studio City, CA bigfish@pacbell.net
 
 

No comments:

Post a Comment