2TB-4TB eSata

[Knezzen]

http://duc.avid.com/showthread.php?t=15603

Here the tip seems to be to remove the second PCI graphics card to free up some bus bandwidth.

[supernova777]

knez this isnt a fight man.. there is no winning lol this is just conversation... with the added benefit of sharing information that might benefit others..

u dont have to make a video i understand + believe that you have problems.. im not trying to say that you dont..

but your not getting my point at all dude..
"pci bus too busy"... is the explanation assigned to teh error code YES
but that actual error isnt being triggered because of a lack of bandwidth of the pci bus.
its that its being denied access because another pci "Bus mastering" card has "taken over" and gained precedence..

http://www.techopedia.com/definition/299/bus-mastering

If a computer contains several components that support bus mastering, hierarchical structure needs to be implemented to prevent several components from trying to use the bus at the same time.

knez is a genius and knows all so he knows already but
for other readers (+ cake eaters)
pci bus mastering.. read the definition on the page..
this is not the normal pci bus.. this is an "enhanced mode"
of a single card that is functioning on the pci bus..
which offers "elevated priority" + direct access to memory etc

The bus master is the “master” and controls the bus pathways that contain the transmission signals and address.

having more than ONE card that has this ability causes PROBLEMS
in a similar way to having more then one ISA card on an oldschool pc using a IRQ
they are both competing for using the same resources.

my comment to pro tools guy above about removing the graphics card *WAS A JOKE*

below is a quote from cubase 5 troubleshooting pdf:
ftp.steinberg.net/Archives/Cubase_VST/Docs_English/Troubleshooting_Cubase_VST_5.pdf
page 15

Graphic cards
Graphic card drivers (e.g for Matrox cards) can block the PCI bus when screen up-
dates occur. Some graphic card manufacturers have discovered that they can in-
crease the card’s performance by designing drivers that send commands over the
PCI bus without checking whether the card is ready to receive them or not (ignor-
ing the queue check of the graphic card). If the card is not ready, the PCI bus is
blocked until the card can receive the commands. As a result no other activities can
occur in the computer, which in turn may cause crackling/clicks (as well as tempo-
rary dropouts and channel swapping)

In order to solve this problem, you can either simply turn off the PCI performance
enhancements in the graphic card driver, get a new and better driver (if available)
or change the settings in the driver so that it doesn’t send commands to the card
until it is ready to receive them

this cubase thing is JUST AN EXAMPLE of how the logic of the operation of the devices is really at fault.. not the devices themselves. there was a more detailed explanation of this like i said that came with cubase 3.7 or 4.0 for pc in the readme file.. they included it with the 5.0 documentation because its a hugely important peice of information to people who at the time were experiencing audio drop outs..and the reason was because their graphics driver was the cause of blocking the pci bus.. but everyoen at the time thought it was that the pci bus was clogged.. and this was found to be incorrect.

this is only relevant because this is exactly the same type of error that causes audiodropouts in other types of configurations aswell.... when a certain device takes complete control of the pci bus + leaves it unavailable to another...
this is not a data thruput limitation of the pci bus.. but rather a resource conflict due to improper hierarchial precedence.

[Protools5LEGuy]

yes protools guy u are right.. easy fix is to take out the graphics card

Chris, don't kill (blame) the messenger... XD

To be more accurate with my example, an Sawtooth/Gigabit ethernet is a cake with 8 pieces

Any 133MHz bus G4 has a cake with 10 pieces.

A MDD w/166 bus is a cake that has 12 pieces.

And about "bad behaviour", bouncing a mix to another drive different from the audiotracks takes 1-2 pieces.

Some services, like AFP or Indexing a drive in background can take 2-4 pieces.

It is normal to get some more errors (At least on LE systems) when protools is set to make "quick punch" recording. It takes 2-3 pieces.

Using 2 graphics cards take 2-4 pieces.

I had tested my G4 DA w Dual GHz CPU with an Rage 128 and DIGI001 in one setup (for more than a year), and Rage+Digi001+AMIII+USB2card+Sonnet ATA66 interface (for more than a year too) and the first setup gives less -6042 DAE errors. 


Chris have a valid point that is Bus Mastering protocol is very powerful and has a lot of headroom for audio tasks, but if Knezzen has first hand experience with the quickest MDD with MIX +++++++ system with a Magma Chassis, how you dare to question it?


The shame is that our more wise-experienced members MacTron and DieHard are Cubase fans and noone has/had a TDM rig. But they have their own troubles mixing M-Audio with UAD-1, Powercores and GeForces 4 Ti aside ATA-SATA cards.

[supernova777]

knez theres a good chance
http://duc.avid.com/showthread.php?t=15603
that link u posted.
that guys problem was probably 100% related to his video card driver

just like the cubase documentation, one of his graphics cards probably had the driver that was bypassing the ready check.... resulting in a timing fuckup + causing a "Device not ready" error for the other devices (which resulted in PCI BUS too busy digidesign error)

the other readme that i read said that they graphics card companies did this to compete with each other starting around 1998 they bypassed this ready check to make the graphics appear quicker then the competitions.. and only a few graphics cards drivers had the ability to toggle this on or off.. most (ati included) had this hardcoded into the driver

this is just speculation on my part

the scsi card could be to blame too.
https://www.google.com/search?q=Adaptec+3940UW+pci+bus+busy&ie=utf-8&oe=utf-8
funny that theres so many results about the adaptec card + the problem
seems to be a relationship there

http://comments.gmane.org/gmane.comp.audio.emagic.logic.tdm/8639
but alot of them also include 2 graphics cards.. + the scsi card like this guy ^^
anyway.. not my problem.. but having 2 seperate graphics cards + a scsi card seems to be a recipe for disaster

http://osdir.com/ml/audio.emagic.logic.tdm/2000-08/msg00285.html

I think the troublemaker is the Adaptec. Digi has
> commented that this SCSI accelerator is really hard on the PCI bus.
> The Atto that they recommend supposedly uses 1/4 of the PCI bus
> resources that the Adaptec does. Next, I would suspect the video
> cards.

im not trying to say that im right FOR YOUR EXACT SITUATION
im just sharing complicated information.
for you to contemplate... im not arguing + fighting with u bud..
get a grip

maybe if u people truelly spoke english you could understand what im saying instead of skimming the copy + picking out bits and jumping to conclusions about what im saying

[supernova777]

they have their own troubles mixing M-Audio with UAD-1, Powercores and GeForces 4 Ti aside ATA-SATA cards.

Diehard has never had a silicon image PCI SATA CARD.. and is basing all of his comments on past experiences from yeras ago

as far as i know Knez also hasnt got a pci sata card... at least not one that he has talked about publicly on the board. which is why i just asked him which card he claims to have tested.

i was going to send diehard a pci sata card but he refused it because hes fucking traumatized from past experiences lol the same reason he refuses to have osx installed (even on a 2nd partition or drive) on any g4.. because he believes the mac boogie man will inject btree errors into his filesystem..
(did u take your chris vitamins?)

[Protools5LEGuy]

Diehard has never had a silicon image PCI SATA CARD.. and is basing all of his comments on past experiences from years ago

DieHard worked on NY city building DAWs for first class studios at least since mid 90s. His building DAW notes are priceless. Those studios probably were based on first class SCSI systems, not cheap SiI 3112 cards...    In a studio that charges 200-300$ an hour a zero-fail tolerance DAW is a must. Probably a dual channel SCSI setup were the bare minimum then.

Some magma-bit3 users say to use ATA-SATA-SCSI cards on the expansion box while others say to use those cards on the host.

If Brian (SonikArkitech BT) were so kind to put some light on this darkness... He uses Mix + system with Korg cards and all the shit we dream about having on our machines.

I would love to hear Protools nubus users talking about that, because they really doesn't have all the headroom we quicksilvers/MDD users have.

From what I readed on D.U.C. (digidesign user conference) SCSI is preferred over ATA66-100 and SATA interfaces (SCSI takes 1 piece of the cake)  Probably 10.000 rpm from SCSI vs 7200 rpm on ATA is the clue.

i was going to send diehard a pci sata card but he refused it because hes fucking traumatized from past experiences lol the same reason he refuses to have osx installed (even on a 2nd partition or drive) on any g4.. because he believes the mac boogie man will inject btree errors into his filesystem..

Better safe than sorry. He has no problems using pure OS9 setups. Probably himself or various clients had that btree error with OS9andOSX while some pure OS9 setups never had that trouble.


BEWARE! Sarcasm at 11!

(did u take your chris vitamins?)

So you are selling blue pills for erection? Or Matrix pills? I thought you were vegan and drug-free, but your House music has make me wonder if you abuse some substances.

Coconut water is a drug. 

[Knezzen]

Chris. I don't have to prove anything to you. Rant all you want. My english is excellent BTW, thanks for bringing it up.

[supernova777]

so yea knez what brand of sata card is it that you have?

[supernova777]

listen guys..
to me, its very obvious the goal/motive of each person is clearly viewable from the content they post..

some people genuinely want to help.. and share + talk about actual usefull information
some people want to boast about what they know
some people just want to be right + change everything into an argument
u just make me lose respect for you when you attack me personally because u cant think of anything else to say.. its really pathetic..

knez..
lukpac..
protoolsguy

im always trying to help people.. im sorry that you seem to want to interpret what words i say as me being boastfull + knowitall.. im not a knowitall.. i actually looked up + researched what i say before i say it..
im not here to shoot my mouth off .. off the cuff... on a whim..

you all have very little good energy to share
perhaps its time i take a break from macos9lives

[IIO]

so, now since there is nobody left who wants to discuss the original topic, this thread could be closed.
 

[lukpac]

so, now since there is nobody left who wants to discuss the original topic, this thread could be closed.
 

I was going to report on my experiences once I wiped/repartitioned. I just haven't gotten that far yet.

[IIO]

 
 
 

[Protools5LEGuy]

so, now since there is nobody left who wants to discuss the original topic, this thread could be closed.
 

IIO, sorry for offtopicing this 2TB'4TB long post. A few post after BT asked it was really clear somehow that You should not buy a drive bigger than 2TB for a PowerMac and that we should partition it in slices no bigger than 220 Gigs...

I was just joking somehow on my own way about who to blame... the boogie

My biggest drive is a 5400rpm SATA II seagate that is only 1.5 Terabytes. And is inside my hackintosh. I cant give any advice about giant drives.

On the PowerPC side, my MDD has 4 ata drives conected. 3 of them are in APM. Nothing strange. But the bigger, a 230 Gig 7200rpm seagate drive only partition is only saw in OSX.

The 4th drive is an ATA drive partitioned on GUID (Yes, you all can call me ignorant) and it appears only on Leo (not sure if Tiger sees it) Panther do not see it, that is sure.

[IIO]

so, now since there is nobody left who wants to discuss the original topic, this thread could be closed.
 

IIO, sorry for offtopicing this 2TB'4TB long post.

i wasnt exactly thinking about you.

but while we are on topic again, while it is uncertain if the size of his drive could be the reason for his protools problem, it is also uncertain if it is not.

personally i think ed is much closer with his idea  that it can happen from using such a fast disk on an ATA bus. s

something which worries me a bit, because if it can affect protools´ disk reading, it might also effect stuff it or toast or my favorite video editing apps.

[DieHard]

I am making this last response in reference to the whole PCI bus mastering discussion so that newbies or people building a DAW can make so quick notes to avoid problems; this is not a debate, this is just real world observations.

Some Background:

The term "Bus Mastering" describes a protocol where the device itself performs the basic computations necessary to perform input/output, thus freeing the CPU for other tasks. UltraDMA (ATA-33) was the first version of IDE to fully use the Bus Mastering protocol. Bus Mastering interfaces and devices are usually faster than PIO.

From Chris: My opinion is that was much more likely a limitation of the CPU's ability to handle and process what was being done on pci bus then the actual pci bus itself.
Diehard.. I would bet u cash doing tests with different levels of macs would reveal the cpu to be the mitigating factor..  comparing a 450mhz sawtooth to a 933 qs to a 1.42mdd and finally a 2.0ghz upgraded cpu.. when the higher models have no issue at all would u say that its because their pci bus is "better"? seriously think back to which macs were actually in use back in the day when u recall these problems... guaranteed it was back around 1998-2001 when they had a dog of a cpu.

No, the CPU has very little to do with the issues and problems we are discussing.  Firstly, these issues on older hardware is not unique to macs... remember, back in 2003, my company not only made between 5 to 15 DAWs a week, but we also made Novell Servers (we were Novell Gold certified) on rack based PCs that served both PCs and Macs.  Some of our accounts in New York City had between 100 to 150 workstations both mac and PC connecting to a single server to store files, these systems worked great as long as guidelines with the PCI bus were established.

So, now I'll get to the point, these systems (with not much CPU power) worked perfectly as long as there were ONLY 1 to 2 Bus mastering cards installed in the server... additional SCSI cards used for Tape storage and other media were all either configured as non-bus mastering, or were created from the factory as non-bus mastering cards. (Sometimes the NIC cards were bus mastering also).  The point is that the moment too many bus mastering PCI cards were used, all sorts of performance issues on the server would crop up; including slow Hard drive writes, NIC packet issues, and many other problems. Remember, to add to the confusion, some cards can be either be configured as bus mastering or not and some systems have dedicated non-bus mastering slots... so read about the actual cards and system you have.

Back to the Macs... the issue regarding too many Bus mastering PCI cards that tax the PCI bus are very real on ALL MODELS OF POWERMAC G4s from Sawtooth thru MDD EVEN with CPU upgrades; hope that makes it clear enough. On newer macs (like a Mac Pro with snow leopard), this, or course is NOT an issue.  We are specifically talking about older Mac Hardware in the G4 era.

From Knez: The PCI bus of the MDD for example is connected to the same controller as the ATA100 controller, but they do not share bandwidth in any other way. Tons of activity on the ATA100 controller does NOT slow down the PCI stuff. Adding a controller to the PCI bus makes it use up bandwith there insted of where it's "supposed to be", and thus leaves less bandwidth to the other stuff on therse.

This is 100% correct and that is why my MDD systems that all have a UAD-1, PowerCore, and Audio Interface PCI card, all have an SSD drive on on the ATA bus, NOT a SATA card.  So the bus mastering of the internal IDE is done by the CPU itself, while the cards do their own, this balances the load on a fast G4 very well with disk I/O bus mastering being done by the CPU and audio data blocks being done by the bus mastering PCI Audio interface
so to summarize...

To all reading these crazy posts that want some real-world guideline and NOT opinions and theories; as a rule of thumb (without going into specific G4 models and configurations that I have tested and built for DAWs)

1) Keep the number on Bus mastering PCI cards (Like SATA, Audio Interfaces, and SCSI) to a Max of 2, in rare cases 3, but test the system; (non bus mastering PCI cards like extra USB will have no effect and add as many as you want, research or test each card)
2) If your interface is FW, then SATA/SCSI Cards will be fine and not cause any issues
3) Always Initially test the system with just Hard Drive I/O and Audio interface, then add cards like PowerCores, UAD-1, and other PCI one at a time and test again

[Protools5LEGuy]

Going to original post questions

Under a pure OS 9 environment, one main issue that has been discussed in other threads is hard drive maintenance;  Any volumes above the 200GB limit may be a real hassle if they become corrupted or need of a defragmentation (audio recording volume). 

Norton Speed disk and other defragmenting (OS 9 programs) will produce "Out of RAM errors" and NOT work.  A work around suggested by some members was to create an OS X bootable DVD with a defrag utility and use that method.  Many OS 9 disk maintenance (repair file/volume issues) will also bomb on volumes greater than 190 to 200GB; so that was the original reason that I suggested all volumes be 190GB or less was because I was under the assumption that the data put on these volumes was important audio project data (or hot pics of sexy babes).

IMO huge volumes that are not repairable are not the way to go.  Most of organize our data into folders, so the additional hierarchy of Volumes seems nice and logical... a Samples Volume (190 GB should be enough for most users), Mac OS (OS & apps), Audio Projects 1, 2, 3... it does not seem unfeasible.  Remember, when you put all the eggs in one basket (or on 1 volume), don't bitch if you crack a few. 

Where is that OSX solution to defrag hard drives ? I have done the maths and your 190 Gb maximum partition size easily beat the 21 partitions limit on a 2Tb drive. You or someone mentioned that that DVD need a Leopard compatible hardware to boot. I guess all G4 users (But apple says 733+ IIRC)

What are your concerns about letting that tool defrag the drive because OS9 just cant handle them?

It could be interesting to install that DVD on a partition and boot that defragger instead an OSX partition that could lead to BTree errors. On a pure OS9 environment. Or better called a OSX free environment.

It is questionable if that defragger should be considered OSX on not 

To all reading these crazy posts that want some real-world guideline and NOT opinions and theories; as a rule of thumb (without going into specific G4 models and configurations that I have tested and built for DAWs)

1) Keep the number on Bus mastering PCI cards (Like SATA, Audio Interfaces, and SCSI) to a Max of 2, in rare cases 3, but test the system; (non bus mastering PCI cards like extra USB will have no effect and add as many as you want, research or test each card)
2) If your interface is FW, then SATA/SCSI Cards will be fine and not cause any issues
3) Always Initially test the system with just Hard Drive I/O and Audio interface, then add cards like PowerCores, UAD-1, and other PCI one at a time and test again

Did you enjoyed the cake example?   

 

[Jakl]

Thanks for that Diehard.

[IIO]

we are deeply off topic again but somehow it is funny how the thread evolves.

Where is that OSX solution to defrag hard drives ?

you dont really need to defrag disks in OSX, but iDefrag works very well if you dont like norton.

I have done the maths and your 190 Gb maximum partition size easily beat the 21 partitions limit on a 2Tb drive.

it is not exactly a limit. it is just  that you have to mount the 22th partition by hand.

and i think it would be fine to create 3x192 for OS9 and audio files, and then make some bigger partitions, like 3x400 in addition.

Did you enjoyed the cake example?   

i did. it was the most helpful contribution so far.

btw. "defrag on dual boot."

there is a simple trick how you can avoid repairing or defragmenting your OSX install into pieces: simply run your norton apps in OS9 from the same partition where your OSX install is – then they wont allow you to operate on that partition by accident.

[DieHard]

Thanks IIO, Knez originally posted the iDefrag info...

DieHard: I use a bootable DVD of iDefrag and defragment my drives that way. Never tried Norton or any other OS9-only defragmentation application on my drives...
...it only works on Mac capable of booting 10.5.x though, since the iDefrag DVD uses a custom version of 10.5 to boot into...
Try it out, it works great and is a great workaround. Might post the DVD image here for all the people not using OSX. Then they can at least burn the DVD and boot into iDefrag

Maybe Knez can post an ISO or Toast image of the DVD he uses... since I am a creature of habbit, I will still keep volumes at 190GB and use Norton Speed Disk when needed. The use of an SSD for current projects, makes defraging obsolete, since I copy the finished project to the non-SSD drive (and thus most files are written linear and not edited after they are finished); eventually, I delete stuff, and recopy, so I will still occasionally defrag; but the days of having to defrag your main audio project drive are over now that SSDs are available.  More recording and less maintenance

[Knezzen]

Sure, I can upload the iDefrag DVD. It's only downloadable from the iDefrag application, so it's quite hard to come by.