Page 6 of 10

Re: The days of one G5 being enough are dead! Here comes the

Posted: Mon Jul 25, 2005 9:07 pm
by Timeline
Hey guys, this is dad talkin here.

remember the last studio machines that were built for studios at the dawn of multitrack DAW's? The SONY Dash 48 tracks.

As a producer I loved the concept of 48 tracks and thought man this is the way to create records.

We didn't care much for the tone so most slaved two analog 24 tracks together for even better sound and comped all day. Every studio that was anybody locked slaves, as many as four at once.

Only the post industry and a few brave souls tried to use DASH for records and most preferred analog. Sony never got it better than 20 bit 48K and they were way expensive and late on arrival.

Studer did the 20 bit dash but by the time they got the bugs out it was too late. The digidesign DAW was already here and it all started.

That was only 7 years ago.

The point is,,, this is where we left off and MOTU should be already good with minimum 48 tracks of 96K 24bit high density recordings with every plug in we might want per channel as if our outboard racks were full up with everything to make the most creative sounding record possible.

We should not be changing buffers, swapping drives and screwing around like we are now in every virtual DAW out there. MINIMUM 48 tracks @ 88.2 or 96 and the sonic quality should be stable with NO hits to the sound. VI's should be able to be swapped in in ANY combination of these 48 tracks with the same 'no degradation' sound quality period.

With VI's becoming the new attachment flavor of MIDI, this changes things and many creative musicians like most of you guys are finding you can have live recorded music and combine it with incredible sounding sample stuff for even more creativity, precision, flexibility & tone like this is a surprise?

We didn't ask for VI's, the equipment mfg. created this capability and it evolved and now they find themselves against a wall in a new battle with current technology to make it work? What up wit dat?

No PCI processor card to handle plugs to relieve CPU and adding features that can bog down the CPU with no off switch tells me their priorities are not in place.

They now have to come up with something for flexible interchange of VI's and they don't have it.

If MOTU delays this fix, someone else will make it work and people may bump to that system. This has happened to way too many companies in our audio history and they were left behind for not thinking ahead.

You just can't blame the artist for being handed a new tool and using it.

had this written earlier and just got home so I threw it in to the frey.

<small>[ July 26, 2005, 12:16 AM: Message edited by: Timeline ]</small>

Re: The days of one G5 being enough are dead! Here comes the

Posted: Mon Jul 25, 2005 10:56 pm
by toodamnhip
Having started this post..I must say I am happy about the dialogue...it is much needed..it seems we have a couple different factions here doesn't it?

We have Factions as follows:

1) Annoyed at comps out of power already and striving for more power, complaining about needing 2 comps, looking for ways to get more juice out of what we have now etc.

2) Those who focus on how the technology is amazing and state we all are ungrateful and what we have today is amazing compared to the past.

I like what Timeline said: can't blame the artist for wanting to use what's out there and all the new stuff.

I also like what another post stated, that being, we are NOT spoiled brats, we are for the most part, asking for but a few VI's to work with a full mix and a low enough buffer to play the damn VI's.

I believe the Genie is out of the bottle.

The past is gone, the technology of VI's is F_cking amazing!
Computer technology needs to step up to the incredible vision of the programmers.

I for one, am truly excited about VI's and feel that , though a pain in the ass, they have incredible potential.

Listen people, computers both suck and are amazing.
When the computers can really cook, then man's amazing mind will truly be unhindered by technological shackles.

I feel the programers are the genius, the future, the amazing MIND of man. The computers, though also an amazing MIND invention, represent the limitations of man at this point. I cheer for the soft ware programmers, this is the true invention and mind of man showing it's best!

Those saying to appreciate the past and how far we come, are voting for limitation...

Who's with me? Insatiable, striving?

And on the other side, who thinks I should look to the past and be grateful?

I am not grateful....yet!

Dave

PS..

I also feel Steve Jobs isn;t grateful yet and that's why he kicked IBM to the curb

<small>[ July 26, 2005, 01:58 AM: Message edited by: toodamnhip ]</small>

Re: The days of one G5 being enough are dead! Here comes the

Posted: Mon Jul 25, 2005 11:24 pm
by BKK-OZ
We have Factions as follows:

1) Annoyed at comps out of power already and striving for more power, complaining about needing 2 comps, looking for ways to get more juice out of what we have now etc.

2) Those who focus on how the technology is amazing and state we all are ungrateful and what we have today is amazing compared to the past.
...and I would like to add a third faction, those who remember what one of the most important pioneers of synths, computers and assorted electronica music-making once said:

"shut up and play yer guitar"
-F Zappa

(not that I am saying anyone here should shut up, just saying that, once you upgrade yer axe to include that 7th string, and its all set-up, housed in a high-bandwidth external enclosure, 'cause thats the new fastest, bestest thing available, at some point, you just gotta play that thing!)

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 12:53 am
by richardein
FWIW, here's some info on Supernode and fibreoptics, culled from the net:

"The SuperNode server is a concurrent-connection, multithreaded application server written in C++ and compiled using GCC. PeerNodes communicate with the SuperNode server using Simple Object Access Protocol (SOAP) over HTTP. For security reasons, the actual SOAP messages are encrypted using the Advanced Encryption Standard (AES-Rijndael). Additionally, SOAP packages are compressed using ZLib to help conserve bandwidth requirements."

" A supernode occurs when your computer is arbitrarily assigned as a hub.

When you are running the software for P2P services, your computer••™s disk space, bandwidth, and processing power are used to help other users on the same system operate their software more efficiently because of their own network or firewall constraints."

"Fibre-optic cabling provides a viable alternative to copper. Unlike its metallic counterpart, fibre cabling does not have the stringent speed and distance limitations that plague network administrators wishing to upgrade their networks. Because it is transmitting light, the limitations are on the transmitting and receiving devices driving it rather than on the cable itself.

Fibre provides several advantages to Lans. Due to its unique properties, fibre-optic cabling is used primarily to overcome the distance limitations of coaxial and twisted pair copper topologies.

Ethernet run over coax (10Base2), for example, has a maximum distance limitation of 185m, and Ethernet run over twisted pair (10BaseT and 100BaseTX) has a limitation of 100m. This can be extended with the use of repeaters, although standard 10Mb Ethernet has a limitation of four repeaters per cable run. Fast Ethernet, on the other hand, can only support two repeaters.

These cable lengths are generally more than adequate for the average Lan, with many companies having favoured coax as the backbone inter-connect between the various floors of a building. As Lans have become more and more distributed, however, coax is no longer an effective solution. Furthermore, with the explosion of Metropolitan Area Networks, the deficiencies of copper cabling are becoming even more apparent.

Fibre optics are not subject to the same restrictions as copper cabling, using light rather than pulses of electricity. Light does diminish over great distances, but if the source is powerful enough and the receiver sensitive enough, light can travel immense distances. Fibre can therefore greatly extend Lan boundaries, with multimode fibre reaching lengths of up to 2km, while single-mode fibre can support cable runs of 5km in half-duplex environments, and much more in full-duplex installations. Single-mode fibre has been known to reach distances of up to 60km without the need for repeaters or additional hardware. When distance is an issue, therefore, fibre provides what may be the only solution.

It isn't just raw distance, however, that gives fibre the edge. Even with the highest-quality coaxial cable or twisted pair (shielded or unshielded), the cable will emit some electromagnetic (EMI) or radio frequency interference (RFI), which can interfere with sensitive measurement equipment. This is especially true as connectors and ground connections age or weaken.

In some environments - medical for example - the potential risk associated with this is just not acceptable, but the cost of alternative cable routing is too high for many practical applications.

Furthermore, it is possible to interrogate the data transmitted by copper media by picking up the RFI from the cable using sensitive detection equipment.

So no matter how many firewalls or other security measures you might physically have in place on your network, the inherent deficiencies of the copper media provide the determined hacker with virtually unguardable access to your data.

With fibre-optic cabling, however, there is no radiated noise. This makes it perfectly safe to install in any sensitive environment. Optical fibre adds extra security protection as well. There are no emissions to pick up and decode, and it is not feasible to 'tap' into it for the purposes of eavesdropping. This makes fibre-optic cabling ideal for use in military networks and other sensitive environments.

As well as emitting its own interference, copper cabling is susceptible to other forms of EMI and RFI from other sources; heavy-duty mains supplies can cause particular problems for copper-cabled Ethernet networks. All of this external interference can corrupt the data being transmitted, resulting in a high volume of network errors. This, in turn, means that data will need to be retransmitted, impacting the network's performance.

This can be a particular problem in noisy industrial environments. Fibre, on the other hand, is immune to interference from outside sources.

The physical appeal of fibre

Along with the electromagnetic properties inherent in using light as a communications medium, fibre-optic cable also has several key physical properties that add to its appeal. For a start, it is smaller in size than twisted pair cables and so takes up less room. Compared with Category 5 UTP, most fibre-optic cable can also endure a tighter bend radius while maintaining specified performance."

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 3:16 am
by Timeline
"shut up and play yer guitar"
-F Zappa

at some point, you just gotta play that thing!)

--------------------
Cheers,
BK
Supernode and fibreoptics
BK, yes That is a timeless statement and thanks for the info on Supernode and fibreoptics. We will see what it brings us soon.

My previous post was simply a brief reminder of history and a marker as to where we should be today and why. Clear 48 track DAW's minimum regardless of how many VI's within the 48 without choking.

I have been saying that were behind with the virtual DAW for some time now and certainly understand why studios went with PT as the standard.

Credit where credit due. Digi really did try to make it seamless back then with hardware technology and today still stand poised to beat the virtual DAWs to interchangeability with VI's if they deem it applicable.

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 4:16 am
by Jaysplace101
Gary, you are right on the money with all of this.

Richard, thanks for the "Supernode" info. Did you find out any ways it could be implemented into a G5 chain? I'm still confused as to if slave was bull-SH'ing us right from the start. I think he was, but looking back, I wonder if he really didn't want to share and made the stories up from then on...???

qo and filtertone. I apologize for my ignorance here, but how do you hook up SATA drives to your G5?

Thanks everyone. Great thread.

j

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 4:29 am
by croyal
Originally posted by toodamnhip:
YEs..this means a second machine for VI's.

To utililize all the wonderful stuff out there...it now takes TWO G5's!
One for DP...the other for VI's.
This hit home hard for me today.
Genie may be out of the bottle, but these systems do have limits. 48 tracks wih everything running and available and instantly switchable would require a CPU for each task:
at least-
1 for DP host
1 for VIs
1 for Reverbs
1 for Summing
1 for eqs.
1 for compressors
1 for good luck

This and more is (and has been) possible with a bunch of G5 servers and network cards. For instance; presenting the 3rd fastest supercomputer in the world!!

http://www.tcf.vt.edu/systemX.html

So, of course the technology exists now to do all these wonderful things.
My earlier point was that we shouldn't expect 1 machine to do what a Mac and a bunch of internal cards will do- as in a TDM system. And, even with a supercharged system there are still limits- ask anyone who has choked a TDM system. Oops, it's hard to choke one because Digidesign tells you all the limits UP FRONT!! So are we being naive or just playing Devil's Advocate by expecting more from a lone dual processor machine?

This type of discussion is moot, as there is nothing stopping anyone from throwng money at the problem- just buy more nodes. Virgina Tech just went on and bought the 1100 Macs and immediately posted limits for simultaneous processing and user storage.

I just don't think it's always productive to point fingers at MOTU when these are hardware (CPU) considerations.

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 4:42 am
by tommymandel
I agree with Mr Marr. . . VI's may be the dark side of the Force - inviting, visually sexy, but ultimately a sure fire one way trip into Hardware Insufficiency. (but like Mr.Marr, I still have tons of synths, Rolands, Prophets, Ensoniqs, a few Yamaha's, Akai samplers....
True, we lose instant recall of a session in its entirety.

tommy, nyc, 2408mkI,SycoLogic Patcher, JLCooper 16/20 Patcher, MicroExpress, LinnSequencer v. 7.03 software, RFoland D50, D550, JD800, JV80, XP50, V-Synth (yeah!), Akai S3000XL, S700, Sequential Circuits Prophet 5, Prophet T8, Ensoniq TS-10, MR-Rack, SD-1, KT-88,, ASR-10, ASR-x, Yamaha CS-2x, EMT-10, Motif6, (no Korgs, you might notice...though I'd like to have a Karma...) DP3.11 with OS 9.2.2, on a G4/450 with PowerLogix 1.2GHz upgrade (2MB Level 2 Cache) 1GB RAM, and Sonnet Tempo pseudo-SCSI PCI card with 3 HD's and 2 monitors. Also PBG4 Titanium 667/DVI, with1MB Level 3 cache (runs great), running OSX10.3.9 with DP 4.6. No VI's: I tried the Mach V v.2 demo, but the PB said, "sorry." Rebirth and GrooveMaker - VI Programs that run separately (under OS9 Boot, and yield interesting loops.)

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 4:43 am
by MT
As I said in my previous post, for you VI/MIDI guys, clustering through Xgrid may be possible now. There's a lot out there about using the Audio/MIDI setup in OSX to slave additional machines off of a main node:

One post claims easy setup and the ability to run Reason, Battery, etc. off of an extra G4
:
Link 1 with screenshots

Here's the related article:

Link 2

Doesn't sound like bullsh•• to me. Someone with the time and extra machine should give it a shot.

MT

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 5:18 am
by filtertone
Originally posted by Jayman:


qo and filtertone. I apologize for my ignorance here, but how do you hook up SATA drives to your G5?

Thanks everyone. Great thread.

j
After looking at a bunch of possibilities, I decided on a Kano Technologies Xspand 4 drive SATA rack along with a Sonnet 8 port SATA PCI card. The guys at Kano who make the rack worked with me on getting the unit unbelievably quiet. Very impressed with these guys. BTW, my old FW800 drives are great for video playback and general backup.

http://www.kanotechnologies.com/enlarge_image/xspand-SATA-rack.cfm

<small>[ July 26, 2005, 08:18 AM: Message edited by: filtertone ]</small>

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 5:35 am
by Jaysplace101
Thanks Filtertone. I just checked this out, and it looks perfect. And, It looks like I could get into a 480 GB system for around $1,200 - with the Sonnet card. Was there something specific they did to make your unit quiet?

Thanks again,

j

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 5:43 am
by filtertone
Originally posted by Jayman:
Thanks Filtertone. I just checked this out, and it looks perfect. And, It looks like I could get into a 480 GB system for around $1,200 - with the Sonnet card. Was there something specific they did to make your unit quiet?

Thanks again,

j
They switched the fans to the Silenx brand -- talk to Robert Young if you end up calling them.

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 6:48 am
by Timeline
Croyal wrote:
I just don't think it's always productive to point fingers at MOTU when these are hardware (CPU) considerations.
One could throw bucks at audio production and create a mini super computer to do audio but I can't afford that, can you?

I'm just saying that it seems to me that it would be more on target to at least keep MOTUs priorities pointed towards performance over tools at this point considering we have SO many tools now.

Ya think?

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 7:29 am
by grimepoch
I think expecting MOTU to enhance your use of the VI limit in DP is not a totally fair in as much as making sure the parts DP is responsible for are as fast and optimized as possible.

You are taking a generic item (CPU) and using it for a multitude of tasks that are not music specific, therefore it suffers from being so open ended. I've been complaining about the need for generic DSP based solutions for a long time, however, people are writing VI's for the main processor. It's hard enough to get everyone to write their plugin in every flavor (VST,AU,MAS,RTAS,DX,LADSPA) of every OS (OSX,WIN,LINUX). Look how many plugins you cannot get for the available DSP cards out there now?

The solution for the VI limit (again, beyond DP bottlenecks) is going to be the computer hardware. Unless the plugin writers are going to work with MOTU on hardware, which I doubt is going to happen, then the solution will ultimately be the hardware for which the software runs.

Said differently, I can take reaktor and create a plugin that will run on a G3, I can also write a plugin that can't work on any existing hardware without dropouts, believe me, I've done it. Why? Because the plugins are open-ended. They are not a finite processing value, as I know you already know, so expecting MOTU to be able to tell you that you can run 50 VIs is ludicrious. There would be not unit of measure.

This problem has long existed, and even moreso, from horrible programming. Why is Word so slow? We've been using WYSIWYG word processors for years yet, consistently, they feel slower. And really, what new feature have they needed that would slow them down so much? Some companies produce horrible code, and they let the hardware speed increases pick up the slack for it.

Sure if you can get more VIs running in Logic, and it performs better, well, I believe that would be a valid argument. But, you have to also remember, it may not be just that simple. It's possible a feature in DP is causing that bottleneck, or a fundemental architecture difference that cannot be overcome. Again, this is just as much speculation as any other suggestion, but at least paints it in another light.

Personally, I think that multi-core systems are going to be the answer, not some PCI card. Not a PCI card that is going to be essentially duplicating process power of the main CPU, why? Quad and higher core powers are just around the corner, it wouldn't make any sense to recreate the wheel for that price point. Not to mention all the code that would need to be written to get it working, and the compatibility issues. Look how hard it is to get 100% reliable interface with a VST to AU wrapper? And that would be simple compared to a card that has to offload CPU usage.

Further, I don't think it is unreasonable to use additional computers as VIs. I've been doing it for years, my first system ran Cakewalk Studio and a second system running samplitude (shudder) and once I had it running, and my mind in that mindset, I had no problems, other than PCI bus issues with a VIA chipset.

Like someone else suggested, I too run my other machines as stand alone synths, and I have zero problems. Sure, I too want a system where I could load up a project and everything is there on one system, or that it loads everything on all machines, but that inconvience is worth it for the extra power and sonic capability I am getting.

But, back to the main point. AU's are not a finite processing quantity. And with no way of gauging them, or limiting them, MOTU can't claim any number of other peoples AU's would work.

And, this problem isn't going away. As we get more power, people write more and more complicated algorithms. It allows us to make synths that can generate much more complex sounds, or provide features we just can't utilize today. So you might as well get used to it.

The convience of digital effects and VIs has a price, and that is your CPU power :)

TIP: And the biggest thing I can recommend to people is this, if you aren't adjusting an effect or VI, CLOSE THE DAMN WINDOW! You would not believe how much power the GUI's take to run. This I cannot stress enough.

Re: The days of one G5 being enough are dead! Here comes the

Posted: Tue Jul 26, 2005 7:56 am
by Timeline
I certainly respect your points encOder.

On the other hand, the first thing that's coming down the pike is throughput due to PCI express of the Intel chip and processors are likely much further off that will make much difference so I think it is well worth it now to consider a dual purpose PCI card that handles audio wire and AU plugs. This would give immediate headroom for independent drives and VI exclusive setups externally.

I don't pretend to know programming like you but at the same time it seems like handling various third party AU's in DP is currently being done very well so why not direct it to another processor on a dedicated MOTU card.

Seems doable to me.

Just wanted to add, It would be also nice to see at least 8 AES connects on a new card too. I would give up an audio wire for that.

<small>[ July 26, 2005, 11:03 AM: Message edited by: Timeline ]</small>