Masterverb Test Thread

PC Configurations, motherboards, etc, etc

Moderators: valis, garyb

User avatar
garyb
Posts: 21562
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: Masterverb Test Thread

Post by garyb » Sat Oct 05, 2013 10:03 am

Eanna wrote:15 dsp scope and 6 dsp pulsar 2 here.
Low ulli available, but pci overloads common enough too when loading large devices..

Wonder if setting which is board zero in cset.ini makes any difference?
nope. the later cards handled PCI traffic better. if the older card is in the computer, it degrades performance.

Eanna
Posts: 598
Joined: Fri Dec 02, 2011 1:57 am
Location: Ireland

Re:

Post by Eanna » Sat Oct 05, 2013 10:34 am

ScofieldKid wrote:This is MSI Neo2-F (Nforce3 Ultra), AMD 4000+ , Nvidia AGP 6800 Ultra, 1 Scope Pro: 9 Masterverb Pro's. This NForce3 Ultra board is performing quite nicely.

Added a Project to my Pro, so now 6+15=21 DSP and 13 Masterverb Pro's. Note: my system has more capacity when the Project is "board0", i.e. DSP's 1,2,3,4,5,6 in the DSP Load window. I would only get between 9 and 12 if the order was reversed of that. The CSET.INI trick lets you switch that fairly easily, so it's good to pay attention to that.

<font size=-1>[ This Message was edited by: ScofieldKid on 2005-12-14 03:43 ]</font>
Looks like I found an answer to my question.. ;-)
I'll try tonight...
Not because it is easy, but because it is hard...

Eanna
Posts: 598
Joined: Fri Dec 02, 2011 1:57 am
Location: Ireland

Re: Masterverb Test Thread

Post by Eanna » Sat Oct 05, 2013 10:36 am

Hi Gary, sorry man wasn't being cheeky! Was trawling thru earlier posts to see if my idea cropped up, didn't see your post before I posted mine..

I'll try tonight and see if there is a difference.
Not because it is easy, but because it is hard...

User avatar
garyb
Posts: 21562
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: Masterverb Test Thread

Post by garyb » Sat Oct 05, 2013 11:16 am

please do. i hope it helps!

User avatar
Bud Weiser
Posts: 1417
Joined: Tue Sep 14, 2010 5:29 am
Location: nowhere land

Re: Masterverb Test Thread

Post by Bud Weiser » Sat Oct 05, 2013 1:33 pm

garyb wrote:
if a single card will go below 13ms @44.1k, it is gen2. early 15dsp cards were gen1, most of them came with 24ADAT i/o. later 15dsp cards were gen2, a few came with 24 ADAT i/o. if you saw both types of 15dsp cards side by side, you'd see that the gen1 card was a little different in size and shape.

all 4dsp cards were gen1.
Thx Gary !

Now, mine is Gen2 because it goes below 13ms @44.1k, has 15 DSPs and 2 ADAT I/O ports (16ADAT i/o).

What was the product/model name when the analog I/Os were cinch (unbalanced) connectors ?

PowerPulsar or SCOPE Classic or SCOPE Pro or ... ???

Bud

User avatar
garyb
Posts: 21562
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: Masterverb Test Thread

Post by garyb » Sat Oct 05, 2013 2:11 pm

all of the above. the i/o plates were all available as a choice. the i/o plate with RCA(cinch) and no z-link connections was called "classic20".

Eanna
Posts: 598
Joined: Fri Dec 02, 2011 1:57 am
Location: Ireland

Re: Masterverb Test Thread

Post by Eanna » Sat Oct 05, 2013 3:31 pm

So....

I've checked my cset.ini, and I was surprised to see just one board listed!

[Host-Config]
Numboards=1
Board0=Host

I don't have either of:
[board0]
or
[board1]
groups in my cset.ini.

I'm nearly sure that in my old rickety PC (pre-June this year) that my cset.ini had the board0/board1 entries... That PC was flakey, hence my 'upgrade' in June... (got a near-new Dell Precision T3500 at a good price).
I have some backups of cset.ini on this PC, and none of them mention two boards.

I'm glad to see that it appears my Pulsar2 appears to be the first one to load DSPs... In the DSP Meter, I first have 1/1 - 1/6, then 2/1 - 2/14.
But wait! 2/14?! I am quite sure that I have a 15-DSP card in there! That's confirmed by the "DSPs 21" in the DSP Meter dialog header...

So what's going on? Is the 15th DSP on those board used for some redundancy? Or some board communications? Does the fact that I've a 'redundant' 15th DSP signify that I need have a Gen-1 board in?

Note that whenever I start Scope, I get:
---------------------------
Sorry!
---------------------------
A capacity limit of your DSP cards has been reached.
The specific cause is:
DSP overload: Big modules do not fit

Followed by:
---------------------------
Capacity limit reached
---------------------------
A capacity limit of your DSP cards has been reached.
The specific cause is:
DSP overload: Big modules do not fit

I dismiss these dialogs, and I can continue to use Scope fine. It's a minor annoyance...

So, in keeping with the idea behind this thread, I started to add some MasterVerb devices to a project with just an Analog In, an Analog Out, and an STM 1632 in.
I got as far as 8 MasterVerb's before I got a PCI-Overflow message. I dismissed that, and added a 9th. Twice more I got the PCI-Overflow, then boom - BSOD. :-(
(losing my post here in the process!)

I've addressed IRQ sharing in the recent past as well as I can.

So what is this test designed to achieve? What are we actually testing here? I can load Zarg synths, big ones, especially since I sorted my IRQ Sharing... My DSP Meter read about a third/two fifths when I hit the BSOD, but I can and do bring my DSP Meter to about 80% on busy projects before I get cagey. What's so special about these MasterVerbs?

(Note too that I get an Incompatible Preset File dialog when adding MasterVerb to a project - dunno why - I just dismiss it, and can continue to load presets fine from "Bank0").

So:
(1) My cset.ini unusually doesn't talk about my two cards, but Scope, when running, finds them. Maybe that's why I have to dismiss the "Big Modules Don't Fit" every time I start Scope?
(2) I am loading my 6DSP Gen2 card first, ahead of my 15DSP Gen1 (quite sure it's a Gen1 now). These two reasons mean I'm slow to mess with cset.ini.
(3) MasterVerb is a pretty poor 8 before PCI-Overflow. Adding a 9th is something I won't be doing again in a hurry...

I see dudes back in 2005 saying they're 'only' getting 14 Masterverbs. My CPU would crap all over their PC's, my graphics card would render the life out of them, my SSD would wipe the floor with their mechanical drives, I've aggressively disabled all I can in Win7 32-bit, and I have a tasty 21 (20 usable it appears) DSPs hanging on tough in there. But I'm loading 8 Masterverbs...

Do you think I should address the board0/board1 business in cset.ini?

Thanks, Eanna
Not because it is easy, but because it is hard...

User avatar
garyb
Posts: 21562
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: Masterverb Test Thread

Post by garyb » Sat Oct 05, 2013 4:04 pm

some people are willing to go to extreme lengths to optimize for the card. that was more true in the past, when some motherboards worked and some didn't. now most motherboards except the ones from the 1155 era work. people are also not setting aside a computer just for audio and destroying all nonessential processes. in addition there were some motherboards that worked REALLY well in the past. those numbers never included 1st gen cards though.

PCI overflow is when the flow of data along the PCI bus gets so congested that it will no longer support realtime audio. the reason a masterverb is the tester is that a reverb is tens and even hundreds of thousand delay lines running from the card down the PCI bus to memory and then back to the card. this is a lot of traffic. the masterverb is a known quantity that everyone can use, so it's a good test of motherboard performance.

i never mess with the cset unless there's a really good reason. just the fact that you could go below 13ms latency @ 44.1k with two cards means that the newer card is the number one(ok, in computer talk that's number 0) card. don't change it.

is the project that has that message old, like pre v5.1? is so, i might rebuild it.

if you have a first gen card, 8 masterverbs is normal, even great. for a 2nd gen card only, 8 isn't great, but it's not bad.

yes, the 15th dsp is redundant which is why it was removed by Creamware to save money in their last days.

Eanna
Posts: 598
Joined: Fri Dec 02, 2011 1:57 am
Location: Ireland

Re: Masterverb Test Thread

Post by Eanna » Sat Oct 05, 2013 4:14 pm

As always, thanks Gary..

If reverbs round trip to memory, seems that auxiliary memory available on scope board would have been a good idea?
Not because it is easy, but because it is hard...

User avatar
garyb
Posts: 21562
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: Masterverb Test Thread

Post by garyb » Sat Oct 05, 2013 4:35 pm

sure, but it was VERY expensive back then and the computer had plenty of it.

as with everything in life, there was give and take in the design.

User avatar
ehasting
Posts: 445
Joined: Sun Oct 26, 2003 4:00 pm
Location: Norway
Contact:

Re: Masterverb Test Thread

Post by ehasting » Sat Oct 05, 2013 9:02 pm

Just to throw in - i have the same error message (PCI error) every time i start up scope.
Two click on yes and its all fine for the rest of the session.

I have always thought its suppose to be like that :).
Higen
http://www.scopeportal.com - Community Sonic Core SCOPE Plugins and Devices
http://www.higen.org

User avatar
Bud Weiser
Posts: 1417
Joined: Tue Sep 14, 2010 5:29 am
Location: nowhere land

Re: Masterverb Test Thread

Post by Bud Weiser » Sun Oct 06, 2013 5:55 am

garyb wrote:all of the above.
ahh,- o.k. ... :lol:
garyb wrote: the i/o plates were all available as a choice. the i/o plate with RCA(cinch) and no z-link connections was called "classic20".
Then I´d call it a "SCOPE Pro Classic" because it´s a gen 2 even it has 15DSPs and vs a "SCOPE Pro Plus" (balanced analog) and "SCOPE Pro Z-Link".
That are the names I´ve found in a "Creamware Audio Products" PDF.

thx

Bud

User avatar
Bud Weiser
Posts: 1417
Joined: Tue Sep 14, 2010 5:29 am
Location: nowhere land

Re: Masterverb Test Thread

Post by Bud Weiser » Sun Oct 06, 2013 5:58 am

garyb wrote: yes, the 15th dsp is redundant which is why it was removed by Creamware to save money in their last days.
Why does or did it exist at all then ?
Became it redundant by software changes ???
When it did nothing at all always, Creamware would have been able to save money straight from the beginning,- no ?

just curious

Bud

User avatar
Bud Weiser
Posts: 1417
Joined: Tue Sep 14, 2010 5:29 am
Location: nowhere land

Re: Masterverb Test Thread

Post by Bud Weiser » Sun Oct 06, 2013 6:11 am

garyb wrote:sure, but it was VERY expensive back then and the computer had plenty of it.

as with everything in life, there was give and take in the design.
Now, we have 12x 32MB RAM in the XITE-1,- 384MB of unused memory ...

Wouldn´t it be an idea to use that for reverbs and/ or all of the memory intensive devices to reduce PCI / PCIe traffic to the computer´s memory.
I don´t think it would be possible to make the computer RAM obsolete but I think it would help to use XITE RAM up to that point where the computer RAM becomes essential.

I´m not able to describe that like an engineer does but as a musician I could imagine modulation-devices using short delays or stuff producing early reflections or smaller rooms as well as all the chorus stuff in synth devices could use that RAM we find in newer SHARC chips.
I don´t know how much RAM is needed for delays beyond 500ms, but eventually everything up to 500ms might work w/ XITE - internal RAM.

There´s a lot unused potential in the XITE up to now as we also see w/ DJMicron´s optimized devices.

Bud

User avatar
garyb
Posts: 21562
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: Masterverb Test Thread

Post by garyb » Sun Oct 06, 2013 10:03 am

yeah well, when things are designed the engineers try to think about the future, but at the same time, they try to make something that works for everybody and is backwards compatible. no doubt there is untapped potential and this could be tapped potential if someone wanted to pay for it.

Analog_Minded
Posts: 18
Joined: Sun May 26, 2013 3:04 am

Re: Masterverb Test Thread

Post by Analog_Minded » Sun Oct 27, 2013 1:41 pm

Ok so here are my test results:

after 11 masterverbs i got pci overflow massage.
How's that performance wise?

My specs:

GA p35 ds3 rev2.0
Q6600
2x 1gb ddrII 6400 cheap ass ram
hd 6950 (will be replaced by fan less card)
2 pulsar II and & Luna II (15 dsp)
1 uad I (other 2 out of system for the moment)

User avatar
garyb
Posts: 21562
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: Masterverb Test Thread

Post by garyb » Sun Oct 27, 2013 4:29 pm

good.

User avatar
Stevil
Posts: 85
Joined: Fri Mar 19, 2004 4:00 pm
Location: palm desert, ca
Contact:

Re: Masterverb Test Thread

Post by Stevil » Thu Nov 14, 2013 11:20 am

i posted this in my build thread, thot i'd share here. PCI caps out at 13. follow the link for system details.

User avatar
Shamk
Posts: 66
Joined: Tue May 14, 2013 10:56 pm
Location: Toulouse, France
Contact:

Re: Masterverb Test Thread

Post by Shamk » Thu Dec 05, 2013 10:45 am

1) MSI P45 NEO3 FR
Intel E8500 on LGA775 (4 PCI)

2) 3x 6DSP Pulsar II on scope 5.1 - 64 bits

4) 12 Masterverbs (13 MV cause PCI overload)
Creamware Scope 5.1 (Scope Power Pulsar 1.01|3 x Pulsar II|Pulsar I|Syncplate), scope abuser since 2004.
Asus Z87-K C2|Intel Haswell Core i7-4770|Kingston KHX1600C9D3K2/8GX - 2 x 4 Gb Ram (1600 MHz)
13 MV @ Masterverb Test

User avatar
Shamk
Posts: 66
Joined: Tue May 14, 2013 10:56 pm
Location: Toulouse, France
Contact:

Re: Masterverb Test Thread

Post by Shamk » Thu Dec 05, 2013 1:48 pm

1) Asus Z87-K C2
Intel Haswell Core i7-4770
Kingston KHX1600C9D3K2/8GX - 2 x 4 Gb Ram (1600 MHz)

2) 3x 6DSP Pulsar II on scope 5.1 - 64 bits

3) Integrated graphics on i7 CPU.

4) 13 Masterverbs (14 MV cause PCI overload)


Now is the time to fight against IRQ Sharing ! :(

@diegomaggi : how did you manage the sharing ?
Creamware Scope 5.1 (Scope Power Pulsar 1.01|3 x Pulsar II|Pulsar I|Syncplate), scope abuser since 2004.
Asus Z87-K C2|Intel Haswell Core i7-4770|Kingston KHX1600C9D3K2/8GX - 2 x 4 Gb Ram (1600 MHz)
13 MV @ Masterverb Test

Post Reply