PlanetZ forums - scopeusers.com

ALL current motherboards that actually have PCI slots have worked well, from what i have seen.

back into the cw/sc after several yrs, i get 9 masterverbs on a 3x pulsar 2 cards in a magma pci expansion box.

intel q9550 cpu - quad core socket 755
asus p5q pro motherboard (p35), only 2 PCI slots, connected up to magma 4pci slot expansion box (housing 3x pulsar 2 cards and tc powercore)
4gb ram
windows 7 x86

that is pretty good. magma has always been the best expansion box. i guess the price shows it.
are you using pci to pcie?

garyb wrote: ↑Wed Aug 04, 2021 5:42 pm that is pretty good. magma has always been the best expansion box. i guess the price shows it.
are you using pci to pcie?

pci to pci interface atm. pcie card would be nice but rare to find 2nd hand and quite expensive. i would like to upgrade to a newer x64 system and move my noah box to a laptop. Pci slots are a quite rare now with all the latest cpus.

Motherboard ASUS Prime B460 plus
CPU Intel i9 10900K 3,7 GHz
RAM 64 GB (2 x 32) DDR4 Corsair Vengeance LPX 3200 MHz cas 16
Windows 7 PRO 64 bit
Scope v5.1
1 Scope Pulsar 2 (6 DSPs) at 44.1 kHz - ULLI latency 13ms

9 Masterverb and get power DSP limit with the 10th in empty project.

I tried also with Masterverb Classic: 9 are ok and get the PCI overflow with the 10th
While with Masterverbs receiving an audio track signal from Wavelab is ok with 7 of them, after that it finishes the DSPs;
9 with Masterverb Classic (it uses a bit less of DSP power than normal Masterverb) instead.


I guess I have to put also the second Pulsar 2 (or the Pulsar 1) I have as backup and redo the tests,
but it's not easy to fit it in the second slot because of the USB and SAT cables in their connectors and the case side that separates from the power PSU unit space.

However consider that I'm not using this new mobo at its best because not all the official Asus drivers for this motheboard install on Win7,
the executables are just for Win10, so it got drivers from the net; I just mounted the Win7Pro SSD I was using.
I wanted to test Cubase 10.5 and Scope inside the system I know well, before upgrading to Win10 Pro (and also because I have projects on working)... but jumping onto Win10 is on schedule because I'm also interested in getting Windows 11.


Anyway this Asus Prime B460 plus (thank you a thousand to the Scope users that discover and tried it ) seems to work great
and much better I think than what I was using before (MSI B75A-G43 since 2013) that with Masterverb test arrives at 6 only and then gives PCI overflow warning.
I remember that the old infamous Asus P4P800 got 8 or 9 Masterverb when I was using it, so I'm happy for now to not having to buy a new usb audio interface for the new modern CPU... almost double power for VST plugins compared to i7 3770K.

Love reading this, thanks.
I've just this afternoon built my Asus Prime B460 plus Win10 system and am typing from it now.
Next I'm now going to physically install my Scope cards so watch this space.

Good news to report:

Asus PRIME B460-PLUS
Intel Core i9 10900K
2x8G CorVenRGBPRO DDR4 3200WHI
Windows 10 Pro
Scope v7.0.1706
2 x Pulsar2 cards
11 Concurrent Masterverbs

The only thing that's weird is I installed the 64 bit installer from here: http://scope.zone/server/public/scope/s ... nglish.exe but in task manager Scope shows up as "Scope (32 bit)".

the 32bit part of the Scope app is simply a gui and loads the dsps. it has nothing to do with audio and it doesn't matter if it's 32 or 64bit. why make two versions? in 64bit Scope, everything that has to do with interfacing with the os is in 64bit.

Another test on

Asus PRIME B460-PLUS
Cpu Intel Core i9 10900KF 3.7GHz
DDR4 64GB G.Skill Trident Z Neo (4x16GB) 3200MHz CL16 1.35V XMP 2.0 [F4-3200C16Q-64GTZN]
Windows 10
Scope v7
Pulsar2 + Scope
11 Masterverbs

3 msec latency seems fine for now.

Using this E3 X11 era Xeon system, 11 Masterverbs (9 Pro) is where I hit PCI limit.

Running on two 6 dsps cards at present, awaiting my 3rd to be refurb'd.

Just optimized my ASUS P4C800 setup (Mobile P4, 2 GHZ, 2GB) with 3 Pulsar 2 cards:
16 Masterverbs, the 17th was unrecoverable.
Seems to perform better with the unique io format than with the former Pulsar One, Two and Luna mixture.
The mobile CPU has 1 pin bent to get down to 1.28 core voltage and emits almost no heat.
It‘s lame with modern VSTs, but that isn‘t it‘s job anyway.

cheers, Tom

Impressive

I've added your result to the spreadsheet, congrats on having the oldest motherboard on there : )

https://docs.google.com/spreadsheets/d/ ... sp=sharing

I just looked in the shared Google Sheets list of Asktoby and see the row of the Gigabyte X58 UD4 mobo.
I remember doing these MasterVerb tests on the same mobo around 2011 and was able to load at least 8 x MasterVerb Pro with 2 x Pulsar II + 1 x Scope (2nd gen). All 3 PCI slots and cards worked perfectly with Scope v5.

I also performed these Masterverb tests with 2 x Pulsar II + 1 x Scope (2nd gen) on a HP XW4600 Workstation from 2008 (Intel X38 chipset) + Intel CPU Q9550@3.4Ghz + 8GB RAM, Windows 7SP1 32bit, Scope v5 .1:

MasterVerbs
Normal load 15
PCI Capacity limit reached at 16

MasterVerb Classic
Normal load 16
PCI Capacity limit reached at 17

MasterVerb Pro
Normal load 13
PCI Capacity limit reached at 14

I also use a HP xw4600 workstation, Intel Core2Duo, 3GHz, 4GB RAM, Scope v7 PCI, Win7 Pro SP1 32Bit.
1Scope 15DSP (gen2), 2 Pulsar-2.

For a valid masterverb test,- do I have to just only load as much Masterverbs as possible into the Scope test-project, or do I have to connect all the already loaded masterverb´s I/Os to something,- p.ex. STM mixer´s send/returns,- and when these are used by the 1st 6 Masterverbs, the additional Masterverbs´ to STM mixer´s direct outs (sends) and common input channels (returns) ?
Do audiosignals have to be fed into the Masterverbs for the test ?

Going to page #1 of this thread, I don´t find any "how to" advice unfortunately.



Bud

Iirc the original instruction was to open an empty project with nothing at all inside and then add Masterverbs (not pro, not classic) until the PCI overflow message appeared.
The score was the number right before the fail.
After the fail Scope suggests to rearrange and that may deliver more instances (as the process may be repeated), but according to my experience it‘s rather unreliable, very system depending.

ps: it probably doesn‘t matter if it‘s a totally blank window or a small project, as long as there aren‘t Asio busses involved. High channel counts have a significant influence on PCI bandwidth.

Thx for reply astroman !

I guess Masterverb uses host computer RAM when at work in a real-world scenario,- no ?
This might not be the case when nothing is connected and consequently no audio data is running thru it´s algorithm.

ASIO doesn´t have to be in the ballpark at all,- but I think virtual cable connections in a "standalone" running SCOPE environment cost some DSP cycles too,- and when audio is running,- maybe coming from a media player playing back thru wave-driver and into multiple sends according to count of Masterverbs already loaded into the project,- cost DSP cycles in addittion and produce bi-directional traffic between Masterverbs and host computer RAM.

With SHARC DSP based technology, XITE-1 included, I recognized, in real world you can only load as much devices until Async is 100% on related DSP chips.
When it happens, you cannot load more devices on other DSPs even DSP load progress bar is somewhere between 50 and 100%.

So, what does it mean when one is able to load p.ex. 16 Masterverbs on a given system but w/ nothing connected to the devices being loaded ?
To me, it´s a theoretical result which has nothing to do w/ using SCOPE in real world.
With connections between devices and running audio/MIDI, the same system giving best results w/ "common" mastervewrb test, might behave different.

For DAWs, there are benchmark projects specifying count of an assortment of CPU intensive VST/AU plugins and tracks running audio.
Several projects,- light, medium and heavy.

Wouldn´t it be an idea finding a consensus about S|C Scope devices causing heavy load and trying to create test projects for typical PCI card combos as also XITE-1D and XITE-1,- all for Scope v7 and using these assorted stock devices ?

I also asked myself what the real bottleneck for SHARC DSP farm products is when being connected to host computers,- PCI card and PCIe card (as also host computer´s PCI/ PCIe slot) data thruput, interconnection of PCI cards (STDM cable), interconnection of DSP slots in XITE boxes,- or count of devices in a given project,- or Async.



Bud

it is actually a pretty accurate measure of performance. even if no audio, the communication lines are open.

The Masterverb test has 1 single purpose: check performance of the PCI bus (dataflow between DSP board and PC Ram).
This is classic Scope‘s weak spot.
Back then the alternative was to either use expensive local Ram (and management) or just transfer via PCI bus.
Even with local Ram a lot of data had to be passed to the CPU anyway and good PCI bus performance was a main focus of motherboard designers. Intel BX comes to mind...
So the decision made sense at that point in time.

As we all know this changed over time... (in general application) PCI was replaced by PCIe, lower voltages etc.
Iirc that‘s what triggered the Masterverb test as a simple tool available to any Scope user.
MV test only tells about the data transfer performance, nothing else.
Delay and reverb are the main aspect, but Asio channels fall into the same category.
If a project has 32 Asio channels, then bandwidth for reverb/delay is significantly cut down - or vice versa.

@ Gary and Astroman ...

thank you for details !
I got it now.



Bud