DSP Overload message

[m.my91]

i fix the problem with a scope 15 dsp and a pulsar 2 .
i always had dsp overload at the middle of my dsp power.
and now it work good.
i do as the manual say.
1: put the scope on pci 1 computer slot.
2: put the pulsar in pci 2 computer slot.
connect two tdm cable (with one tdm connector in the middle unplug).
after installing scope go on "cset ini".
and modify like this:

[Host-Config]
Numboards=2 (because you have two card scope only install one!)
Board0=Host (0 :scope 15 dsp is the host card)

[Board0] ("mix the dsp of the two card in one" pulsar 6 dsp first and scope 15 dsp second in the dsp meter of scope)
Boardid=1

[Board1]
Boardid=0


like this you say to the system that there are 2 cards on the computer and the plugins load work good until 90 % of your dsp power.
if you don't have a scope 15 dsp alway put the biggest card in pci one and the smallest after.

for a pulsar2+pulsar+luna

i would make this:0:(pulsar 2 on pci 1,1: pulsar on pci 2,2:luna on pci 3 (2 stdm cable).

[Host-Config]
Numboards=3 (because you have 3 card )
Board0=Host (0 :pulsar2 is the host card)

[Board0] ("mix the dsp of the the pulsar2 and luna)
Boardid=2

[Board1] (pulsar 4 dsp in the middle pass through dsp)
Boardid=1

[Board2] (luna 2 and pulsar 2 mix dsp)
Boardid=0

(don't forget the jumper on the pci card if you work on xtc mode).
for me it works .
phil

[Marco]

I hope this will work on mine too. But this shows the Problem with scope, this should be automatic done by the Software itself, but there is no update now, because nö employes. But I want to use scope for music production and not for seaching and solving Probleme. Now rme card is bought, and we will see what is going on....maybe a combination of both cards will solve the Problem with scopes pci bottleneck. Otherwise my patience is empty and scope leave the house!

[Bud Weiser]

I hope this will work on mine too. But this shows the Problem with scope, this should be automatic done by the Software itself, but there is no update now, because nö employes. But I want to use scope for music production and not for seaching and solving Probleme. Now rme card is bought, and we will see what is going on....maybe a combination of both cards will solve the Problem with scopes pci bottleneck. Otherwise my patience is empty and scope leave the house!

Thx for your important info !
You better go and improve your english.
We already know SCOPE isn´t ideal for you anymore and you bought a RME card.

B.t.w. I have a RME card too, many here have.
It´s not all about RME OR S|C.
But never forget you run latest SCOPE on offically discontinued hardware.

Bud

[garyb]

this is an old hack. it does not refer the v7, afaik, and even if it did, first generation Scope cards will always reduce PCI performance.

[Marco]

Hallo Guys, with this board trick, I didnt really know until now, and this after so many years. The Scope seems to work proberly.

BTW I really live in germany, and sorry for the bad english, but I dont have friends to exercise. And just reading in the internet is not enough. So you must live with my bad english, included myself

But I hope my music is much better than mi english.

And NO I love scope, if it runs.

[nebelfuerst]

I also used this trick with renumbering the boards to make my 14DSP-board to be the master.
Nevertheless, the performance of Xite1 is a big step forward compared to pci.
As I'm still on pci, but I also own a Xite1, I just did the famous Masterverb-test on the same PC-Hardware ( Asus P7P55D ).
Result: Xite1 can handle 23 Masterverbs.

As my setup didn't reach the limits of pci yet, my migration towards Xite1 is a process of years

[Marco]

https://1drv.ms/u/s!ApPWzfjyiAbwmDOaSBobSyOJOQGl

My happy testsong made with 6 Hardware Synthesizers mixed in scope used some FX.

[Bud Weiser]

BTW I really live in germany, and sorry for the bad english, but I dont have friends to exercise. And just reading in the internet is not enough. So you must live with my bad english, included myself

same here !

But I hope my music is much better than mi english.

I hope too !

Bud

[Bud Weiser]

Result: Xite1 can handle 23 Masterverbs.

Masterverb test doesn´t count for XITE-1

Bud

[garyb]

Masterverb test doesn´t count for XITE-1

Bud

because PCIe has a bunch more bandwidth...

[valis]

And our legacy 'IRQ sharing' issues are also a thing of the past with PCIe, which is really another way of saying the same thing

[Bud Weiser]

Masterverb test doesn´t count for XITE-1

Bud

because PCIe has a bunch more bandwidth...

So, you say it still makes sense performing the masterverb test on XITE ?

IIRC, I got the same result dawman posted some time ago,- 60 masterverb /XITE-1,- even on my old WinXP machine,- ASUS P5WD2 Premium, Intel 955X chipset, PCIe 1.5, D945 dual core proc, Mushkin DDR2.
Then I wondered why I got so bad results w/ big ZARG synths and in opposite to other´s here.

Now I wonder why Nebelfuerst´s limit is 23.

Must be many (unknown) factors affecting performance.



Bud

[Bud Weiser]

And our legacy 'IRQ sharing' issues are also a thing of the past with PCIe, which is really another way of saying the same thing

Ehmmm ... what´s up w/ the USB controllers sharing IRQ w/ XITE PCIe card ?
Or graphics card and driver ?
Or M.2 NVME PCIe x4 drives,- which sometimes (system dependend) hog the PCIe bus, when the always busy OS is installed on these and possibly sharing lane(s) w/ PCIe x1 slot for XITE PCIe card ?



Bud

[garyb]

usb might make trouble, might not. it's best to disable those controllers that share with Scope hardware. there are always plenty in current motherboards.

WAAAYYY too much thought about this for XITEs. PCI cards are more sensitive. PCI is more limited in resources, and timing.

graphics are not usually a problem in current systems.

[Bud Weiser]

graphics are not usually a problem in current systems.

Good to know ´cause I wondered if usage of a dedicated graphics card OR the GPU (inside CPU) might make a difference.

And,- according to my second to last post,- I still wonder why there are such differences in masterverb test results possible on XITE-1.
23 vs 60 ... not peanuts IMO.



Bud

[garyb]

it's not significant.
both are beyond the practical limit. there are many factors including how the test is done(retry or not?), how well the motherboard works(performance is not always equal), and possibly the graphics card. generally, i believe cards are better, but onboard graphics are ok, too. in any case, PCIe generally does not have the issue. the whole drama of PCI overflow is certainly based around PCI cards. everyone is always trying to exceed limits, but the gain of doing so are often minimal. as you know, it's more about having a useful tool, than simply specifications.

believe me, stubbornness is a bigger issue. before all-in-one boxes, we spent our time on music and audio. there was no other choice. i've seen the music business(especially the music making business) go from a sound thing, to computers. i think priorities are bit bit twisted, i mean obviously computers are super useful, but for what purpose? well, the rant is probably out of place, but in the quote above, i was referring mainly to IRQ problems. IRQ problems may or may not be involved in PCI overflow messages, but only in the sense that when the Scope hardware cannot communicate with the processor in a timely fashion, data is lost(clicks and pops). if the timing is REALLY messed up, then the logjam can be interpreted as "not enough bandwidth" or PCI overflow, even if technically there may be bandwidth available. it's a real-time thing. audio is time dependent.

soooo-the difference in reports in not statically meaningful because it's only a few reports, but neither system would be prone to PCI overflow messages with an XITE. i have stacked a lot of masterverbs with XITEs, it gets boring putting all those things in the routing window. an XITE-1D can max out the dsps with masterverbs before overflow. a masteverb test is only meaningful with an XITE if the nuber is very small, which would mean a hardware or setup problem.

[Bud Weiser]

it's not significant.
both are beyond the practical limit.

ah ... o.k.

... in any case, PCIe generally does not have the issue. the whole drama of PCI overflow is certainly based around PCI cards.

So, when messages like "no more (SAT) connections between DSP X and DSP Y" appear,- this has nothing to do w/ "PCI" overflow,- possibly just only because we´re on PCIe w/ XITE-1 ?

... everyone is always trying to exceed limits, but the gain of doing so are often minimal. as you know, it's more about having a useful tool, than simply specifications.

believe me, stubbornness is a bigger issue. before all-in-one boxes, we spent our time on music and audio. there was no other choice. i've seen the music business(especially the music making business) go from a sound thing, to computers. i think priorities are bit bit twisted, i mean obviously computers are super useful, but for what purpose? well, the rant is probably out of place, ...

the latter doesn´t matter and I can only agree.

... only in the sense that when the Scope hardware cannot communicate with the processor in a timely fashion, ...

Well, isn´t it the same when these "out of connections" errors appear,- a communication error and timing issue ?

I pick on this because always, when I couldn´t load more, there were several,- at least 2,- warnings ...
a) DSP limit reached (... remove device etc.)
b) No more SAT connections between ...
Is there a correlation between a) and b) ?

if the timing is REALLY messed up, then the logjam can be interpreted as "not enough bandwidth" or PCI overflow, even if technically there may be bandwidth available. it's a real-time thing. audio is time dependent.

So, when I see DSP load of about 50% and cannot load more w/ these "connection warnings" coming up,- that´s a real-time issue too ?
You know,- I had such scenarion w/ some devices, possibly in that specific combination I used in project window at that time,- and even I was able loading Masterverbs ´til the cows come home.

... i have stacked a lot of masterverbs with XITEs, it gets boring putting all those things in the routing window.

Yes,- indeed !!!

an XITE-1D can max out the dsps with masterverbs before overflow.

Ahh,- didn´t know ... so it depends on number of (the large 333MHz ?) DSPs available and not on the 6 old 60MHz DSPs ?

Originally and when the "connection warnings" appeared, I guessed the old/small 60MHz SHARCs were the bottleneck/culprit,- but now it seems it can´t be because there´s identical number of these in both,- the XITE-1 and XITE-1D.

... a masteverb test is only meaningful with an XITE if the nuber is very small, which would mean a hardware or setup problem.

o.k., I see ...

best



Bud

[garyb]

sat connections are between chips in the XITE. there are a limited number of possible connections. this is based on the architecture of the Sharcs and is not related to PCI overflow.

PCI overflow is a limitation of the PC and "no more sat connections" is a limit of the Sharcs.

which dsps are loaded can really affect the sat limit...

xiteadsp.jpg (192.77 KiB) Viewed 7238 times

remember this map?

[Bud Weiser]

PCI overflow is a limitation of the PC and "no more sat connections" is a limit of the Sharcs.

which dsps are loaded can really affect the sat limit...

remember this map?

Yes, I do,- but it didn´t help always.



Bud

[garyb]

everything has limits.
some plugins need parts to be rebuilt for the XITE architecture. that's up to the plugin maker to figure out. many of the 3rd party plugins have XITE and PCI versions. many of the plugins that you might see limits because of sat connections are 15 years old or older.

nothing will help anything always, most limits are easily avoided, but may require choices. i know that it would be great if everything worked as one might imagine it, but on the other hand, maybe it wouldn't...

things that stretch sat connections to their limits might be best locked to a middle dsp, or something...