New PC -> Error: PCI Bandwidth limit reached - HELP!

[Sheldon]

12 is pretty healthy. a system should run great even if it only does 6 or 7. it really depends on how you use your stuff.

if you use resources stupidly, in ways that make no musical sense, then there's never enough resources....ever.

So 12 masterverbs is limit for pci bandwidth? Does the pulsar 2 use 32 bit 66 Mhz PCI bus?

On 9 dsp i 11 masterverb, on 21dsp I did 12.

Frankly I don't know how much Bw each takes. But 12 is probably rather close to the limit. Maybe garyb have some insight here?

When you load 12 masterverbs, what was the dsp load?

[garyb]

pci is a 33mhz clock.

why are you worried about using 12 reverbs?

[ehasting]

on 9, the dsp load is max i think.. on 21dsp the dsp is not even half.
I can max out the dsps on the mother board, 12 masterverb is more then plenty!

And the PCI bandwidth limitations will probably never ever be a problem to me.. i need max 2 reverbs pr. project.. in most cases 1

[Sheldon]

pci is a 33mhz clock.

why are you worried about using 12 reverbs?

So all pulsar2 cards use pci bandwidth about 133 MB/s, but not 266 MB/s? I just try to understand where the reason of bandwidth lack on sandy bridge platforms. I had a thought that pci-e-pci bridge is locked to only 33 Mhz, but pulsar2 uses 66 Mhz bus and thats why there is bandwidth lack (about twice decreasing). But if you say the card uses 32 bit 33 Mhz bus i'm confused. I wrote to asrock tech support but they are keeping silence.

on 9, the dsp load is max i think.. on 21dsp the dsp is not even half.
I can max out the dsps on the mother board, 12 masterverb is more then plenty!

And the PCI bandwidth limitations will probably never ever be a problem to me.. i need max 2 reverbs pr. project.. in most cases 1

It looks like that 'resources' of 12 dsps will be limiting the bandwidth of pci.

[garyb]

i think you misunderstand.

reverbs are a measure of bandwidth because there are thousands of delay lines requiring communication between the card and system memory and that must happen via the pci bus. 12 masterverbs indicates a very healthy amount of bandwidth. if you are only streaming asio, this doesn't even matter. EVERY process that uses the LIMITED resources(ALL resources in EVERY system are limited) will limit what's available for all other operations.

the bandwidth on the 1155 motherboards is poor because the method of connecting the bus to the rest of the system is inneficient, not because of the bus clock speed. in any case, there's nothing to understand about it, because understanding the exact reason won't make it work any better. Sonic|Core tech support isn't going to give any further information about this, it's pretty immaterial. the motherboard either works, or it doesn't. the real problem is that in the latest motherboards, the industry(intel) decided that since the pci bus is heading the way of the isa bus, that they only needed to support basic legacy devices(i'm sure eliminating pci support from the chipset and letting the motherboard manufacturers use a 3rd party's chip to control pci traffic makes sense as far as making the new system of pci-e even more robust, since it's one less job to be handled, if pci doesn't matter. timing is everything in the flow of data, especially for a realtime audio system). since there have been lots of complaints(this doesn't just target Scope cards), there are chipsets with natively supported pci busses being developed and produced. since we haven't seen them at this point, there's no telling how well they will work, but my guess is that they'll be fine. you probably have to wait a bit for such a socket 1155 motherboard to be common, but i'm sure it won't be long. then again, i always like to wait a few bios updates before buying a motherboard with the latest chipset.

ultimately, we're talking about cards that in many cases are 10-15 years old and older. the socket 1156 and 775 motherboards offer almost identical performance in the REAL world as the newer offerings. there isn't really any problem. eventually, if you want to use your 15-20 year old sound card, you'll have to put it in an older machine. there are pci-e Scope products avialable NOW that are not subject to pci limitations. this is how it is, a tempest in a teakettle.

[Sheldon]

i think you misunderstand.

reverbs are a measure of bandwidth because there are thousands of delay lines requiring communication between the card and system memory and that must happen via the pci bus. 12 masterverbs indicates a very healthy amount of bandwidth. if you are only streaming asio, this doesn't even matter. EVERY process that uses the LIMITED resources(ALL resources in EVERY system are limited) will limit what's available for all other operations.

the bandwidth on the 1155 motherboards is poor because the method of connecting the bus to the rest of the system is inneficient, not because of the bus clock speed. in any case, there's nothing to understand about it, because understanding the exact reason won't make it work any better. Sonic|Core tech support isn't going to give any further information about this, it's pretty immaterial. the motherboard either works, or it doesn't. the real problem is that in the latest motherboards, the industry(intel) decided that since the pci bus is heading the way of the isa bus, that they only needed to support basic legacy devices(i'm sure eliminating pci support from the chipset and letting the motherboard manufacturers use a 3rd party's chip to control pci traffic makes sense as far as making the new system of pci-e even more robust, since it's one less job to be handled, if pci doesn't matter. timing is everything in the flow of data, especially for a realtime audio system). since there have been lots of complaints(this doesn't just target Scope cards), there are chipsets with natively supported pci busses being developed and produced. since we haven't seen them at this point, there's no telling how well they will work, but my guess is that they'll be fine. you probably have to wait a bit for such a socket 1155 motherboard to be common, but i'm sure it won't be long. then again, i always like to wait a few bios updates before buying a motherboard with the latest chipset.

ultimately, we're talking about cards that in many cases are 10-15 years old and older. the socket 1156 and 775 motherboards offer almost identical performance in the REAL world as the newer offerings. there isn't really any problem. eventually, if you want to use your 15-20 year old sound card, you'll have to put it in an older machine. there are pci-e Scope products avialable NOW that are not subject to pci limitations. this is how it is, a tempest in a teakettle.

Thank you for comprehensive reply. But i want to detail some things. Lets have a look to the chain of all stuff interaction

Pulsar 2 <-133 MB/s-> Asmedia ASM1083 bridge <-1x PCI-E 1000 MB/s-> P67 Chipset <-20 Gb/s DMI-> CPU <--> Memory

As you can see something wrong with realization pci-e - pci bridge, intel chipset provides a huge chanel to pci needs, but it's not used. I agree that the one thing that appears in comparison with chipsets with native pci support is additional latency, but lack bandwidth it's not fault of intel chipset. I think we can guess that 12 masterverbs requires all the pci bandwidth so 1 masterverb needs about 11 MB/s. I can only load 4 masterverbs so it's about 44 MB/s. Why? It seems to me that asrock developers just share 133 MB/s on 3 slots on my motherboard.

[garyb]

because it really doesn't work like it should.
you can guess and theorize all you want.
you can lay blame anywhere you feel that it's appropriate.
it still won't work.

perhaps, it's a run of bad chips.
perhaps it's faulty engineering that doesn't really meet pci spec(i'm betting on this one, since the cards work with so many other designs).

it would take a lot less time to just put the card somewhere that it can be happy.

it's not JUST bandwidth. it's timings and access too, and in synergistic combination.

perhaps there's a new bios that will fix the problem.
perhaps your motherboard is faulty.

in any case, if you want to use the card, you need to do something else if the regular fixes(bios setting, bios rev, power settings and irqs) don't help. there's no way for you to reengineer the system. if you could do it, you surely wouldn't need my(or anyone else here for that matter) input or insight. that's for sure!

by the way-i'm almost POSITIVE that NO motherboard reaches the true theoretical pci bandwidth limit. figures like that are a myth. even the length of the traces on the motherboard and the proximity of devices on the motherboard effect the actual usable bandwidth. all motherboards are a comprimise. manufacturers optimize performance based on what they think the users want and need. sometimes they just make crap.

[Beacon]

Hallo everyone !
(I'm newer here)
I must reprise the topic to make a question:
it's possible to resolve the problem with this adapter ?
http://www.startech.com/Cards-Adapters/ ... d~PEX1PCI1

My configuration:
- Asus P8Z68-V LE
- Intel Core i7 2700K (Sandy Bridge)
- 2 Pulsar II
- 1 UAD1

... or is it better to wait for a patch ?
Thank you

[garyb]

i would think that later bios updates will improve performance.
i'm just waiting to hear about such good news. i don't think that there is much that S|C can actually do about this as it's a manufacturing decision that the motherboard designers make/made.

[Beacon]

i'm just waiting to hear about such good news.

... me too ...
In the meantime, what do you think about the PCIe to PCI Adapter Card ?
(pulsar II is a low profile PCI card ?)

[garyb]

nope, not low profile.

the adaptor will probably work, but basically, it's the same thing as what's happeneing on the motherboard. an extra chip is required to handle the pci traffic and then the data is fed to the pci-e slot. that's the design problem. the motherboard's chipset no longer handles pci traffic or has an actual pci bus. the pci to pci-e adaptor is built-in to the motherboard. chances are, that just like the motherboard's pci slots, you'll get less than ideal performance. depending on how you use the system, you may or may not have a problem. then again, some of these newer motherboards have had very poor pci performance, so an adaptor might help.

you won't be able to just use a slot adaptor without heavy case modification, unless an extrenal box is involved. low profile cards didn't even exist when Scope card were made. depending on the external pci-pci-e box used, the perfomance might be pretty good, but that kind of box isn't cheap.

[Beacon]

Thank you Garyb.
It's time to move on Xite ...

[djmooncat]

After buying a new system and not really thinking about the requirements of my scope PCI cards (I should have had look at the trusty planet Z forums) I got an ASUS P8Z77-V LE which has 3 PCI slots, which I thought would be perfect.

Well I can confirm GaryB's information from Sonic Core these 1155 chipset/motherboards do not work well with a SCOPE PCI card. I get PCI time out errors with only 4 reverb devices and the problems start very soon when using one the synths. In fact I had one delay a mixer and a synth and I got the PCI error, I've done the BIOS settings but made no difference

I forgot about all this PCI bandwidth stuff, my previous system has been running along nicely for years, so I think I'll probably put my cards back in there and not use my new system for my DAW

Bit gutted but then I didn't do my usual thorough checking as I was in a rush to replace the old system for some application development stuff I'm doing which needed more CPU horse power for running lots of virtual machines. I'd assumed I'd be able to do both. But that's not the case, slightly expensive mistake.........

Cheers,

Mark

[m.my91]

Hallo everyone !
(I'm newer here)
I must reprise the topic to make a question:
it's possible to resolve the problem with this adapter ?
http://www.startech.com/Cards-Adapters/ ... d~PEX1PCI1

My configuration:
- Asus P8Z68-V LE
- Intel Core i7 2700K (Sandy Bridge)
- 2 Pulsar II
- 1 UAD1

... or is it better to wait for a patch ?
Thank you

hello man
i had your motherboard "p8z68-v lx" and always had "pci limit reached" error with this motherboard.
i understand after that in the new motherboard there'is a "chip" beetween pci bus and processor witch causing
this 'pci limit reacch" error.
so i find a new intel chipset with a "NATIVE PCI" (pci connect to the bus like the old computer).
this chipset is the B75 B77.
forget the Z68 for creamware card.

[garyb]

i'm sorry to tell you that the PCI slots on socket 1155 motherboards just don't work correctly.
the newer socket 1150 motherboards generally work quite well.

[m.my91]

Hi all,

My bandmate bought a new PC but gets PCI-bandwith errors. he doesn't speak english so I offered him to put this help request on planetz ... I am a using RME stuff but do also own to pulsar 2 cards, so I am a little bit familier with scope ...

Ok, we did some testing on MY old system and everything worked as expected even a16 connected to one of my 2 pulsar cards. so I thing A16 should be OK.

Some info about the new PC:

Hardware:
ASUS P8P67, i7, 16 GB RAM,3 HD,Scope 15 DSP,A16 ultra,3 monitors 24",ATI Radeon 6700 series

Software:
Windows 7 64 bit,Cubase 64 bit,Scope 5.1,
Omnisphere, Trilian, RMX 64 bit plugins

BIOS already updated
graphics card driver up to date.
A16 checked on an old system - excellent

A16Ultra connected as slave to Scope card (Master) via ADAT. Sync via ADAT, 44,1 KHz.
All unnecessary Services disabled
all energy savings disabled (in Windows and BIOS)
bluetooth disabled, realtek audio disabled, usb 3.0 support disabled, 1394a disabled ...

Error: PCI Bandwidth limit reached
1. we get this message even with only 8 channel connections routed directly inout (without any other mnodules, synths ...
2. Using the scopecard alone works like a charm
3. wihout starting cubase we are also able to load many modules.

as soon as we start cubase the possibility to chrash/get the above error message increases. we can provoke the error just by playing on the keyboard (e.g. modulation wheel ) and sending midi. after a few seconds the message appears. sometimes the pc even crashes (BSOD

We have already contacted SC support - no way ... they cant't help.

any ideas, hints on what we could do?
should we try 32-bit cubase?does that make sense?
Is there a way to force the systems to use an interrupt exclusively for the slot with the card (ASUS manual shows serveral "shared" in the IRQ-matrix)
greetz
tom

attached 2 pics of routing an error message

hello man
i think it's a problem with your motherboard.
p67 not compatible with old pci gear:
http://semiaccurate.com/2010/08/19/inte ... i-support/
some new motherboard doesn't integrate "natice pci bus" like the old computer.
the new motherboard add a chip between pci bus and the processor and the creamware card doesn't understand this chip message.
i know that intel build a new chipset especialy for using old pci card in a new computer.
this chipset are the B75 or B77.

[m.my91]

Hi all,

My bandmate bought a new PC but gets PCI-bandwith errors. he doesn't speak english so I offered him to put this help request on planetz ... I am a using RME stuff but do also own to pulsar 2 cards, so I am a little bit familier with scope ...

Ok, we did some testing on MY old system and everything worked as expected even a16 connected to one of my 2 pulsar cards. so I thing A16 should be OK.

Some info about the new PC:

Hardware:
ASUS P8P67, i7, 16 GB RAM,3 HD,Scope 15 DSP,A16 ultra,3 monitors 24",ATI Radeon 6700 series

Software:
Windows 7 64 bit,Cubase 64 bit,Scope 5.1,
Omnisphere, Trilian, RMX 64 bit plugins

BIOS already updated
graphics card driver up to date.
A16 checked on an old system - excellent

A16Ultra connected as slave to Scope card (Master) via ADAT. Sync via ADAT, 44,1 KHz.
All unnecessary Services disabled
all energy savings disabled (in Windows and BIOS)
bluetooth disabled, realtek audio disabled, usb 3.0 support disabled, 1394a disabled ...

Error: PCI Bandwidth limit reached
1. we get this message even with only 8 channel connections routed directly inout (without any other mnodules, synths ...
2. Using the scopecard alone works like a charm
3. wihout starting cubase we are also able to load many modules.

as soon as we start cubase the possibility to chrash/get the above error message increases. we can provoke the error just by playing on the keyboard (e.g. modulation wheel ) and sending midi. after a few seconds the message appears. sometimes the pc even crashes (BSOD

We have already contacted SC support - no way ... they cant't help.

any ideas, hints on what we could do?
should we try 32-bit cubase?does that make sense?
Is there a way to force the systems to use an interrupt exclusively for the slot with the card (ASUS manual shows serveral "shared" in the IRQ-matrix)
greetz
tom

attached 2 pics of routing an error message

B75 B77
"LEGACY PCI NATIVE"
not
BRIDGED PCI make this error becauce bridged pci use pci-express bus for feeding creamware card and it doesne't work well.
phil

[garyb]

why do nyou continually give out false info?

Z87 and Z97 have bridged PCI to PCIe and they work fine.