Topic: RME Support please help on SB motherboard PCI issues!!

Is there anyway we can get a sticky for the topic of Sandy Bridge (socket 1155) motherboards PCI slot issues.  Many people want to upgrade to sandy bridge, or want to buy a HDSP PCI card (the 9632 in my case), but are weary due to the PCI issues most people have had.

The product testers or at the very least the great staff who works on the kick ass drivers that make RME what they are must have ran the RME HDSP 9632 on socket 1155 motherboards, right?  Is there any way we can get some feedback in regards to their experiences with certain 1155 motherboards having no issues with latency or even getting the RME PCI cards to work on them.

Ive read in a few places that the GIGABYTE GA-Z68X-UD4-B3 is working pretty well.  Other than that I haven't seen much on the subject.

Re: RME Support please help on SB motherboard PCI issues!!

I'm afraid I only have a first generation i7 board here, not sure about our developers' PCs at the moment. Can you provide some links referring to the issues in question?


Regards
Daniel Fuchs
RME

Regards
Daniel Fuchs
RME

Re: RME Support please help on SB motherboard PCI issues!!

Why don't you just use the Gigabyte board that was also recommended to you on Gearslutz? wink

If some boards work and others don't then it's more an issue of the new boards than the RME drivers (especially since these worked fine in older boards).

Re: RME Support please help on SB motherboard PCI issues!!

There are two issues with Sandybridge motherboard PCI slots.  First is that they are NOT native PCI, instead they use a bridge chip which I imagine is what the first poster is asking about (Do RME PCI cards work correctly in Sandybridge motherboard PCI slots?) and (What if any latency is added to the RME PCI card(s) because of the bridge chip?).  Next, because of the bridge chip, all the Sandybridge PCI slots SHARE A SINGLE IRQ (How does that effect RME PCI cards, especially with two or three RME PCI cards installed?).  Since you have a test machine running Sandybridge, have you put an RME PCI card in it?  Multiple RME PCI cards in it?  If so did they run without problems and if so what was the latency if latency increased?  The only difference between your first generation Sandybridge motherboard and the current generation is the fix of the SATA hard drive controller chip bug so your findings should apply to the current generation motherboards.  I currently run a three RME 9652 PCI card setup on an old single core AMD rig (I need that many outputs because I use it with a large frame analog mixer).  I'm thinking about upgrading computers also and a Sandybridge 2600k rig is one of the better bang for the buck setups from Intel right now (in particular I was looking at the Asus P8P67 (just the standard version of the board), it has three PCI slots (http://www.asus.com/Motherboards/Intel_ … /#overview).  AMD's Bulldozer line is due out in September so I'm keeping my eye on that platform also to see if a three PCI slot motherboard is available.  With either platform I want to continue using my three RME 9652 PCI setup and any info on RME PCI compatibility/performance would be helpful as I'm not a crash test dummy kind of guy.  While I'm on the subject of running multiple RME PCI cards I have a couple more questions.  Not that long ago you put out beta drivers that allowed for a six RME PCI setup, did this become a permanent feature?  What were the results of using that many cards?  What results have you had with using PCI slot expansion boxes and filling them with RME PCI cards?

Re: RME Support please help on SB motherboard PCI issues!!

I have a pair of Dynaudio bm5a, turntables, mixers, a couple controllers among other devices collecting dust due to frustration over my shit P4 running a bunch of not so stable vst.  So I have been strictly djing dubstep and dancehall the past 2 years or so.  Was trying to squeeze a PC, the RME 9652 and Cubase 6 out of $2000 or a little more to start back on it production wise.  I had my heart set on the 2600k, seeing how well it has preformed in benchmarks and the raving reviews from some fellow producers.  As PCI looks to be going the way of IDE, and seeing how the RME 9652 has been out for 7+ years and is still a solid card, I would like to (hate this damn term) future proof my audio interface.  Hopefully PCI-e will continue past the upcoming 2011 Ivy Bridge motherboards, as PCI 3.0 is on the horizon.

2600k for me.  Smiles all around!

Re: RME Support please help on SB motherboard PCI issues!!

IIRC, the Q67, Q65, and B65 do offer a single legacy PCI lane (for "business/workstation" MoBo's) - but Intel's own block diagrams don't confirm this?  I can find numerous blog postings that state this, but Intel doesn't seem to confirm this.  Go figure...

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8

Re: RME Support please help on SB motherboard PCI issues!!

The Q67 chipset won't allow for any overclocking (probably so with the Q65 and the hard to find B65) so anyone going that route save your money and buy a locked CPU (non "K" chip) so no real solution using them. ---- Bump ---- Any word on answering any of my above questions?  Did you do any testing of multiple cards?

Re: RME Support please help on SB motherboard PCI issues!!

Seems like this is an Intel issue and not an RME issue IMNSHO.  RME's PCI cards operate within the legacy PCI spec.  How the Sandy Bridge PCIe-to-PCI bridge chips would affect this is anyone's guess.  I'd liken this to the Firewire controller debacle of the past decade IMNSHO.  All RME can do is design to the standard spec and hope these types of bridge chips won't break standard functionality.  RME certainly can't make IRQ's out of thin air if the bridge chip is forcing a single IRQ on multiple cards...

You could always go with an x58/1366 system which has better RAM bandwidth than SB and still get native PCI support (I run such a system with the P6X58D-E, but just swapped to the HDSPe MADI card only for its added features above the HDSP MADI card, not because I specificaly wanted a PCIe card).  Microcenter in the USA sells the i7-960 for $179 all day long.  Hell of a deal IMO...

Last option?  Go PCIe for your modern cutting-edge Sandy Bridge PC, or take a chance with the PCI Bridge Chip.

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8

Re: RME Support please help on SB motherboard PCI issues!!

Any new information on this topic?

Re: RME Support please help on SB motherboard PCI issues!!

How about this: Startech PCI to PCIe adapter.

I haven't tried it but at least one poster on this forum has and posted that it worked for him. Unsupported by RME, of course...

PC1 = HDSPe PCIe: DF-ADI-8 DS / HDSPe PCIe: MF2
MBP = HDSPe Expresscard: MF1

11 (edited by SLL 2012-02-24 04:19:26)

Re: RME Support please help on SB motherboard PCI issues!!

The StarTech adapter is still a bridge. Buf of cause, the bridging firmware could be made better on the adapter, and by that make less problems for legacy PCI cards.

But I agree on the topic. Lots of us are still using RME HDSP PCI cards. There is also lots of PCI -> PCIe topics here on this forum and on gearslutz, where you can find lots of info on what mobo to buy. The early favour SB mobo with PCI slots was the mentioned GIGABYTE GA-Z68X-UD4-B3, which I think should work out fine. But I sense, that there is a new boss in town... the ASUS P8Z68-V PRO/GEN3 (or deluxe model). This is also the board I would go for, if I wasn't thinking on SB-E :-) Could be, that I'm forced to choose the Z68 mobo, because a combined SB-E DAW / gaming rig will cost a fortune.

But for me, there is a very important thing, that I really can't find anything about. This is the low latency performance of bridged PCI slots with RME cards. How low can you go... so to speak - with the buffers, before pops clicks appears in the audio path?! I won't be too happy (actually crying), if I can't run any projects (smaller or bigger) at a low 64 or 32 buffer setting with my HDSP 9652. This would make me sad, wanting a PCIe card (could be, that this is what RME is waiting for) :-)

SLL

RME Raydat, Asus Z370-A prime, i7 8700, Noctua D15s cooler, Corsair RM850x, Crucial 32GB DDR4 ram

12 (edited by Randyman... 2012-02-29 03:37:31)

Re: RME Support please help on SB motherboard PCI issues!!

I'm not taxing the I/O in my mixing DAW (P8Z68V-Gen3 w/ 2500K @ 4.5GHz) - but it gladly plays back projects @ 32-Samples w/o a peep.  My "big boy" tracking DAW (x58 / i7-930 @ 4.3GHz) uses quite a bit more CPU with the same project @ 32-Samples, but then it has a HDSPe MADI card - so lots more overhead and "active" ASIO I/O going on.

That said, I don't track on my mix (2500K) system, so any bridge-chip issues would only affect playback...

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8

13

Re: RME Support please help on SB motherboard PCI issues!!

Randyman, I guess you're using a RME legacy PCI card, and nice to know that you then actually can playback projects at a 32 samples buffer, which is the smallest buffer setting on my RME HDSP card, maybe on all RME cards?

When you say "plays back projects...", do you mean bigger projects like with more audio tracks, VSTi's, plugins? If so, and you're using a legacy RME PCI card, that we now for sure know, that the Asus P8Z68V-Gen3 + RME PCI audio plays back projects 100% stable without any audio pops or glitches, and at a lowest buffer setting of 32 samples. And this is through an onbaord a PCI -> PCIe bridge chip. This is great info :-)

Then I know, that my HDSP card will function in this mobo, probably at all buffer settings.

An interesting thing could be, to load up bigger projects on a 32 samples buffer, and see how the audio behave. Could be nice to know when the audio starts to burp and mess up the audio quality.

SLL

RME Raydat, Asus Z370-A prime, i7 8700, Noctua D15s cooler, Corsair RM850x, Crucial 32GB DDR4 ram

Re: RME Support please help on SB motherboard PCI issues!!

Yep - "Legacy" PCI HDSP card on a my new z68/2500K (which means PCI bridge chip by defintion) running 32-Samples does fine for playback of large projects in Nuendo 4 - mainly Audio Tracks with tons of Groups/Busses and oodles of VST Plug-ins running.  I have a separate PC for VSTi's and such.

I did mention that the overclocked 2500K/1155 seemed to use less CPU % than my overclocked "big boy" i7 930/1366 - well that was not the case.  The 1366 still edges it out at similar CPU speeds.  I'm assuming the 1366's insane RAM bandwidth (Tri-Channel) and overclocked QPI really helps in this respect (as I've witnessed in the past).  I don't think the PCI Bridge Chip is causing any bottlenecks or otherwise affecting performance of my 2500K/z68 - but again - this is only with a few output channels "Active" comapred to tons of active I/O on the 1366 Tracking Rig.

The fact that "Overclocking" a Sandy Bridge "K" CPU only affects the CPU's frequency does not offer the same Overclocking benefits as older i5/i7 (and C2D/C2Q) setups where you raise the bclk (base clock) and the FSB/QPI/DMI and RAM bandwidth all get a nice bump in performance along with the CPU Frequency.  I believe that is what allows the older 1366/i7 930 to still out-perform my 1155/2500K at similar CPU clock speeds...

You can still score a killer deal on a 1366 system (Microcenter had the i7-960 for $179!!!) - and you get native PCI as a bonus.  1366 Might be food for thought if you might need large I/O counts over PCI.

Or go for LGA2011 like I just did (still waiting to finish the build)!  Spanks the 1366, too!!! smile

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8