Page 1 of 1

Posted: Thu May 02, 2002 8:07 am
by liulai
Hi guys,
Just a quick thought, in desperate need for peer feedback. How quick and effective do you think CW tech support is?

As you may have read elsewhere, I haven't been able to configure and run my Pulsar II on my current Dual Ahtlon MP rig. I have tried calling in several times and all I was asking was a straightforward answer as to whether I will _ever_ be able to run my card on that system, without having to buy a separate system (which is paradoxical to say the least, since integration and space-saving are the very reasons why I bought my Pulsar II).

So far, more than a month later since my first attempts, no clear answer.

Thoughts appreciated,
Vanni

Posted: Thu May 02, 2002 8:35 am
by subhuman
In Creamware's Service FAQ, it is clearly stated (and has been for months) that they do not recommend dual processor systems currently. They list i815, i845, and i850 systems as the prefered PC platform. This is because certain aspects of the system rely on the chipsets to provide realtime access to main memory (reverbs, samplers), and most dual processor chipsets provide much less.

If you have a Tyan Tiger MP (rev 1, or rev2), you probably never get very good performance on your PCI bus, there were hardware issues which Tyan addressed with the Tyan Tiger MP rev1.03(rev3), which I have and use just fine, with only the occasional copy protection violation (caused by running in ACPI mode). If you have the Thunder, well I hate to say it, but that board has a lot of onboard devices that sit on the PCI bus, disabling them <i>might</i> make it useful for audio, but it's absolutely more for the server environment.

Creamware has also started asking the users of duals for help (on their own Creamware Forum) so they can build in better support for duals in SFP. This is an issue they do want to fix, but (and this is MY OPINION ONLY), many dual boards will probably never work very well due to the way the PCI bus is implemented. But you can be sure Creamware is doing what they can to make this work better...


<font size=-1>[ This Message was edited by: subhuman on 2002-05-02 09:36 ]</font>

Posted: Thu May 02, 2002 7:05 pm
by liulai
Refreshing answer, dear subhuman. I have noticed that the Tech Support guys (Rembert Gantke in person, who mailed me a link to a SPF beta!) are quite keen on testing Pulsar behaviour with Dual CPU systems...on our own skin!

I have spend 5 days (five!), daytiem and late nights trying to set up a decently working system. All I could incur were copy-protection errors and, when lucky enough, two ASIO channel stuttering something out-of-sync.

This is definitely not the kind of performance that one expects for _any_ PC configuration.

Also, I can repeat you that my Thunder K7 has never pointed me to any consistent PCI-related issue. The onboard ATI graphic chip is deactivated by jumper and takes no IRQ. The dual-SCSI channel takes a single IRQ and, as for the two LAN cards, I use them anyway (one goes to my Internet gateway Linux box, the other occasionally to my laptop).

I heard of the problems with the earlier Tiger MP but I have a quite late revision of the S2462 and don't think any of the issues I have dramatically experienced on my skin (and my HDs!) are related to the mobo as "it is". Perhaps things are better with the AMD760MPX chipset...

Posted: Fri May 03, 2002 9:16 am
by subhuman
Here is a list of hardware problems with the dual AMD boards

<a href=http://www.amd.com/us-en/assets/content ... pdf>AMD768 errata</a>

Here's a snippet if you don't feel like loading Ad*be Acr*bat Reader:

<i>
Table 2. Cross-Reference of Product Revision to Errata
Erratum Number and Description
Revision Number
B1 B2
10 Multiprocessor System May Hang While in FULL APIC Mode and IOAPIC Interrupt is
Masked X X
21 PCI Discard Card Timer Expiration Causes Data Corruption on PCI Primary Bus X X
22 Writes to Real Time Clock Must be Byte Width X X
23 AC ‘97 Modem Controller May Transmit Incorrect Data to AC-link X X
24 Single Processor Systems Will Not Resume from C2, C3, or POS Low Power States X X
25 PCI Bridge Does Not Handle Device Initiated Target Abort Cycles Correctly X X
26 USB Controller May Cause Secondary PCI Bus Contention X
</i>

Posted: Sun May 12, 2002 7:16 pm
by ohmelas
And the moral of the story is....

1. CW America has been kind, responsive and forthright. They're usually nice people. Paul's always been courteous and well meaning.

2. I've heard on this forum that CW Germany isn't always the greatest when dealing withe their American customers and that often because of centralization in their corporate policy that ocean in the middle creates a problem for world-wide service.

3. CW America usually doesn't appear to have the infrastructure to handle some of these issues, in my opinion, since I'm willing to bet a support database hasn't really been compiled on these particular topics. You're looking at the closest thing to it! Their IT support budget must be limited not by their intentions by by CW Germany. This is why PlanetZ is what it is and the CW Forum leaves a lot to be desired.

I wonder if CW could tell us how many calls they get regarding these issues and to what percentage they're getting so that when their de-bugging team goes into software revision they can work on the most important things first. Statistically speaking, this could help us all out a lot.

It seems that Subhuman does a lot to clearify these issues since he's trying all kinds of things in his line of work. In reality those tests leave a lot of us tinkering folks, out of alignment with CW's recommendations and ideal systems. But Subhuman doesn't work for CW to the best of my knowledge and it an advanced user trying to get by like the rest of us.

My Advice:

1. Get a system that is recommended for CW. It's not as forgiving as some other systems like RME or MOTU because it does more.

2. Get a dual processor in the fall. The new Cubase will be out. CW will have worked out the bugs on this system and a better software will be available for you to take advantage of dual chips. I had the same problem with a BP6 duall 500. When I went to a single P4 I was happy and my errors stopped. I bought a MOBO that was recommended as working with the CW system with the exception of the dual head MATROX and IDE-RAID which they, Sub, told me not to buy but was able to work around it anyway. At my own risk.

Keep the faith dude.