Future-proof your old Pulsar II PCI Cards

[stonberg]

Modern LGA motherboards with PCI slots are becoming as rare as the earth enrichment material which appears from a rocking horse's behind.

My current PC is of the LGA 1155 variety, which I think is around the sandy bridge time. Since then Intel have moved on to socket 1150 and now socket 1151.

My 2x PCI Pulsar II cards are now 16 years old but still going strong and still as good as if not better than anything else available.

Eventually I want to upgrade my PC, but keep my cards. What to do?

Introducing...

https://www.scan.co.uk/products/startec ... ansion-bay

PCIe addressing/discovery is identical to PCI, so as far as the driver is concerned, this enclosure will be completely transparent and it will be able to talk to the Pulsar II cards as if they were on a PCI bus. It's not available just yet, but is up for pre-order (not that I'd recommend doing such a thing).

[yayajohn]

can it fit a pro card?

[stonberg]

Would that be the 15-chip card? The...Sonic Rocket Booster was it?

Ah, hang on, Startech do this one:

https://www.startech.com/Cards-Adapters ... PEX2PCIE4L

Only 2x PCI slots instead of 4x, but you do get 2x PCIe slots also. It's much longer than the 4x PCI slot one.

Q: did your Pro cards fit into a normal PC ATX form factor case? If so, I would think that they would fit in the 4x PCI slot enclosure. What length are the cards?

[garyb]

ok for one card. will reduce PCI performance.

[stonberg]

ok for one card. will reduce PCI performance.

I really don't see how that would happen.

[fidox]

http://forums.scopeusers.com/viewtopic.php?f=31&t=34281

it's somehow similar post here :

you have similar setup like me, sandy bridge times.
my main audio computer is now 4 years old, so i still have some time, intel i7 2600k.
my second computer is now 6 years old (775 slot), still working nice, just if using newer vst plugins my quad core q9400 coughing lots
i have both computer connected together, so i'm thinking lately to purchase new configuration, slowly build, maybe every month one or two purchases (not much money just for music producing)
the best choice would be replacing quad core q9400 and keep i7 computer.
So, i'll still have new configuration for some years.
Why ?
I just want to keep all my creamware cards and not thinking to sell any of them.
If i want to replace or add bigger card, then i'll need more money for card + new computer configuration (in about 1 or 2 years),
but there are less and less mobos with 3xpci slots.
Well, just thinking loud, but i like to share my thoughts and passion to scope cards

[garyb]

ok for one card. will reduce PCI performance.

I really don't see how that would happen.

actual user experience doesn't lie.

even the Magma boxes suffer from reduced PCI performance. the problem isn't just bandwidth, there is also TIMING, something that is very important to a realtime system(and music ). the Magmas certainly outperform the boxes you mentioned, but they cost almost as much as an XITE-1(they can be found used these days).

the biggest problem with the Startech boxes have been the power supplies. that's why i say one card. Startech did make a kit that used a standard ATX case and power supply. this one would be my choice. in any case, the card will be crippled in that environment. it may work more than well enough for you, but i think keeping it in an older machine makes more sense when you get to that point in the future. right now, there are working machines that are plenty fast.

if having a Scope card and a bleeding edge machine is important, you can probably afford an XITE anyway.

[astroman]

there are tons of old Intel boards out there... P3, PIV
so much that you'll never ever run out of supply, practically garbage to grab for free
the faint of heart may keep 5-10 on the shelf just in case

add a 16 channel Adat board to your latest and greatest supa-dupa-box and bus the stuff via lightpipe
problem solved

it will work just as the Pro Tools TDM system that I recently setup in a 15 year old G4 Mac
the sh*t was irresistably cheapo - 200€ for 2 cards (12 DSPs) and an 8-channel io-interface
I don't plan to record with PT - but it makes a nice FX-box... attached to an iPad
sidenote: I've been shocked about how crappy the DSP stuff was implemented in PT

cheers, Tom

[stonberg]

can it fit a pro card?

From StarTech's website:

- Supports 32-bit PCI cards less than 6.7” in length; 64-bit PCI 12.28” cards (full-length) are not supported

That's for the 4x PCI, shorter enclosure, not for the full length card one.

[stonberg]

(Sorry for delay in responding, I've been on 'holiday')

ok for one card. will reduce PCI performance.

I really don't see how that would happen.

even the Magma boxes suffer from reduced PCI performance. the problem isn't just bandwidth, there is also TIMING, something that is very important to a realtime system(and music ).

What timing are you referring to? No computer system is realtime. Guaranteed delivery within a specified (or known) time frame is as good as it gets. Audio data to/from the card (i.e. pretty much all the data that's sent up and down the PCI bus) will be buffered at both ends and have a known round-trip latency. I doubt the published 'ULLI' bothers to take into account the time it takes to send the data up and down the PCI bus as that will be orders of magnitude smaller than the latency introduced by the data buffering.

What performance? Even if you were passing stereo 100 channels of 24 bit 44.1KHz audio data up and down the PCI bus you'd get nowhere near saturating its bandwidth.

There is absolutely no bandwidth restriction whatsoever on the PCIe side of things. The PCI side will run at PCI speeds. The system will introduce a small amount of latency doing the PCIe<->PCI (serial<->parallel) protocol conversion, but that'll be done in hardware (maybe with a small amount of bare metal firmware code) and so will be of the order of a few microseconds.

> the biggest problem with the Startech boxes have been the power supplies. that's why i say one card.

That might actually be a problem. Power consumption of 1x Pulsar II card is 12W according to the manual's appendix. So the unit will not be able to support more than 2x cards; 3 will be right at the published limit of 36W.

> in any case, the card will be crippled in that environment.

How will it be 'crippled'?

> it may work more than well enough for you

So it will be crippled but work better than ever?

That reminds me of my body as I get older...

> if having a Scope card and a bleeding edge machine is important, you can probably afford an XITE anyway.

Er...no. I do not need the I/O and I don't have the best part of four grand lying around to spend on an interface

I like my Pulsar cards and want to hang on to them. Given that I've already paid about £215 for what was effectively a 32-bit to 64-bit driver update, spending a similar amount to be able to continue using my cards is not unreasonable.

I'm quite happy to be a guinea pig for the enclosure, but that's going to be at least several months away. My PC is still good enough for the time being.

[astroman]

ok... it's your cash...
I've had a Sandy Bridge system myself, which completely failed with Scope cards in PCI slots
so I bought one of those PCIe to PCI slot converters and mounted the card with removed backplate
(the case was significantly higher than a regular one, so it fit inside, though untightened to anything)

there was no magically increased performance whatsoever - it just recovered more quickly from errors
all those bus specs are technical figures
a PCI card is a PCI card - with a certain way IO is programmed
for the same reason PCIe can achieve higher rates, but it cannot turn PCI stuff into something different

I'm not the only one whose tests in that domain failed...

cheers, Tom

[garyb]

stonberg, you are full of incorrect suppositions...

Scope cards are not Windows soundcards. they are external audio hardware that uses the computer for a GUI and power. they require their own operating system. Scope card ARE realtime devices, unlike the computer.

what you paid for was NOT a driver update. the driver part is the EASY part. SonicCore would have given the driver away for free, if that's all the update was. what you paid for was the new DSP operating system/interface/GUI. while it works inside Windows, it is not a part of Windows.

no, external boxes do not work correctly. it may work well enough for you, however. the only real solution at this time, is to use a motherboard with PCI slots that works correctly, if you want the thing to work right. the Startech box works ok for one card, but not two. it would take a Magma box to get the minimal performance needed for multiple cards to say, "well, it does work". Mama boxes cost almost as much as upgrading to an XITE, if you bought one new.

don't believe me? that's fine, try it and see. if it works well enough for you, i'm very happy.

[astroman]

There is absolutely no bandwidth restriction whatsoever on the PCIe side of things. The PCI side will run at PCI speeds. The system will introduce a small amount of latency doing the PCIe<->PCI (serial<->parallel) protocol conversion, but that'll be done in hardware (maybe with a small amount of bare metal firmware code) and so will be of the order of a few microseconds...

in a more general context than mentioned above:
the point is not about bandwidth alone but about un-interupted communication if (near) realtime audio is the topic
max datarates are based on burst transfers which may be interupted arbitrarily (say a copy operation to disk)
there's no reserved path (at a constant rate) for audio that superceeds OS actions
since a desktop OS performs a lot of crap that doesn't matter in DAW context, effectively just a fraction of cycles are available
it would be smart to be able to set priority on the interface level, but no such thing exists

cheers, Tom

[babaorum]

I think I'll prefer to rack (3U) my actual PC with the PCI cards (I know it works huge well)
before to buy a new one instead of take an extension box.