[LAD] Is RME HDSPe AES supported by alsa?

[Joshua D. Boyd]

Niels Mayer wrote:
> On Wed, May 5, 2010 at 3:55 PM, Reuben Martin <reuben.m at gmail.com 
> <mailto:reuben.m at gmail.com>> wrote:
> 
>     IIRC, they put such a huge matrix on it because they designed one
>     chip to slap on all their cards in that family. Saves money to just
>     design and fab one chip instead of a separate chip for each unit.
> 
> 
> RME uses FPGA's, which is why they're so expensive 

<snip>

> It seems like after so many years of production, now that they've 
> stabilzed on a particular architecture, they could have come up with 
> custom VLSI to replace the FPGA and save everybody a bunch of money; and 
> allow their cards to be used by a wider number of musicians whose 
> budgets send them to "prosumer" cards with chips 
> like http://www.via.com.tw/en/products/audio/controllers/envy24/ ; these 
> implement their 20-channel 36-bit wide digital mixer in VLSI, and derive 
> significant economies of scale from the fact that they can sell one of 
> their chips in every PC made.

I wouldn't agree that RME should switch to VLSI.  I also wouldn't say it 
is entirely fair to blame the cost on them using FPGAs.

FPGAs can be quite cheap.  Ignoring the mixer for a moment, the FPGA for 
a 64 channel matrix + numerous input and output channels + high speed 
link to other cards + PCI interface would be about about $15-20 in 
quantities of one.  Expect discounts for 1000 units, and more discounts 
for 10,000 units.

A VLSI would cost millions to make.  They might have to move 100,000 
units using the exact same chip at the exact same card price to break 
even, and if RME was that high volume of a vendor, then their FPGA 
prices would be much lower than the list prices one can find online.