[asterisk-dev] Support for Dialogic DM3 cards

[Miroslav Nachev]

Gerry,

I think that the prices of Intel boards are additional barrier for
using them with open source products. It is a shame that the BOM is
under $100 USD and the price is more than $5,000 USD.
Also, I think that up to few months on the market will be available
boards similiar or better than Intel Communications boards at price in
times less than Intel's. I am sure for that because I have internal
information for boards where are used TI DSPs for decoding and
encoding of G.72xx, Speex, GSM, Echo Cancellation, DTMF Detection and
Generation, Fax Detection/Recv/Send, T.38, H.264, MPEG-4, MP3, etc.
So, after few months Intel will stop selling such cards or will
decrease the prices in times.


-- 
Best regards,
 Miroslav                            mailto:miro at space-comm.com

Monday, March 27, 2006, 7:23:20 PM, you wrote:

GG> While Steve is absolutely correct that the older Dialogic cards were
GG> only half-duplex, the newer "JCT" series of cards - and all of the DM3
GG> series cards - do support full-duplex audio streaming through a single
GG> DSP, so they can fit into the Asterisk architecture quite nicely - once
GG> you tune those buffer sizes, which we are doing in our channel driver.
GG> :-)

GG> More broadly, though, to the bigger question of DSP vs HMP (Host Media
GG> Processing: Intel-speak for audio processing through the host CPU),
GG> obviously there are many situations where HMP is the ticket. Certainly
GG> Steve's HMP routines in Asterisk have shown the power of this technology
GG> and we at Intel sell our own HMP. However, there are some situations
GG> where - for right now, anyway - onboard DSP support, especially when
GG> coupled with an onboard T1/E1 interface for advanced call control, can
GG> be more efficient, especially in higher call volumes. 

GG> I think the fact that Digium themselves seem to be moving towards more
GG> onboard DSP-type processing in their products bears out this fact much
GG> more than anything that I could say, anyway.

GG> Finally, for those of you who met Dwayne (the Intel programmer leading
GG> the chan-dialogic driver effort) or myself at VON, then you already know
GG> that we are driving towards releasing the channel driver with the next
GG> release of Asterisk Business Edition, due RSN. As an earlier poster
GG> noted, for several reasons, our channel driver is available only in
GG> Asterisk Business Edition. As an old-time open source believer, it pains
GG> me personally for this to be the case, but at this point it is a matter
GG> of having an ABE-only channel driver or no channel driver.

GG> Gerry
GG>    

GG> There are 10 kinds of people in the world, those who understand binary
GG> and those who don't.
GG>  
GG> Gerry Gilmore
GG> Field Applications Engineer
GG> Intel Corporation
GG> (http://www.intel.com)
GG>  

GG> -----Original Message-----
GG> From: asterisk-dev-bounces at lists.digium.com
GG> [mailto:asterisk-dev-bounces at lists.digium.com] On Behalf Of Steve
GG> Underwood
GG> Sent: Saturday, March 25, 2006 10:09 AM
GG> To: Asterisk Developers Mailing List
GG> Subject: Re: [asterisk-dev] Support for Dialogic DM3 cards

GG> Wai Wu wrote:

>>---- i could comment on 'whats *loose* around here' but, there is no
GG> reason too.
>>
>>---- either you believe in asterisk and  what it can do for you.  (or
GG> you don't)
>>
>>---- if you are the latter, i'd suggest you leave before others sense
GG> your
>>---- lack of commitment.
>>
>>---- god forbid, you are a dialogic mole <G>.
>>
>>
>>Gee, you guys sound like a cult. LoL. 
>>  
>>
GG> I think that guy sounds more like a cretin.

>>Seriously, I am looking for ways to better integrate Dialogic/Aculab
GG> cards with Asterisk. The way it is done right now is very wasteful on
GG> the hardware (two speech resources per channel) when routing calls from
GG> PSTN to, say, IAX trunks. When I know more about the * source code,
GG> there might opportunity for a solution to this. 
>>  
>>
GG> The problem lies with the hardware, at least for Dialogic. There are
GG> Aculab drivers for *, but I don't know much about the Aculab cards. I do
GG> know quite a lot about the Dialogic cards, and they are very limiting.
GG> Most of the voice resources, and all the older one, are half-duplex. You
GG> can only do VoIP calls with a pairs of these resources. Even then, the
GG> performance sucks. Dialogic use huge amounts of buffering. This is great
GG> for IVR applications, as it makes the software timing very relaxed. For
GG> VoIP it makes the latency awful. There is a good reason why Dialogic
GG> built a completely separate line of cards for VoIP - the IVR + H.100 bus
GG> voice oriented cards really don't cut it.

GG> Regards,
GG> Steve


GG> _______________________________________________
GG> --Bandwidth and Colocation provided by Easynews.com --

GG> asterisk-dev mailing list
GG> To UNSUBSCRIBE or update options visit:
GG>    http://lists.digium.com/mailman/listinfo/asterisk-dev
GG> _______________________________________________
GG> --Bandwidth and Colocation provided by Easynews.com --

GG> asterisk-dev mailing list
GG> To UNSUBSCRIBE or update options visit:
GG>    http://lists.digium.com/mailman/listinfo/asterisk-dev