Re: Another ? about Speex and CPU
[email protected] Wed, 29 Feb 2012 23:29:35 -0500
| Newsgroups | gmane.comp.audio.compression.speex.devel |
|---|---|
| Message-ID | <[email protected]> |
Yes, it does and thanks for the fast answer! The AVR32 runs at same speed but has a built in floating point unit for float calcs and a DSP instruction Set. But If I use the fixed point version I guess it doesn't change anything then. So another question: - Is there a difference of performance/quality between the floating point and the fixed point? Actually, what's the main difference? Thanks again. Dominic Quoting Jean-Marc Valin <[email protected]>: > I don't know the AVR32 chip. However for an ARM7TDMI running at 50 MHz, > assuming a good compiler and some hand-coded assembly, you would be very > close to the achievable limit (couldn't say on which side). So if your > AVR32 is faster than an ARM7TDMI, it's likely achievable. If it's > slower, then it's likely not achievable. > > Hope that helps, > > Jean-Marc > > On 12-02-29 10:21 PM, [email protected] wrote: >> Hello, >> >> I am interested to use Speex but need to know If I can use it under >> our project's constraint. We need to use a low cost micro controller >> on a very small PCB. This means there's no space for an external DSP >> unit. >> >> Can someone tell me if using an AVR32 (Atmel uc) which has a floating >> point calculation unit and DSP instruction set and running at ~50Mhz >> for a narrow band 8Kbps sampling rate maybe be feasible or out of reach? >> >> AT32UC3C0512C: >> Low Power 32-bit AVR® Microcontroller >> - Compact Single-cycle RISC Instruction Set Including DSP Instruction Set >> - Built-in Floating-Point Processing Unit (FPU) >> - Read-Modify-Write Instructions and Atomic Bit Manipulation >> - Performing 1.49 DMIPS / MHz >> o Up to 91 DMIPS Running at 66 MHz from Flash (1 Wait-State) >> o Up to 49 DMIPS Running at 33 MHz from Flash (0 Wait-State) >> >> Or >> >> An ARM7TDMI core running at ~50Mhz for a narrow band 8Kbps sampling rate? >> >> I need to know this before I get deeper into Speex. >> >> Thanks a lot! >> Dominic Bazinet >> >> _______________________________________________ >> Speex-dev mailing list >> [email protected] >> http://lists.xiph.org/mailman/listinfo/speex-dev >> >> > >