gideon@taux01.UUCP ( Gideon Intrater) (08/17/89)
I heard that TI has announced recently a new DSP chip (think its called - 320C50). Does anyone know what can this beast do ? Is it any faster than the Motorola DSP96000 ? Can it do Floating-Point or just Fixed-Point arithmetic ? -- Gideon Intrater gideon@nsc.nsc.com National Semiconductor, P.O.Box 3007, Herzlia B 46104, Israel Phone: +972-52-522303, Fax: +972-52-558322
jbuck@epimass.EPI.COM (Joe Buck) (08/18/89)
In article <2380@taux01.UUCP> gideon@nsc.nsc.com ( Gideon Intrater) writes: >I heard that TI has announced recently a new DSP chip (think its called - >320C50). Does anyone know what can this beast do ? Is it any faster than the >Motorola DSP96000 ? Can it do Floating-Point or just Fixed-Point arithmetic ? TMS320C30. It is a 32-bit floating point DSP; it does not resemble the TMS320C2x (20 or 25) chips in any way, and it would have been more honest for TI to make an unrelated name for the chip. It does a floating multiply in parallel with a floating addition (the destinations can be different, not just a multiply-and-accumulate) in a single 60ns cycle, and like most DSP chips, it has zero-overhead loops. Integer multiplies are only 24x24 bits (32 bit result). It's in about the same league as the DSP96000 in terms of performance, though I don't know as much about the 96k as about the C30. By the way, can you get 96000's yet? By the way, Todd Day (todd@ivucsb.sba.ca.us) is now conducting the official vote for the creation of comp.dsp. Send yes votes to yes@ivucsb.sba.ca.us and no votes to no@ivucsb.sba.ca.us. -- -- Joe Buck jbuck@epimass.epi.com, uunet!epimass.epi.com!jbuck
grosen@amadeus.ucsb.edu (Mark D. Grosen) (08/18/89)
There is a new TMS320C50 chip (as well as the TMS320C30), which was announced at the International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in May. This is a continuation of the fixed-point generation. It is basically a souped up TMS320C25, but with some neat new features. From the TI TMS320C25 DSP Preview Bulletin: * 35 and 50 ns single-cycle time * single cycle mult-and-add * zero-overhead loops * zero-overhead context switches (dual register set) * 8K (16 bit) program/data RAM * 544 dual-access data RAM * 2K boot ROM * on chip serial port and timer * software wait state generators * provisions for adding custom peripherals to the internal bus * IEEE JTAG test bus on chip Mark D. Grosen ARPA: grosen@amadeus.ucsb.edu Signal Processing Lab / Communications Research Lab ECE Dept. University of California Santa Barbara, CA 93106
mdeale@cosmos.acs.calpoly.edu (Myron Deale) (08/18/89)
In article <3533@epimass.EPI.COM> jbuck@epimass.EPI.COM (Joe Buck) writes: >In article <2380@taux01.UUCP> gideon@nsc.nsc.com ( Gideon Intrater) writes: >>I heard that TI has announced recently a new DSP chip (think its called - >>320C50). Does anyone know what can this beast do ? Is it any faster than the >>Motorola DSP96000 ? Can it do Floating-Point or just Fixed-Point arithmetic ? > >TMS320C30. It is a 32-bit floating point DSP; it does not resemble There is a 'C50. EE Times, among other publications, has a pre- announcement of the chip: 16-bit integer (ie. fixed-point), 57MHz clock, 35 ns MAC, about 1e6 transistors. Mainly intended for embedded control applications, has some glue-type peripheral enhancements on-board. In addition, the 'C50 has 8K Words program/data RAM on-board, supports a four-wire serial test bus (ala IEEE p1149.1), and "the C50 handles interrupts with no overhead, for faster processing of even complex interrupt routines." Doesn't look all that impressive at first glance, in fact the interrupt stuff reminds me a little of TI's non-conformist and sometimes dingy past. Not as fancy as the C30, however, there appears to be enough good stuff to recommend it. I wonder how well it performs in comparison to the DSP16A (from ATT), the 25 ns version? >in about the same league as the DSP96000 in terms of performance, >though I don't know as much about the 96k as about the C30. By the >way, can you get 96000's yet? The 96000 looks slightly faster than the C30. A more humane register set-up too, as far as programmers are concerned. I would guess they are sampling, with production a ways off. That's a guess, I'll have to call to make sure. [Follup-To: comp.dsp, if/when it's created] > >By the way, Todd Day (todd@ivucsb.sba.ca.us) is now conducting the >official vote for the creation of comp.dsp. Send yes votes to >yes@ivucsb.sba.ca.us and no votes to no@ivucsb.sba.ca.us. >-- >-- Joe Buck jbuck@epimass.epi.com, uunet!epimass.epi.com!jbuck -Myron // mdeale@cosmos.acs.calpoly.edu
garyb@iotek.UUCP (Gary Burrell) (08/18/89)
In article <2380@taux01.UUCP> you write: > >I heard that TI has announced recently a new DSP chip (think its called - >320C50). Does anyone know what can this beast do ? Is it any faster than the >Motorola DSP96000 ? Can it do Floating-Point or just Fixed-Point arithmetic ? > >-- >Gideon Intrater gideon@nsc.nsc.com >National Semiconductor, P.O.Box 3007, Herzlia B 46104, Israel >Phone: +972-52-522303, Fax: +972-52-558322 The TMS320C50 has been anounced, it is a 35 ns fixed point DSP and an upgrade to TI's TMS320C2X family. A quick summary of features is: 1 level stack on 13 of the CPU registers to allow for a zero level context switch. More Memory 8 K x 16 on chip program/data RAM 544 x 16 dual-access data RAM Full Duplex serial port Interval Timer Software Wait State Generator Parallell input/output JTAG IEEE 1149.1 Test bus According to the information I recieved from TI on this chip the asembler-linker will be available 4Q 88, Samples in 1Q 90 and full production later in 1990, The C compiler will be available sometime in 1990. For more information contact TI, however earlier in the summer when I requested the data sheet etc. on this chip they sent me a non-discloser document to sign (I wasn't that interested in the chip :-) ), so I do not have any further information on this chip other than what is in the device preview (Doc # SPRT095). Disclammer: I have no association with TI, the above information was taken from a TI Preview Bullitin. <<<<<<******>>>>>> Gary R. Burrell, Iotek Inc, |*| E-Mail: garyb@iotek.uucp |*| 1127 Barrington St., Suite 100, |*| Fax: (902)420-0674 |*| Halifax, N.S., B3H 2P8, Canada |*| Phone: (902)420-1890 |*| Damm it Jim I'm a Doctor not a Computer Scientist! *************************************