Shop OBEX P1 Docs P2 Docs Learn Events
Cyclone 10 early info — Parallax Forums

Cyclone 10 early info

jmgjmg Posts: 15,171
in Propeller 2
An email says Altera have first info out, on their Cyclone 10 series.

https://www.altera.com/products/fpga/cyclone-series/cyclone-10.html

Light on info of real MHz and $ numbers, but looks similar to MAX 10 with both BGA and E144 packages.

The software might have a prelim target device support, allowing some speed comparisons with Cyclone V in a P2 or P1V build ?
«134

Comments

  • cgraceycgracey Posts: 14,134
    Wow. I didn't know a Cyclone 10 was even in the works. The Cyclone series is all about function density, so that means another doubling of performance-over-price, compared to the current Cyclone V.
  • jmgjmg Posts: 15,171
    cgracey wrote: »
    Wow. I didn't know a Cyclone 10 was even in the works. The Cyclone series is all about function density, so that means another doubling of performance-over-price, compared to the current Cyclone V.

    You could ask Altera about prelim speed/fit files for the Cyclone 10, and see what P2 /P1V run up at ?


  • Cluso99Cluso99 Posts: 18,069
    I haven't been able to get on to the Altera website for days!

    BTW Think they just skipped from Cyclone V to 10 as a marketing gimmick ;)
    i.e. A Cyclone 10 must be better than a Spartan 6 because you know, 10 > 6.
  • evanhevanh Posts: 15,858
    We'll be up to 88k in no time.
  • cgraceycgracey Posts: 14,134
    It looks like none of these devices are as big as the current Cyclone V -A9 device. They are all somewhat smaller.

    The bright spot is that "2x higher performance for up to half the cost" is claimed. Now, do you think they meant 2x performance, or actually 2x HIGHER performance? I'm hoping for the latter, which is 3x performance. Either way, it could run Prop2 at 160MHz, which is full speed.
  • nutsonnutson Posts: 242
    Strange combination, the GX family looks a new high end design (M20K, 10G tranceivers, floating point, fractional PLL, Mlab etc) while the LP family looks lik new Cyclone IV member, lacking most of the new features of the Cyclone V family, even going back to M9K
  • Cluso99Cluso99 Posts: 18,069
    Hope P2 is available well before these devices become available ;)
  • ozpropdevozpropdev Posts: 2,792
    Not a good sign when you get emails from Intel talking up Cyclone 10 while their website has been down for 4 day now. :lol:
  • jmgjmg Posts: 15,171
    edited 2017-02-15 04:54
    I get this from Altera.com
    Product Line            10CL006 10CL010 10CL016 10CL025 10CL040 10CL055 10CL080 10CL120
KLE                           6      10      16      25      40      55      80     120
Memory Blocks (9K)           30      46      56      66     126     260     305     432
Memory Block (Kb)           270     414     504     594   1,134   2,340   2,745   3,888
18 x 18 multipliers          15      23      56      66     126     156     244     288
Phase-locked loop (PLL)       2       2      4       4        4       4       4       4
Global clock networks        10      10      20      20      20      20      20      20
LVDS channels                65      65     137     124     124     132     178     230
E144 (22x22mm) GPIO.LVDS  88.22   88.22   78.19   76.18      NA      NA      NA      NA
M164 (8x8mm,0.5mm)           NA    71.22   71.22     NA      NA      NA      NA      NA
U256 (14x14mm,0.8mm      176.65   176.65  162.53 150.52      NA      NA      NA      NA
  • Ken GraceyKen Gracey Posts: 7,390
    Let me guess: we need to make another FPGA board in preparation for the Cyclone 10!

    Somebody say it ain't so in order to get me to sleep tonight.

    Ken Gracey
  • cgraceycgracey Posts: 14,134
    No need for a new fpga board.
  • jmgjmg Posts: 15,171
    Ken Gracey wrote: »
    Let me guess: we need to make another FPGA board in preparation for the Cyclone 10!

    Somebody say it ain't so in order to get me to sleep tonight.

    Ken Gracey

    I don't think it is quite that bad.

    Luckily, it looks like the present 80MHz P2 speed, can (just) test USB code.

    As Chip mentions above, many of these Cyclone 10's are smaller than the P2 123 FPGA, and there are many choices already for partial P2 emulation. (ie less than 16 COGS)

    Also, you often find Eval boards come on stream before you can buy FPGAs, and the Speed files come before that.

    P2 123 allows full P2 emulation, so that is not going away.

    When a C-10 series speed file is available (might be now, if you ask Altera nicely), you can see what MHz a C-10 allows.
    Then, if that reports significantly faster MHz, you could try to get a C-10 Eval Board, to test P2 at a higher MHz.

    These smaller C-10's, may have more potential running P1V code.

    Another part Parallax could look seriously at, for P1V, is the Lattice ICE40UP5K-SG48ITR ~ $6.00
    ( has an Eval board iCE40UP5K-B-EVN $65)

    That's a practical (at least relative to BGA alternatives) QFN48 package, 39io with 1024kb of RAM. (4x a P1)

  • Cluso99Cluso99 Posts: 18,069
    Ken Gracey wrote: »
    Let me guess: we need to make another FPGA board in preparation for the Cyclone 10!

    Somebody say it ain't so in order to get me to sleep tonight.

    Ken Gracey
    No, No, No !!!
    If the answer is yes, then stop P2 development now and save any more $.
  • PublisonPublison Posts: 12,366
    edited 2017-02-22 14:18
    Maybe new info:

    https://www.altera.com/products/fpga/cyclone-series/cyclone-10.html
  • kwinnkwinn Posts: 8,697
    OMG, the links are approaching novel size!
  • Cluso99Cluso99 Posts: 18,069
    Still cannot reach Altera - link fails :(
  • ozpropdevozpropdev Posts: 2,792
    Cluso99 wrote: »
    Still cannot reach Altera - link fails :(
    Same here.
    All the above links fail for me.
    Even the local Intel FPGA site fails when you click on the "Learn about Cyclone FPGA's" link.
  • kwinnkwinn Posts: 8,697
    The link Publison posted worked for me.
  • ozpropdevozpropdev Posts: 2,792
    No access outside USA?
  • AribaAriba Posts: 2,690
    ozpropdev wrote: »
    No access outside USA?
    No problem from Europe.
    You can try a free Proxy-server in Europe or USA.
    If I do the same with an Australian Proxy, I also get no access.

    Andy


  • RaymanRayman Posts: 14,576
    edited 2017-02-17 12:07
    This may be a sore point with Chip, but how much can we count on 160 MHz?
    I think we're OK at 80 MHz, but 160 MHz would definitely be better.

    I'm guess that the design tools Chip is using now give a solid indication of upper speed limit, right?

    Anyway, if 160 MHz is for sure, it would be nice to test it out with FPGA...
  • cgraceycgracey Posts: 14,134
    Rayman wrote: »
    This may be a sore point with Chip, but how much can we count on 160 MHz?
    I think we're OK at 80 MHz, but 160 MHz would definitely be better.

    I'm guess that the design tools Chip is using now give a solid indication of upper speed limit, right?

    Anyway, if 160 MHz is for sure, it would be nice to test it out with FPGA...

    The device will synthesize to 160MHz, no problem. And that's worst process, -10% voltage, and 125-degree C temperature.
  • RaymanRayman Posts: 14,576
    Can your test chip tell you if the analog I/o stuff will go to 160 MHz?
  • kwinnkwinn Posts: 8,697
    ozpropdev wrote: »
    No access outside USA?

    No problem from Canada.
  • ozpropdevozpropdev Posts: 2,792
    Ariba wrote: »
    ozpropdev wrote: »
    No access outside USA?
    No problem from Europe.
    You can try a free Proxy-server in Europe or USA.
    If I do the same with an Australian Proxy, I also get no access.

    Andy

    Thanks Andy.
    It didn't get to try the proxy trick as the problem seems to be fixed now.
    All links to Altera work now. :)

  • Cluso99Cluso99 Posts: 18,069
    edited 2017-02-17 23:39
    At last, altera.com and the link above are accessible from Oz - took a while!!!

    Curiously the Cyclone 10 has gone back to 4 input LEs like the Cyclone IV. The V had 6 input LEs and seemed to be less efficient in the P1V than the Cyclone V.

    No price/availability info yet.
  • AleAle Posts: 2,363
    Cluso: There are two versions of the cyclone 10, one the LP seems to have 4 input LUTs the other one (GX) names the ALMs in the brief. But the offer seems to have smaller members than what CV offers.
  • TorTor Posts: 2,010
    The link works, but strip out everything after .html please.. that's just tracking info.
  • PublisonPublison Posts: 12,366
    Tor wrote: »
    The link works, but strip out everything after .html please.. that's just tracking info.

    Thanks. Done!

  • jmgjmg Posts: 15,171
    edited 2017-06-20 20:49
    Bumping with this new email :

    "Software and Silicon Available for Intel® Cyclone ® 10 LP Devices
    Start your next design today with Intel® Cyclone® 10 LP silicon and the latest release of Intel Quartus® Prime design software v17.0. This software provides support for all device and package options so you can get started immediately with the specific device you need.
    Intel Cyclone 10 LP FPGAs have half the power at half the cost. Ideal for high-volume, cost-sensitive functions among a broad spectrum of general logic applications."

    Mouser show prices, LE count is in the part number

    10CL006YE144C8G 60 $6.7400 RAM 270000b
    for comparison, a MAX 10
    10M04SCE144C8G Altera $11.43000 1 RAM 193536b

    10CL025YU256C8G 119 $18.7100 RAM : 594000b
    10CL025YE144C8G 60 $19.460
    for comparison, a MAX 10
    10M25SCE144C8G 60 $31.610 RAM : 691200b


    Cyclone 10 have less IO in the E144 than MAX 10, but do seem to be cheaper per LE.
    Still no signs of MHz value, I guess a build of a P1V on each could tell us packing and MHz ?

Sign In or Register to comment.