Member Since: February 4, 2010

Country: United States

  • Hmmm. Describing the SAMD21 as being the newest version of the Atmega328 seems a tad misleading. About the only thing they have in common are that they’re made by Atmel. They have a completely different and incompatible CPU core, a different set of peripherals that’s not going to be compatible with existing libraries, and you can’t use the SAMD21 with 5V shields and accessories designed for the Atmega328 without frying it. Arguably it’s less suitable as an upgrade than some of the non-Atmel, non-Arduino boards which are not just compatible with the Arduino IDE, but also have 5V tolerant I/O and are supported by a bunch of third-party libraries.

  • Citation: (I think there are probably other FTDI chips beyond the one named that have this issue, too.)

  • It’s kind of scary actually. Supposedly, some other models of genuine FTDI chip don’t ignore the EEPROM write commands used in the bricking code, so if their driver ever misidentifies one of those as an FT232RL for whatever reason and runs the malicious code it’ll quite happily brick those in exactly the same way.

  • The clock speed’s actually misleading, though. Although it runs at a higher clock speed than most Cortex-M3 and M4 boards, many common arithmetic operations take far more clock cycles than on Cortex-M3/M4. For example, on paper the Galileo is actually slower at integer multiplication than the Arduino Due - it’s one clock cycle at 84 MHz on the Due, but between 5 and 12 cycles at 400 MHz on the Galileo depending on the values being multiplied. Bit shifts, rotates, and integer division all take several times as many clock cycles on Galileo too. The Due doesn’t have hardware floating point, but the Cortex-M4F boards that do (like the STM32F4DISCOVERY) apparently beat it even more comprehensively and spectacularly at floating point operations. I expect Cortex-A series SoC are even more impressive.

    This is essentially a slightly improved 486. Intel are using 1989-era CISC technology to try and compete with 2010-era RISC, and it’s working about as well as you’d expect

  • From what I recall, the Intel Galileo is actually closer to a hybrid between a 486 and a first-generation Pentium processor than it is to the Atom processors you’re comparing it to. In particular, it has no support for any of the newer instruction sets like MMX, SSE, or SSE2 - a lot of modern software expects these, especially the kind of x86-only or tuned-for-x86 code that might otherwise give this an advantage over ARM platforms.

  • Ah, right - are you using the resistive touch panel controller for the other 4 ADC channels then? Only just noticed that the A10 user guide mentioned it could be configured as standalone ADC.

  • Anyone thinking of buying should bear in mind that, unless I’m mistaken, the ADCs are a lot slower and lower-resolution than the ones on the Arduino - 6 bits and 250 Hz sampling rate max (no, that’s not a typo). Probably OK for reading a potentiometer or maybe monitoring battery levels if you don’t care much about resolution but not much use for faster or higher-precision applications.

  • Livesuit and PhoenixCard are different programs for loading an operating system image into the onboard Flash. They both have to be run on a normal PC, and if I’m remembering correctly Livesuit requires you to connect the board over USB whereas PhoenixCard writes an SD card you can use to load the OS on. (I’ve got a Cubieboard which uses the same hardware and software.)

  • It appears to be closed source and available either as a free-as-in-beer cloud web application or a paid commercial desktop application with a per-seat licensing fee, as far as I can tell.

  • Interesting, guessing you’ve discontinued making the XMOS XS1-L1-64 Development Board then? (There aren’t that many modern microcontrollers that need a reset controller like the NCP303, and that’s apparently the one that XMOS use in their reference designs.)

No public wish lists :(