during porting RTEMS to HiFive1 I found out that it has very small RAM and only one accessible UART. It would be great to have more RAM (at least 64 KiB), more accessible UARTs, may be I2C and CAN buses. I think it would be great for demonstration of Core capabilities and for educational purposes. At the same, I think that the current board is too expensive for most developers. So, is it possible that there will be a new board with greater capabilities soon?
The HiFive1 is not the last RISCV board that will ever be made, it’s the first. Later boards will no doubt have more RAM, speed, peripherals.
As for price, things such as Arduino Uno or Raspberry Pi use existing SoCs that are near obsolete and have already shipped in possibly billions of devices and are available at very low prices. The HiFive1 uses a brand new SoC that has never shipped in anything else. Big difference.
I think it’s very good value for someone with a true interest in getting their feet wet with RISCV hardware.
It’s certainly a little pricey to be an Arduino replacement. However, compared to the cost of an FPGA development kit, which is the only other way I’m going to get my hands on RISC-V on actual hardware, it’s a bargain.
I hope so. But rather than adding more peripherals to the E300 core I hope they go up the foodchain and get a U500 into silicon, also at a price point competitive with FPGA dev kits, so the community can start developing outside the small microcontroller scope.
Right. By producing a custom piece of silicon at a very low price - low enough that a small startup could produce a run of chips - in a short timeframe it’s a powerful demonstration to potential SiFive customers that the economics and logistics of their business model are viable.
lowRISC sadly don’t seem to be making very fast progress.
I’m hoping that @yunsup’s presentation at Hot Chips very soon will be about the much anticipated multi core 1.5 GHz 64 bit linux capable SoC and/or a board with it on.
I’d like to second the request for more ram! The ram is currently the biggest restriction on the device. My desktop computers from the 90’s were far less powerful than this little microcontroller, but they had megabytes of ram making them far more useful than a single function device.
A way to run code directly from external SRAM would be amazing!
The vast majority of Arduinos sold (Uno, Leonardo, Pro, Pro Mini, Nano) have one or two KB of SRAM and 16 or 32 KB of flash for the program. Even the Mega (used in countless 3D printer designs) has only 8 KB of SRAM, half that of the HiFive1.
The x86-based Arduino 101 has 24 KB SRAM, and the ARM based Arduino Zero has 32 KB.
16 KB is fine in this class of board.
I of course hope for a very different class of board soon, with GHz+ CPU and a gig of two of RAM.
To what will you attach this SRAM? You’ll need an entirely new SoC with an external bus, not just a new board. Or else use some of the I/O pins to make a serial or narrow parallel bus. It won’t be fast.
The justification is that huge numbers of applications of microcontrollers need little more than a few dozen bytes of internal state.
If you have different needs then there are other classes of SoC, such as the obsolete smartphone SoCs used in the Ordoid, Raspberry Pi and many others like them.