rohitksingh has quit [Remote host closed the connection]
rohitksingh has joined ##openfpga
rohitksingh has quit [Remote host closed the connection]
rohitksingh has joined ##openfpga
rohitksingh has quit [Remote host closed the connection]
rohitksingh has joined ##openfpga
rohitksingh has quit [Remote host closed the connection]
rohitksingh has joined ##openfpga
emeb_mac has joined ##openfpga
<zyp>
are there any ecp5 boards with usable IO available for purchase anywhere apart from the official boards from lattice?
<zyp>
I know of the chinese led driver boards (and are awaiting a couple in the mail), but I understand they are set up for output only, and the IO voltage on the other side of the level translators is not what I want
<zyp>
and then there's the elgato camlink which doesn't seem to have any IO apart from HDMI
<zyp>
everything else seems to be «coming soon» or «assemble it yourself»
<emeb_mac>
zyp: I just recently saw someone using those Colorlight boards but replacing the 5V drivers with bidir FET switch chips which fortunately had the same pinout.
<peepsalot>
is there any overall side-by-side comparison of various open soft cores out there? with info like feature sets, relative resource usage, etc.?
<tnt>
zyp: and the elgato hdmi is not connected to the fpga directly (there is a hdmi receiver cihp) so you can't even use those for anything else than real hdmi.
<tnt>
peepsalot: not that I know of. There is so much variation ... also depending on the target fpga, resource usage can vary greatly.
<peepsalot>
tnt: i'm mostly curious about particularly small implementations. sort of wondering what the smallest feature set needed to run linux for example.
<peepsalot>
i suppose there's probably some disagreements about who holds the "record"
<tnt>
"particularly small" and "run linux" are not quite the same objective :p
<tnt>
I don't know about other architecture but for RISC-V your best bet is probably the VexRiscv implementation, it can be configured to run linux.
<tnt>
What's your target fpga ?
<peepsalot>
cyclonev / de10-nano
<tnt>
And why do you want to run linux in the first place ? Because other than a cool tech demo, it's going to be fairly useless except in very specific cases (like prototyping a HDL design you want to make an ASIC of or stuff like that).
<peepsalot>
idk, not really a requirement, just learning what options are out there, and trying to get an idea of complexity for such a core
<tnt>
mmu-less linux might bring it to a more reasonable overhead. I think I saw some riscv support for that yet but I don't know if anyone has run it on a fpga soft core yet.
<sorear>
you still need 8-16 MB of RAM, which is going to be larger than any reasonable MMU
Asu has joined ##openfpga
<tnt>
sorear: well you need that in both cases.
<tnt>
An hyperram controller is not that big. ~ 500 LCs on an ice40. (and half of that is implementing high-speed serdes manually, so on a ecp5 that has it built in, probably 300 LUT/FFs)