============================================================== Guild: wafer.space Community Channel: Information / general / Small Risc-V cores After: 08/31/2025 23:59 Before: 10/01/2025 00:00 ============================================================== [09/21/2025 19:28] rebelmike [09/21/2025 19:28] rebelmike We've realised that the biggest constraint is really the size of the pads. You're going to want a QSPI interface to some RAM and flash, so that's 7 or preferably 12 IOs. And then 4 or 8 IOs for actually doing something. By the time you've made that pad ring you might as well put at least enough RAM in it for registers. But yes, I believe SERV can be configured to store registers in external RAM, that will be extremely slow though! I think going for RV32E (only 15 registers) makes sense. Also if you support the compressed instructions (which is good to save flash bandwidth if you have no/minimal icache) they mostly can only use registers that are in RV32E anyway. [09/21/2025 19:29] rebelmike I put together a spreadsheet for possible pad setups on a subdivided chip. I don't know if the "pad strip" idea (pads on one side only) works in practice. https://docs.google.com/spreadsheets/d/1qEPO6t-Uz7HYwtpl9eECe0vx1UJSmiynZBl_LVMHF0k/edit?gid=0#gid=0 {Embed} https://docs.google.com/spreadsheets/d/1qEPO6t-Uz7HYwtpl9eECe0vx1UJSmiynZBl_LVMHF0k/edit?gid=0 gf180 subdivision 2025-09_media/AHkbwyLfIqJIM4orbn_-RBNL4V6lwW4Kf6L4faZvDz-1951E [09/21/2025 19:32] tholin I believe smaller pads are theoretically possible if you're willing to reduce output drive strength and reduce the effectiveness of the ESD protection. [09/21/2025 19:35] 246tnt Technically nothing forces you to make a ring ... you could have a "strip"of pads on one side only. [09/21/2025 19:35] 246tnt (I mean from a functional stand point, that might not be directly supported in the tooling) [09/21/2025 19:36] mithro_ There is also this thing called "circuit under pads" [09/21/2025 19:39] 246tnt Is that allowed here ? And compatible with the bonding method ? I know in both ihp and sky, placing anything active under the pad is a not an option. And on sky in early mpw some people had high failure rate when bonding because there was some stuff under the pad that was getting crushed. [09/21/2025 19:39] mithro_ https://gf180mcu-pdk.readthedocs.io/en/latest/physical_verification/design_manual/drm_09_3.html [09/21/2025 19:39] mithro_ > Active circuits are allowed when ball-type wire-bonding process is used and when bumping/flip chip technology is used. No circuit under pad is allowed when wedge-type wire-bonding process is used. The following additional rules apply to designs when active circuits are used under the pad. [09/21/2025 19:51] mithro_ 15 pins == 4 for spi bus, 3 for power, 8 for I/O? [09/21/2025 19:56] mithro_ BTW I expect it'll take us a while to figure out what options make sense [09/21/2025 20:01] rebelmike I was thinking the 15 would be 7 for QSPI, 2 for power, clock, reset, 4 for IO. But actually I think you could get 16 in that width and have an extra ground. [09/21/2025 21:11] mithro_ I tend to think of 10 I/O as a good number as it ends up with a byte and 2 "control/signal lines" [09/21/2025 21:56] rebelmike Yeah, only 4 GPIOs is a bit sad. Going to pins on 2 sides and 3x3 division might be a better option. Or 5x2 with a single strip but the other way. ============================================================== Exported 15 message(s) ==============================================================