Re:OpenTitan: An open-source silicon root of trust from lowRISC
by monocasa on Tuesday December 17, 2019 @12:18AM
(#21810859)
What's the gate count? It should be a good fit for libresilicon, a 1um open source process.
Re:OpenTitan: An open-source silicon root of trust from lowRISC
by namibj on Monday December 16, 2019 @11:45PM
(#21810677)
The question would be: how expensive is it to pay off the NDA-schenannigans for a mostly-obsolete node at a small, practically defunct fab? It should be a competitive advantage at that point.
Re:OpenTitan: An open-source silicon root of trust from lowRISC
by paulgerhardt on Monday December 16, 2019 @11:47PM
(#21810701)
When I last quoted, about $200k for the Hong Kong guys and 4-5 million for the Minnesota/Florida folks. Hong Kong was a much bigger process node.
Re:OpenTitan: An open-source silicon root of trust from lowRISC
by namibj on Tuesday December 17, 2019 @02:20AM
(#21811416)
4-5 million seems worth to try a kickstarter or so for.
If it succeeds, it'll open all this needed info. Kickstarter pledges convert at some ratio into credits that can be used to pay the fab.
Re:OpenTitan: An open-source silicon root of trust from lowRISC
by ljhsiung on Tuesday December 17, 2019 @12:10AM
(#21810818)
There is Verilator (https://github.com/verilator/verilator).
While it, frankly, doesn't hold a candle to the proprietary tools like Synopsys, it's a starting point and serves as a wonderful way to do research and actually learn about hardware.
Put another way, you wouldn't trust a proprietary cryptographic algorithm, why would you trust a black box chip?
Current chips (including OpenTitan) still need to be fabricated at third party foundries using that foundry's proprietary process design kits ('PDKs') and signing two layers of NDA's. As multiple talks at CHES showcase every year there are a number of shenanigans that can be pulled off between a company like Google releasing verilog code for tapeout and the foundry handing back a 'compiled' chip.
The state of open silicon is somewhat analogous to GNU's efforts to create an open source operating system prior to Linux coming along and implementing an open kernel. We still lack open pdk's though it looks like there have been some recent breakthroughs with two foundries in the US and one in Hong Kong.
Recent advances in open FPGA toolchain environments at least let us simulate open solutions here like OpenTitan even if we can't yet tape them out transparently.