(no commit message)
[libreriscv.git] / nlnet_2022_ongoing.mdwn
1 # NL.net proposal
2
3 2022-08-107
4
5 ## Project name
6
7 Libre-SOC Ongoing 2022/3
8
9 ## Website / wiki
10
11 <https://libre-soc.org/nlnet_2022_ongoing>
12
13 Please be short and to the point in your answers; focus primarily on
14 the what and how, not so much on the why. Add longer descriptions as
15 attachments (see below). If English isn't your first language, don't
16 worry - our reviewers don't care about spelling errors, only about
17 great ideas. We apologise for the inconvenience of having to submit in
18 English. On the up side, you can be as technical as you need to be (but
19 you don't have to). Do stay concrete. Use plain text in your reply only,
20 if you need any HTML to make your point please include this as attachment.
21
22 ## Abstract: Can you explain the whole project and its expected outcome(s).
23
24 Libre-SOC aims to create a Supercomputing-class entirely Libre Hybrid
25 CPU-VPU-GPU. In proposal 2022-08-51 we aim to begin the long process
26 of submitting the required Scalable Vector Extension to the OpenPOWER
27 Foundation: this Grant Request focusses more on continuing to
28 *implement* that Scalable Vector Extension.
29
30 With the entire project being 100% FOSSHW and managed strictly as a
31 Libre Project under strict Full Transparency conditions the end result
32 is a High-performance Processor Initiative that EU Citizens can trust.
33
34 # Have you been involved with projects or organisations relevant to this project before? And if so, can you tell us a bit about your contributions?
35
36 As mentioned in 2022-08-51,
37 a lot! a full list is maintained here <https://libre-soc.org/nlnet_proposals/>
38 and includes
39
40 * the world's first FOSSHW IEEE754 Formal Correctness Proofs for fadd, fsub, and fma, with support for FP Formal Proofs added to symbiyosis;
41 * the world's first in-place Discrete Cosine Transform algorithm;
42 * Significant improvements to Europe's only silicon-proven FOSSHW VLSI toolchain (coriolis2, by LIP6 Labs of Sorbonne University)
43 to do an 800,000 transistor fully automated RTL2GDSII
44 tape-out;
45 * development of a 180nm Power ISA 3.0 "Test ASIC", the largest fully FOSSHW
46 ASIC ever taped-out in Europe (and funded by Horizon 2020)
47 * development of an Interoperability "Test API" for Power ISA systems,
48 with thousands of unit tests.
49
50 and much more. The side-benefits alone for EU citizens are enormous.
51
52 # Requested Amount
53
54 EUR 100,000.
55
56 # Explain what the requested budget will be used for?
57
58 (Note having completed 2022-02-012 we meet the conditions for a
59 larger budget request)
60
61 Whilst 2022-08-51 focusses on submitting SVP64 to the OpenPOWER ISA WG,
62 and satisfying Voting Members of its suitability, we need to proceed
63 with implementing SVP64 and underlying infrastructure:
64
65 * Dynamic Partitioned SIMD for nmigen
66 * Completion of IEEE754 FP Formal Correctness Proofs
67 * Completion of an In-Order Single-Issue core implementing SVP64
68 * Addition of the IEEE754 FPU to the Core
69 * Addition of other ALUs and pipelines (bitmanip, video)
70 implementing new Draft instructions from 2022-08-051
71 * Addition of SMP (multi-core) support
72 * Running under Verilator and on FPGAs (big ones) which will
73 need to be investigated, bought, and the Libre-Licensed tools support
74 potentially added or improved
75 * Continued documentation, attendance of Conferences online
76 * Begin investigating Multi-Issue Out-of-Order, continuing
77 the 6600 Scoreboard research from 2019-02-012
78 * Establishment and management of Continuous Integration
79 infrastructure and upgrading the Libre-SOC IT systems
80 (currently a single 4GB VM)
81 * If there is sufficient budget we would like to begin investigating
82 OpenCAPI (we have access to two Bitmain 250 FPGAs thanks to UOregon)
83
84 several more practical details which help very much to ensure that the
85 efforts to date, funded very kindly by NLnet, reach fruition as part
86 of providing EU Citizens with a powerful Libre alternative processor
87 option.
88
89 # Compare your own project with existing or historical efforts.
90
91 As hinted at in 2022-08-051
92 we are basically developing a Cray-style Supercomputer, leveraging
93 the Supercomputing-class Power ISA
94 and extending it. Similar historic ISAs include
95 Cray Y/MP, ETA-10, Cyber CDC 205. More recent is the NEC SX Aurora.
96 They are all proprietary systems: Libre-SOC's efforts are entirely
97 FOSSHW.
98
99 Whilst the European Processor Initiative is focussing exclusively
100 on RISC-V, due to the amount of time it takes to assess an ISA's
101 suitability it has to be said that it is being discovered, very slowly,
102 that RISC-V is not suited to High-Performance Supercomputing
103 workloads. The best explanation online is here:
104 <https://news.ycombinator.com/item?id=24459041>
105
106 Therefore this project is a really important alternative
107 being based on a much more suitable High-performance
108 base that has the backing of
109 IBM for over 25 years, and is now an Open ISA.
110 <https://openpowerfoundation.org/blog/final-draft-of-the-power-isa-eula-released/>
111
112 ## What are significant technical challenges you expect to solve during the project, if any?
113
114 Processor design is HARD. This is dramatically underestimated. We are
115 therefore taking a careful and considered incremental approach, using
116 Software Engineering programming techniques, developing unit tests
117 at every level and ensuring rigorous documentation and Project coordination
118 guidelines are adhered to.
119
120 We also make significant use of automation,
121 compiler technology and abstraction
122 which would never be considered by Hardware-only VLSI Engineers.
123 By taking a step back we simplify the approach to one that is
124 manageable by a much smaller team.
125
126 ## Describe the ecosystem of the project, and how you will engage with relevant actors and promote the outcomes?
127
128 As in 2022-08-051
129 we are already set to submit presentations through multiple Conferences
130 as has been ongoing since 2019 as can be seen at <https://libre-soc.org/conferences> and will continue to submit press releases to
131 OPF <https://openpowerfoundation.org/blog/libre-soc-180nm-power-isa-asic-submitted-to-imec-for-fabrication/>. Our entire development is public
132 so is accessible to all.
133
134 # Extra info to be submitted
135