RISC-V Design ยท All levels
RVV Vector Basics
Extensions: Vector & Crypto: RVV decouples architectural vector length from implementation width through vtype and vl, so software scales across cores with different VLEN values. Hardware executes vector instructions over available lanes with tail and mask semantics preserving deterministic behavior when loop trip counts are not multiples of lane width. Correct implementation requires precise exception behavior, clear scalar-vector state transitions, and stable save-restore of vector CSRs across interrupts and context switches.
What this topic teaches
RVV Vector Basics trains mechanism-first reasoning for RISC-V design closure. RVV decouples architectural vector length from implementation width through vtype and vl, so software scales across cores with different VLEN values. Hardware executes vector instructions over available lanes with tail and mask semantics preserving deterministic behavior when loop trip counts are not multiples of lane width. Correct implementation requires precise exception behavior, clear scalar-vector state transitions, and stable save-restore of vector CSRs across interrupts and context switches.
Senior-engineer framing question
When Sustained vector utilization across mixed-width kernels while keeping scalar fallback correctness intact. moves, can you isolate first failing mechanism, request decisive evidence, assign owner, and decide release-safe action?
RISC-V PIPELINE DIAGRAM - RVV Vector Basics
PC -> IF -> ID -> EX -> MEM -> WB
| | | | |
i-cache decode ALU/BR LSU regfile write
\ |
+-> branch resolve + redirect
Hot paths:
- branch + load-use dependencies in ID/EX
- memory latency stretching MEM stage
- writeback arbitration for integer/vector units
Focus: map symptom to first failing stageArchitecture visuals
Draw before you tune. Use these visuals in design reviews, interview loops, and post-silicon triage.
Decode and control map
DECODE CONTROL MAP - RVV Vector Basics
opcode/funct3/funct7 controls asserted
----------------------- ---------------------------------------
LUI / AUIPC rd_write, imm_select(U), alu_add_pc
JAL / JALR rd_write, pc_redirect, link_write
BRANCH cmp_enable, branch_type, pc_redirect
LOAD mem_read, rd_write, wb_sel(memory)
STORE mem_write, store_size, addr_calc
OP-IMM alu_enable, imm_select(I), rd_write
OP alu_enable, src2_reg, rd_write
SYSTEM / CSR csr_readwrite, trap_check, privilege_gate
VECTOR (V extension) vdecode, lane_mask, vtype_updatePrivilege stack
PRIVILEGE MODE STACK - RVV Vector Basics
+------------------------------+
| Machine mode (M) |
| firmware, PMP, trap root |
+---------------+--------------+
|
delegated traps
v
+------------------------------+
| Supervisor mode (S) |
| kernel, page tables, drivers |
+---------------+--------------+
|
ecall / syscall
v
+------------------------------+
| User mode (U) |
| applications, libraries |
+------------------------------+
Key rule: each upward transition records cause + PC in trap CSRs.Translation path
MMU PAGE WALK DIAGRAM - RVV Vector Basics
virtual address
|
+--> TLB lookup hit? ---- yes ---> physical address -> cache/memory
| |
| no
v
satp root PPN + VPN indices
|
+--> level-2 PTE fetch (valid?)
| |
| +-- no -> page fault trap
v
level-1 PTE fetch -> level-0 PTE fetch
|
+--> permissions check (R/W/X, U/S, A/D)
|
+-- fail -> access fault trap
+-- pass -> install TLB entry -> continueVector lane lens
VECTOR LANE VIEW - RVV Vector Basics
VLEN register file
|
+--> lane0: ALU/MUL/permute
+--> lane1: ALU/MUL/permute
+--> lane2: ALU/MUL/permute
+--> lane3: ALU/MUL/permute
...
mask register -> per-lane predicate enable
load/store unit -> strided/segmented access queue
Throughput model:
effective ops/cycle = active_lanes * issue_rate * mask_density
Focus: balance lane utilization and memory feedOwnership layers
RISC-V OWNERSHIP LAYERS - RVV Vector Basics
layer owner closure artifact
-------------------- ---------------------------- -----------------------------
ISA compliance architecture/spec team unpriv + priv test evidence
decode/control front-end RTL owner decode matrix + assertions
pipeline timing microarchitecture owner hazard/perf regression trends
memory + MMU LSU/MMU owner TLB/pagewalk trace checks
privilege/CSR path firmware + kernel interface trap/interrupt conformance
vector subsystem vector RTL + compiler owner lane-utilization profilesEvidence required
Primary metric: Sustained vector utilization across mixed-width kernels while keeping scalar fallback correctness intact..
Primary artifact: Vector bring-up checklist covering vtype/vl programming, tail/mask policy, context-switch state, and scalar fallback tests..
Owners to include: ISA and compiler lead, vector microarchitecture owner, OS and runtime owner, verification lead.
One reproducible workload and one stable comparator run.
One run with locked environment metadata for causal confidence.
Root-cause tree
ROOT CAUSE TREE - RVV Vector Basics
Sustained vector utilization across mixed-width kernels while keeping scalar fallback correctness intact. regressed
|
reproducible on fixed seed?
/ \
no yes
| |
env/tool drift first failing domain?
/ | \
decode execute memory/MMU
| | |
control map bypass/FU TLB/walk/perm
|
privilege/CSR side effects checked?
Stop at first confirmed mechanism, then assign explicit owner + fix proof.Movement trend
BEFORE / AFTER TREND - RVV Vector Basics
Sustained vector utilization across mixed-width kernels while keeping scalar fallback correctness intact.
^
| o target band
| o after fix + reruns
| o
| o baseline (failing)
+--------------------------------------------------> iteration
capture issue isolate mechanism close + monitor
Use this view to confirm the gain is causal and stable across seeds.Key takeaways
Classify mechanism before proposing fixes.
Tie every claim to one proving artifact.
Close with owner accountability and rollback criteria.
Common pitfalls
Averaging away tail behavior and mode-specific failures.
Blending results from mismatched build/runtime metadata.
Declaring closure before cross-workload validation.