\frame{
\begin{center}
- \huge{Simple-V RISC-V Extension for Vectors and SIMD}\\
+ \huge{Simple-V RISC-V Parallelism Abstraction Extension}\\
\vspace{32pt}
\Large{Flexible Vectorisation}\\
\Large{(aka not so Simple-V?)}\\
- \Large{(aka How to Parallelise the RISC-V ISA)}\\
+ \Large{(aka A Parallelism API for the RISC-V ISA)}\\
\vspace{24pt}
\Large{[proposed for] Chennai 9th RISC-V Workshop}\\
\vspace{16pt}
https://sigarch.org/simd-instructions-considered-harmful
\item Setup and corner-cases alone are extremely complex.\\
Hardware is easy, but software is hell.
- \item O($N^{6}$) ISA opcode proliferation!\\
+ \item O($N^{6}$) ISA opcode proliferation (1000s of instructions)\\
opcode, elwidth, veclen, src1-src2-dest hi/lo
\end{itemize}
}
\frame{\frametitle{Quick refresher on RVV}
\begin{itemize}
- \item Extremely powerful (extensible to 256 registers)\vspace{10pt}
- \item Supports polymorphism, several datatypes (inc. FP16)\vspace{10pt}
- \item Requires a separate Register File (32 w/ext to 256)\vspace{10pt}
- \item Implemented as a separate pipeline (no impact on scalar)\vspace{10pt}
- \end{itemize}
- However...\vspace{10pt}
+ \item Effectively a variant of SIMD / SIMT (arbitrary length)\vspace{4pt}
+ \item Fascinatingly, despite being a SIMD-variant, RVV only has
+ O(N) opcode proliferation! (extremely well designed)
+ \item Extremely powerful (extensible to 256 registers)\vspace{4pt}
+ \item Supports polymorphism, several datatypes (inc. FP16)\vspace{4pt}
+ \item Requires a separate Register File (32 w/ext to 256)\vspace{4pt}
+ \item Implemented as a separate pipeline (no impact on scalar)
+ \end{itemize}
+ However...
\begin{itemize}
- \item 98 percent opcode duplication with rest of RV (CLIP)
+ \item 98 percent opcode duplication with rest of RV
\item Extending RVV requires customisation not just of h/w:\\
gcc, binutils also need customisation (and maintenance)
\end{itemize}
adding a level of indirection,
SV expresses how existing instructions should act
on [contiguous] blocks of registers, in parallel, WITHOUT
- needing new any actual extra arithmetic opcodes.
+ needing any new extra arithmetic opcodes.
\item What?
Simple-V is an "API" that implicitly extends
existing (scalar) instructions with explicit parallelisation\\
\frame{\frametitle{What's the value of SV? Why adopt it even in non-V?}
\begin{itemize}
- \item memcpy becomes much smaller (higher bang-per-buck)
+ \item memcpy has a much higher bang-per-buck ratio
\item context-switch (LOAD/STORE multiple): 1-2 instructions
\item Compressed instrs further reduces I-cache (etc.)
- \item Greatly-reduced I-cache load (and less reads)
- \item Amazingly, SIMD becomes (more) tolerable (no corner-cases)
+ \item Reduced I-cache load (and less I-reads)
+ \item Amazingly, SIMD becomes tolerable (no corner-cases)
\item Modularity/Abstraction in both the h/w and the toolchain.
\item "Reach" of registers accessible by Compressed is enhanced
\item Future: double the standard INT/FP register file sizes.
\frame{\frametitle{How does Simple-V relate to RVV? What's different?}
\begin{itemize}
- \item RVV very heavy-duty (excellent for supercomputing)\vspace{8pt}
- \item Simple-V abstracts parallelism (based on best of RVV)\vspace{8pt}
- \item Graded levels: hardware, hybrid or traps (fit impl. need)\vspace{8pt}
- \item Even Compressed become vectorised (RVV can't)\vspace{8pt}
- \item No polymorphism in SV (too complex)\vspace{8pt}
+ \item RVV very heavy-duty (excellent for supercomputing)\vspace{4pt}
+ \item Simple-V abstracts parallelism (based on best of RVV)\vspace{4pt}
+ \item Graded levels: hardware, hybrid or traps (fit impl. need)\vspace{4pt}
+ \item Even Compressed become vectorised (RVV can't)\vspace{4pt}
+ \item No polymorphism in SV (too complex)\vspace{4pt}
\end{itemize}
What Simple-V is not:\vspace{4pt}
\begin{itemize}
- \item A full supercomputer-level Vector Proposal
+ \item A full supercomputer-level Vector Proposal\\
+ (it's not actually a Vector Proposal at all!)
\item A replacement for RVV (SV is designed to be over-ridden\\
by - or augmented to become - RVV)
\end{itemize}
registers are reinterpreted through a level of indirection
\item Primarily at the Instruction issue phase (except SIMD)\\
Note: it's ok to pass predication through to ALU (like SIMD)
- \item Standard (and future, and custom) opcodes now parallel\vspace{10pt}
+ \item Standard and future and custom opcodes now parallel\\
+ (crucially: with NO extra instructions needing to be added)
\end{itemize}
- Note: EVERYTHING is parallelised:
+ Note: EVERY scalar op now paralleliseable
\begin{itemize}
\item All LOAD/STORE (inc. Compressed, Int/FP versions)
\item All ALU ops (Int, FP, SIMD, DSP, everything)
- \item All branches become predication targets (C.FNE added?)
+ \item All branches become predication targets (note: no FNE)
\item C.MV of particular interest (s/v, v/v, v/s)
\item FCVT, FMV, FSGNJ etc. very similar to C.MV
\end{itemize}
\item "2nd FP\&INT register bank" possibility, reserved for future\\
(would allow standard regfiles to remain unmodified)
\item Element width concept remain same as RVV\\
- (CSRs give new size to elements in registers)
+ (CSRs give new size: overrides opcode-defined meaning)
\item CSRs are key-value tables (overlaps allowed: v. important)
\end{itemize}
Key differences from RVV:
\begin{itemize}
\item Predication in INT reg as a BIT field (max VL=XLEN)
\item Minimum VL must be Num Regs - 1 (all regs single LD/ST)
- \item SV may condense sparse Vecs: RVV lets ALU do predication
- \item Choice to Zero or skip non-predicated elements
+ \item SV may condense sparse Vecs: RVV cannot (SIMD-like):\\
+ SV gives choice to Zero or skip non-predicated elements\\
+ (no such choice in RVV: zeroing-only)
\end{itemize}
}
for (i = 0; i < VL; i++)
if (ireg[predr] & 1<<i) # predication uses intregs
ireg[rd+id] <= ireg[rs1+irs1] + ireg[rs2+irs2];
- if (reg\_is\_vectorised[rd]) \{ id += 1; \}
- if (reg\_is\_vectorised[rs1]) \{ irs1 += 1; \}
- if (reg\_is\_vectorised[rs2]) \{ irs2 += 1; \}
+ if (reg\_is\_vectorised[rd] ) \{ id += 1; \}
+ if (reg\_is\_vectorised[rs1]) \{ irs1 += 1; \}
+ if (reg\_is\_vectorised[rs2]) \{ irs2 += 1; \}
\end{semiverbatim}
\begin{itemize}
if (unit-strided) stride = elsize;
else stride = areg[as2]; // constant-strided
for (int i = 0; i < VL; ++i)
- if (preg\_enabled[rd] && ([!]preg[rd] & 1<<i))
+ if ([!]preg[rd] & 1<<i)
for (int j = 0; j < seglen+1; j++)
if (reg\_is\_vectorised[rs2]) offs = vreg[rs2+i]
else offs = i*(seglen+1)*stride;
\end{frame}
-\frame{\frametitle{Register key-value CSR store}
+\frame{\frametitle{Register key-value CSR store (lookup table / CAM)}
\begin{itemize}
\item key is int regfile number or FP regfile number (1 bit)
\frametitle{Register key-value CSR table decoding pseudocode}
\begin{semiverbatim}
-struct vectorised fp\_vec[32], int\_vec[32]; // 64 in future
-
+struct vectorised fp\_vec[32], int\_vec[32];
for (i = 0; i < 16; i++) // 16 CSRs?
tb = int\_vec if CSRvec[i].type == 0 else fp\_vec
idx = CSRvec[i].regkey // INT/FP src/dst reg in opcode
tb[idx].elwidth = CSRvec[i].elwidth
tb[idx].regidx = CSRvec[i].regidx // indirection
+ tb[idx].regidx += CSRvec[i].bank << 5 // 0 (1=rsvd)
tb[idx].isvector = CSRvec[i].isvector
tb[idx].packed = CSRvec[i].packed // SIMD or not
- tb[idx].bank = CSRvec[i].bank // 0 (1=rsvd)
+ tb[idx].enabled = true
\end{semiverbatim}
\begin{itemize}
\item All 32 int (and 32 FP) entries zero'd before setup
- \item Might be a bit complex to set up in hardware (TBD)
+ \item Might be a bit complex to set up in hardware (keep as CAM?)
\end{itemize}
\end{frame}
\frame{\frametitle{Predication key-value CSR store}
\begin{itemize}
- \item key is int regfile number or FP regfile number (1 bit)\vspace{6pt}
- \item register to be predicated if referred to (5 bits, key)\vspace{6pt}
- \item INT reg with actual predication mask (5 bits, value)\vspace{6pt}
- \item predication is inverted Y/N (1 bit)\vspace{6pt}
- \item non-predicated elements are to be zero'd Y/N (1 bit)\vspace{6pt}
+ \item key is int regfile number or FP regfile number (1 bit)
+ \item register to be predicated if referred to (5 bits, key)
+ \item INT reg with actual predication mask (5 bits, value)
+ \item predication is inverted Y/N (1 bit)
+ \item non-predicated elements are to be zero'd Y/N (1 bit)
+ \item register bank: 0/reserved for future ext. (1 bit)
\end{itemize}
Notes:\vspace{10pt}
\begin{itemize}
\item Table should be expanded out for high-speed implementations
- \item Multiple "keys" (and values) theoretically permitted
+ \item Key-value overlaps permitted, but (key+type) must be unique
\item RVV rules about deleting higher-indexed CSRs followed
\end{itemize}
}
\begin{semiverbatim}
struct pred fp\_pred[32], int\_pred[32];
-
for (i = 0; i < 16; i++) // 16 CSRs?
tb = int\_pred if CSRpred[i].type == 0 else fp\_pred
idx = CSRpred[i].regkey
- tb[idx].zero = CSRpred[i].zero
- tb[idx].inv = CSRpred[i].inv
- tb[idx].predidx = CSRpred[i].predidx
+ tb[idx].zero = CSRpred[i].zero // zeroing
+ tb[idx].inv = CSRpred[i].inv // inverted
+ tb[idx].predidx = CSRpred[i].predidx // actual reg
+ tb[idx].predidx += CSRvec[i].bank << 5 // 0 (1=rsvd)
tb[idx].enabled = true
\end{semiverbatim}
\begin{itemize}
- \item All 32 int and 32 FP entries zero'd before setting
- \item Might be a bit complex to set up in hardware (TBD)
+ \item All 32 int and 32 FP entries zero'd before setting\\
+ (predication disabled)
+ \item Might be a bit complex to set up in hardware (keep as CAM?)
\end{itemize}
\end{frame}
\begin{semiverbatim}
def get\_pred\_val(bool is\_fp\_op, int reg):
tb = int\_pred if is\_fp\_op else fp\_pred
- if (!tb[reg].enabled):
- return ~0x0 // all ops enabled
- predidx = tb[reg].predidx // redirection occurs HERE
+ if (!tb[reg].enabled): return ~0x0 // all ops enabled
+ predidx = tb[reg].predidx // redirection occurs HERE
+ predidx += tb[reg].bank << 5 // 0 (1=rsvd)
predicate = intreg[predidx] // actual predicate HERE
if (tb[reg].inv):
predicate = ~predicate // invert ALL bits
% if there are available parallel ALUs to do so.
-\begin{frame}[fragile]
-\frametitle{ADD pseudocode with redirection (and proper predication)}
-
-\begin{semiverbatim}
-function op\_add(rd, rs1, rs2) # add not VADD!
- int i, id=0, irs1=0, irs2=0;
- rd = int\_vec[rd ].isvector ? int\_vec[rd ].regidx : rd;
- rs1 = int\_vec[rs1].isvector ? int\_vec[rs1].regidx : rs1;
- rs2 = int\_vec[rs2].isvector ? int\_vec[rs2].regidx : rs2;
- predval = get\_pred\_val(FALSE, rd);
- for (i = 0; i < VL; i++)
- if (predval \& 1<<i) # predication uses intregs
- ireg[rd+id] <= ireg[rs1+irs1] + ireg[rs2+irs2];
- if (int\_vec[rd ].isvector) \{ id += 1; \}
- if (int\_vec[rs1].isvector) \{ irs1 += 1; \}
- if (int\_vec[rs2].isvector) \{ irs2 += 1; \}
-\end{semiverbatim}
-
- \begin{itemize}
- \item SIMD (elwidth != default) not covered above
- \end{itemize}
-\end{frame}
-
-
\frame{\frametitle{Implementation Options}
\begin{itemize}
\item Absolute minimum: Exceptions: if CSRs indicate "V", trap.\\
- (Requires as absolute minimum that CSRs be in H/W)
+ (Requires as absolute minimum that CSRs be in Hardware)
\item Hardware loop, single-instruction issue\\
(Do / Don't send through predication to ALU)
\item Hardware loop, parallel (multi-instruction) issue\\
Notes:\vspace{4pt}
\begin{itemize}
\item 4 (or more?) options above may be deployed on per-op basis
- \item SIMD always sends predication bits through to ALU
+ \item SIMD always sends predication bits to ALU (if requested)
\item Minimum MVL MUST be sufficient to cover regfile LD/ST
\item Instr. FIFO may repeatedly split off N scalar ops at a time
\end{itemize}
% but MODIFYING the remaining "vectorised" op, subtracting the now
% scalar ops from it.
-\frame{\frametitle{Predicated 8-parallel ADD: 1-wide ALU}
+\frame{\frametitle{Predicated 8-parallel ADD: 1-wide ALU (no zeroing)}
\begin{center}
\includegraphics[height=2.5in]{padd9_alu1.png}\\
{\bf \red Predicated adds are shuffled down: 6 cycles in total}
}
-\frame{\frametitle{Predicated 8-parallel ADD: 4-wide ALU}
+\frame{\frametitle{Predicated 8-parallel ADD: 4-wide ALU (no zeroing)}
\begin{center}
\includegraphics[height=2.5in]{padd9_alu4.png}\\
{\bf \red Predicated adds are shuffled down: 4 in 1st cycle, 2 in 2nd}
}
+\begin{frame}[fragile]
+\frametitle{ADD pseudocode with redirection (and proper predication)}
+
+\begin{semiverbatim}
+function op\_add(rd, rs1, rs2) # add not VADD!
+ int i, id=0, irs1=0, irs2=0;
+ predval = get\_pred\_val(FALSE, rd);
+ rd = int\_vec[rd ].isvector ? int\_vec[rd ].regidx : rd;
+ rs1 = int\_vec[rs1].isvector ? int\_vec[rs1].regidx : rs1;
+ rs2 = int\_vec[rs2].isvector ? int\_vec[rs2].regidx : rs2;
+ for (i = 0; i < VL; i++)
+ if (predval \& 1<<i) # predication uses intregs
+ ireg[rd+id] <= ireg[rs1+irs1] + ireg[rs2+irs2];
+ if (int\_vec[rd ].isvector) \{ id += 1; \}
+ if (int\_vec[rs1].isvector) \{ irs1 += 1; \}
+ if (int\_vec[rs2].isvector) \{ irs2 += 1; \}
+\end{semiverbatim}
+
+ \begin{itemize}
+ \item SIMD (elwidth != default) not covered above
+ \end{itemize}
+\end{frame}
+
+
\frame{\frametitle{How are SIMD Instructions Vectorised?}
\begin{itemize}
\item SIMD ALU(s) primarily unchanged
- \item Predication is added down each SIMD element (if requested,
- otherwise entire block will be predicated as a group)
+ \item Predication added down to each SIMD element (if requested,
+ otherwise entire block will be predicated as a whole)
\item Predication bits sent in groups to the ALU (if requested,
otherwise just one bit for the entire packed block)
\item End of Vector enables (additional) predication:
completely nullifies end-case code (ONLY in multi-bit
predication mode)
\end{itemize}
- Considerations:\vspace{4pt}
+ Considerations:
\begin{itemize}
\item Many SIMD ALUs possible (parallel execution)
\item Implementor free to choose (API remains the same)
% or they can be used to cover several operations on totally different
% vectors / registers.
-\frame{\frametitle{Predicated 9-parallel SIMD ADD}
+\frame{\frametitle{Predicated 9-parallel SIMD ADD (Packed=Y)}
\begin{center}
\includegraphics[height=2.5in]{padd9_simd.png}\\
{\bf \red 4-wide 8-bit SIMD, 4 bits of predicate passed to ALU}
\frame{\frametitle{Why are overlaps allowed in Regfiles?}
\begin{itemize}
- \item Same register(s) can have multiple "interpretations"
+ \item Same target register(s) can have multiple "interpretations"
+ \item CSRs are costly to write to (do it once)
\item Set "real" register (scalar) without needing to set/unset CSRs.
\item xBitManip plus SIMD plus xBitManip = Hi/Lo bitops
\item (32-bit GREV plus 4x8-bit SIMD plus 32-bit GREV:\\
\end{itemize}
Note:
\begin{itemize}
- \item xBitManip reduces O($N^{6}$) SIMD down to O($N^{3}$)
+ \item xBitManip reduces O($N^{6}$) SIMD down to O($N^{3}$) on its own.
\item Hi-Performance: Macro-op fusion (more pipeline stages?)
\end{itemize}
}
\item scalar-to-vector (w/ 1-bit dest-pred): VINSERT
\item vector-to-scalar (w/ [1-bit?] src-pred): VEXTRACT
\item vector-to-vector (w/ no pred): Vector Copy
- \item vector-to-vector (w/ src pred): Vector Gather
- \item vector-to-vector (w/ dest pred): Vector Scatter
+ \item vector-to-vector (w/ src pred): Vector Gather (inc VSLIDE)
+ \item vector-to-vector (w/ dest pred): Vector Scatter (inc. VSLIDE)
\item vector-to-vector (w/ src \& dest pred): Vector Gather/Scatter
\end{itemize}
\vspace{4pt}
Notes:
\begin{itemize}
\item Surprisingly powerful! Zero-predication even more so
- \item Same arrangement for FVCT, FMV, FSGNJ etc.
+ \item Same arrangement for FCVT, FMV, FSGNJ etc.
\end{itemize}
}
CSRvect1 = \{type: F, key: a3, val: a3, elwidth: dflt\}
CSRvect2 = \{type: F, key: a7, val: a7, elwidth: dflt\}
loop:
- setvl t0, a0, 4 # vl = t0 = min(4, n)
+ setvl t0, a0, 4 # vl = t0 = min(min(mvl, 4, n))
ld a3, a1 # load 4 registers a3-6 from x
slli t1, t0, 3 # t1 = vl * 8 (in bytes)
ld a7, a2 # load 4 registers a7-10 from y
\end{frame}
-\frame{\frametitle{Under consideration}
+\frame{\frametitle{Under consideration (some answers documented)}
\begin{itemize}
\item Should future extra bank be included now?
\item 8/16-bit ops is it worthwhile adding a "start offset"? \\
(a bit like misaligned addressing... for registers)\\
or just use predication to skip start?
+ \item see http://libre-riscv.org/simple\_v\_extension/\#issues
\end{itemize}
}
\item An extra pipeline phase almost certainly essential\\
for fast low-latency implementations\vspace{4pt}
\item With zeroing off, skipping non-predicated elements is hard:\\
- it is however an optimisation (and could be skipped).\vspace{4pt}
+ it is however an optimisation (and need not be done).\vspace{4pt}
\item Setting up the Register/Predication tables (interpreting the\\
CSR key-value stores) might be a bit complex to optimise
(any change to a CSR key-value entry needs to redo the table)