\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 A Parallelism API for the RISC-V ISA)}\\
\vspace{24pt}
\Large{[proposed for] Chennai 9th RISC-V Workshop}\\
- \vspace{24pt}
+ \vspace{16pt}
\large{\today}
\end{center}
}
\frame{\frametitle{Quick refresher on SIMD}
\begin{itemize}
- \item SIMD very easy to implement (and very seductive)\vspace{10pt}
- \item Parallelism is in the ALU\vspace{10pt}
- \item Zero-to-Negligeable impact for rest of core\vspace{10pt}
+ \item SIMD very easy to implement (and very seductive)\vspace{8pt}
+ \item Parallelism is in the ALU\vspace{8pt}
+ \item Zero-to-Negligeable impact for rest of core\vspace{8pt}
\end{itemize}
Where SIMD Goes Wrong:\vspace{10pt}
\begin{itemize}
\item See "SIMD instructions considered harmful"
- https://www.sigarch.org/simd-instructions-considered-harmful
- \item Corner-cases alone are extremely complex.\\
+ 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 and s/w also need customisation (and maintenance)
+ gcc, binutils also need customisation (and maintenance)
\end{itemize}
}
\item Why?
Implementors need flexibility in vectorisation to optimise for
area or performance depending on the scope:
- embedded DSP, Mobile GPU's, Server CPU's and more.\vspace{4pt}\\
+ embedded DSP, Mobile GPU's, Server CPU's and more.\\
Compilers also need flexibility in vectorisation to optimise for cost
of pipeline setup, amount of state to context switch
- and software portability\vspace{4pt}
+ and software portability
\item How?
- By implicitly marking INT/FP regs as "Vectorised",\\
+ By marking INT/FP regs as "Vectorised" and
+ adding a level of indirection,
SV expresses how existing instructions should act
- on [contiguous] blocks of registers, in parallel.\vspace{4pt}
+ on [contiguous] blocks of registers, in parallel, WITHOUT
+ needing any new extra arithmetic opcodes.
\item What?
Simple-V is an "API" that implicitly extends
- existing (scalar) instructions with explicit parallelisation
- (i.e. SV is actually about parallelism NOT vectors per se)
+ existing (scalar) instructions with explicit parallelisation\\
+ i.e. SV is actually about parallelism NOT vectors per se.\\
+ Has a lot in common with VLIW (without the actual VLIW).
\end{itemize}
}
\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)\vspace{10pt}
- \item context-switch (LOAD/STORE multiple): 1-2 instructions\vspace{10pt}
- \item Compressed instrs further reduces I-cache (etc.)\vspace{10pt}
- \item greatly-reduced I-cache load (and less reads)\vspace{10pt}
- \end{itemize}
- Note:\vspace{10pt}
+ \item memcpy becomes much smaller (higher bang-per-buck)
+ \item context-switch (LOAD/STORE multiple): 1-2 instructions
+ \item Compressed instrs further reduces I-cache (etc.)
+ \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.
+ \end{itemize}
+ Note:
\begin{itemize}
\item It's not just about Vectors: it's about instruction effectiveness
\item Anything implementor is not interested in HW-optimising,\\
\frame{\frametitle{How does Simple-V relate to RVV? What's different?}
\begin{itemize}
- \item RVV very heavy-duty (excellent for supercomputing)\vspace{10pt}
- \item Simple-V abstracts parallelism (based on best of RVV)\vspace{10pt}
- \item Graded levels: hardware, hybrid or traps (fit impl. need)\vspace{10pt}
- \item Even Compressed become vectorised (RVV can't)\vspace{10pt}
+ \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{10pt}
+ 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, or just be replaced by - RVV)
+ by - or augmented to become - RVV)
\end{itemize}
}
\frame{\frametitle{How is Parallelism abstracted in Simple-V?}
\begin{itemize}
- \item Register "typing" turns any op into an implicit Vector op\vspace{10pt}
+ \item Register "typing" turns any op into an implicit Vector op:\\
+ 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}
- Notes:\vspace{6pt}
+ Note: EVERY scalar op now paralleliseable
\begin{itemize}
\item All LOAD/STORE (inc. Compressed, Int/FP versions)
- \item All ALU ops (soft / hybrid / full HW, on per-op basis)
- \item All branches become predication targets (C.FNE added)
+ \item All ALU ops (Int, FP, SIMD, DSP, everything)
+ \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}
}
-\frame{\frametitle{Implementation Options}
-
- \begin{itemize}
- \item Absolute minimum: Exceptions (if CSRs indicate "V", trap)
- \item Hardware loop, single-instruction issue\\
- (Do / Don't send through predication to ALU)
- \item Hardware loop, parallel (multi-instruction) issue\\
- (Do / Don't send through predication to ALU)
- \item Hardware loop, full parallel ALU (not recommended)
- \end{itemize}
- Notes:\vspace{6pt}
- \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 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}
-}
-% Instr. FIFO may need its own slide. Basically, the vectorised op
-% gets pushed into the FIFO, where it is then "processed". Processing
-% will remove the first set of ops from its vector numbering (taking
-% predication into account) and shoving them **BACK** into the FIFO,
-% but MODIFYING the remaining "vectorised" op, subtracting the now
-% scalar ops from it.
-
-\frame{\frametitle{Predicated 8-parallel ADD: 1-wide ALU}
- \begin{center}
- \includegraphics[height=2.5in]{padd9_alu1.png}\\
- {\bf \red Predicated adds are shuffled down: 6 cycles in total}
- \end{center}
-}
-
-
-\frame{\frametitle{Predicated 8-parallel ADD: 4-wide ALU}
- \begin{center}
- \includegraphics[height=2.5in]{padd9_alu4.png}\\
- {\bf \red Predicated adds are shuffled down: 4 in 1st cycle, 2 in 2nd}
- \end{center}
-}
-
-
-\frame{\frametitle{Predicated 8-parallel ADD: 3 phase FIFO expansion}
- \begin{center}
- \includegraphics[height=2.5in]{padd9_fifo.png}\\
- {\bf \red First cycle takes first four 1s; second takes the rest}
- \end{center}
-}
-
-
-\frame{\frametitle{How are SIMD Instructions Vectorised?}
-
- \begin{itemize}
- \item SIMD ALU(s) primarily unchanged\vspace{6pt}
- \item Predication is added to each SIMD element\vspace{6pt}
- \item Predication bits sent in groups to the ALU\vspace{6pt}
- \item End of Vector enables (additional) predication\vspace{10pt}
- \end{itemize}
- Considerations:\vspace{4pt}
- \begin{itemize}
- \item Many SIMD ALUs possible (parallel execution)
- \item Implementor free to choose (API remains the same)
- \item Unused ALU units wasted, but s/w DRASTICALLY simpler
- \item Very long SIMD ALUs could waste significant die area
- \end{itemize}
-}
-% With multiple SIMD ALUs at for example 32-bit wide they can be used
-% to either issue 64-bit or 128-bit or 256-bit wide SIMD operations
-% or they can be used to cover several operations on totally different
-% vectors / registers.
-
-\frame{\frametitle{Predicated 9-parallel SIMD ADD}
- \begin{center}
- \includegraphics[height=2.5in]{padd9_simd.png}\\
- {\bf \red 4-wide 8-bit SIMD, 4 bits of predicate passed to ALU}
- \end{center}
-}
-
-
\frame{\frametitle{What's the deal / juice / score?}
\begin{itemize}
- \item Standard Register File(s) overloaded with CSR "vector span"\\
+ \item Standard Register File(s) overloaded with CSR "reg is vector"\\
(see pseudocode slides for examples)
- \item Element width and type concepts remain same as RVV\\
- (CSRs are used to "interpret" elements in registers)
- \item CSRs are key-value tables (overlaps allowed)\vspace{10pt}
+ \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: overrides opcode-defined meaning)
+ \item CSRs are key-value tables (overlaps allowed: v. important)
\end{itemize}
- Key differences from RVV:\vspace{10pt}
+ Key differences from RVV:
\begin{itemize}
- \item Predication in INT regs as a BIT field (max VL=XLEN)
+ \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 NO ZEROING: non-predicated elements are skipped
+ \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}
}
\frametitle{ADD pseudocode (or trap, or actual hardware loop)}
\begin{semiverbatim}
-function op_add(rd, rs1, rs2, predr) # add not VADD!
+function op\_add(rd, rs1, rs2, predr) # add not VADD!
int i, id=0, irs1=0, irs2=0;
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}
+ \item Above is oversimplified: Reg. indirection left out (for clarity).
\item SIMD slightly more complex (case above is elwidth = default)
\item Scalar-scalar and scalar-vector and vector-vector now all in one
\item OoO may choose to push ADDs into instr. queue (v. busy!)
\frametitle{Predication-Branch (or trap, or actual hardware loop)}
\begin{semiverbatim}
-s1 = reg_is_vectorised(src1);
-s2 = reg_is_vectorised(src2);
+s1 = reg\_is\_vectorised(src1);
+s2 = reg\_is\_vectorised(src2);
if (!s2 && !s1) goto branch;
for (int i = 0; i < VL; ++i)
- if cmp(s1 ? reg[src1+i] : reg[src1],
- s2 ? reg[src2+i] : reg[src2])
- preg[rs3] |= 1 << i;
+ if (cmp(s1 ? reg[src1+i]:reg[src1],
+ s2 ? reg[src2+i]:reg[src2])
+ ireg[rs3] |= 1<<i;
\end{semiverbatim}
\begin{itemize}
\item If s1 and s2 both scalars, Standard branch occurs
\item Predication stored in integer regfile as a bitfield
\item Scalar-vector and vector-vector supported
+ \item Overload Branch immediate to be predication target rs3
\end{itemize}
\end{frame}
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]
+ if (reg\_is\_vectorised[rs2]) offs = vreg[rs2+i]
else offs = i*(seglen+1)*stride;
vreg[rd+j][i] = mem[sreg[base] + offs + j*stride]
\end{semiverbatim}
\end{frame}
-\frame{\frametitle{Why are overlaps allowed in Regfiles?}
+\frame{\frametitle{Register key-value CSR store (lookup table / CAM)}
\begin{itemize}
- \item Same register(s) can have multiple "interpretations"\vspace{6pt}
- \item xBitManip plus SIMD plus xBitManip = Hi/Lo bitops\vspace{6pt}
- \item (32-bit GREV plus 4x8-bit SIMD plus 32-bit GREV)\vspace{6pt}
- \item RGB 565 (video): BEXTW plus 4x8-bit SIMD plus BDEPW\vspace{6pt}
- \item Same register(s) can be offset (no need for VSLIDE)\vspace{6pt}
+ \item key is int regfile number or FP regfile number (1 bit)
+ \item treated as vector if referred to in op (5 bits, key)
+ \item starting register to actually be used (5 bits, value)
+ \item element bitwidth: default, dflt/2, 8, 16 (2 bits)
+ \item is vector: Y/N (1 bit)
+ \item is packed SIMD: Y/N (1 bit)
+ \item register bank: 0/reserved for future ext. (1 bit)
+ \end{itemize}
+ Notes:
+ \begin{itemize}
+ \item References different (internal) mapping table for INT or FP
+ \item Level of indirection has implications for pipeline latency
+ \item (future) bank bit, no need to extend opcodes: set bank=1,
+ just use normal 5-bit regs, indirection takes care of the rest.
+ \end{itemize}
+}
+
+
+\frame{\frametitle{Register element width and packed SIMD}
+
+ Packed SIMD = N:
+ \begin{itemize}
+ \item default: RV32/64/128 opcodes define elwidth = 32/64/128
+ \item default/2: RV32/64/128 opcodes, elwidth = 16/32/64 with
+ top half of register ignored (src), zero'd/s-ext (dest)
+ \item 8 or 16: elwidth = 8 (or 16), similar to default/2
\end{itemize}
- Note:\vspace{10pt}
+ Packed SIMD = Y (default is moot, packing is 1:1)
+ \begin{itemize}
+ \item default/2: 2 elements per register @ opcode-defined bitwidth
+ \item 8 or 16: standard 8 (or 16) packed SIMD
+ \end{itemize}
+ Notes:
+ \begin{itemize}
+ \item Different src/dest widths (and packs) PERMITTED
+ \item RV* already allows (and defines) how RV32 ops work in RV64\\
+ so just logically follow that lead/example.
+ \end{itemize}
+}
+
+
+\begin{frame}[fragile]
+\frametitle{Register key-value CSR table decoding pseudocode}
+
+\begin{semiverbatim}
+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].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 (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)
+ \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 xBitManip reduces O($N^{6}$) SIMD down to O($N^{3}$)
- \item Hi-Performance: Macro-op fusion (more pipeline stages?)
+ \item Table should be expanded out for high-speed implementations
+ \item Key-value overlaps permitted, but (key+type) must be unique
+ \item RVV rules about deleting higher-indexed CSRs followed
\end{itemize}
}
-\frame{\frametitle{Why no Zeroing (place zeros in non-predicated elements)?}
+\begin{frame}[fragile]
+\frametitle{Predication key-value CSR table decoding pseudocode}
+
+\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 // 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\\
+ (predication disabled)
+ \item Might be a bit complex to set up in hardware (keep as CAM?)
+ \end{itemize}
+
+\end{frame}
+
+
+\begin{frame}[fragile]
+\frametitle{Get Predication value pseudocode}
+
+\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
+ predidx += tb[reg].bank << 5 // 0 (1=rsvd)
+ predicate = intreg[predidx] // actual predicate HERE
+ if (tb[reg].inv):
+ predicate = ~predicate // invert ALL bits
+ return predicate
+\end{semiverbatim}
\begin{itemize}
- \item Zeroing is an implementation optimisation favouring OoO\vspace{8pt}
- \item Simple implementations may skip non-predicated operations\vspace{8pt}
- \item Simple implementations explicitly have to destroy data\vspace{8pt}
+ \item References different (internal) mapping table for INT or FP
+ \item Actual predicate bitmask ALWAYS from the INT regfile
+ \item Hard-limit on MVL of XLEN (predication only 1 intreg)
+ \end{itemize}
+
+\end{frame}
+
+
+\frame{\frametitle{To Zero or not to place zeros in non-predicated elements?}
+
+ \begin{itemize}
+ \item Zeroing is an implementation optimisation favouring OoO
+ \item Simple implementations may skip non-predicated operations
+ \item Simple implementations explicitly have to destroy data
\item Complex implementations may use reg-renames to save power\\
Zeroing on predication chains makes optimisation harder
+ \item Compromise: REQUIRE both (specified in predication CSRs).
\end{itemize}
- Considerations:\vspace{10pt}
+ Considerations:
\begin{itemize}
- \item Complex not really impacted, Simple impacted a LOT
- \item Overlapping "Vectors" may issue overlapping ops
+ \item Complex not really impacted, simple impacted a LOT\\
+ with Zeroing... however it's useful (memzero)
+ \item Non-zero'd overlapping "Vectors" may issue overlapping ops\\
+ (2nd op's predicated elements slot in 1st's non-predicated ops)
\item Please don't use Vectors for "security" (use Sec-Ext)
\end{itemize}
}
% if there are available parallel ALUs to do so.
-\frame{\frametitle{Predication key-value CSR store}
+\frame{\frametitle{Implementation Options}
\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 register to store actual predication in (5 bits, value)\vspace{6pt}
- \item predication is inverted (1 bit)\vspace{6pt}
- \item non-predicated elements are to be zero'd (1 bit)\vspace{6pt}
+ \item Absolute minimum: Exceptions: if CSRs indicate "V", trap.\\
+ (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\\
+ (Do / Don't send through predication to ALU)
+ \item Hardware loop, full parallel ALU (not recommended)
\end{itemize}
- Notes:\vspace{10pt}
+ Notes:\vspace{4pt}
+ \begin{itemize}
+ \item 4 (or more?) options above may be deployed on per-op basis
+ \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}
+}
+% Instr. FIFO may need its own slide. Basically, the vectorised op
+% gets pushed into the FIFO, where it is then "processed". Processing
+% will remove the first set of ops from its vector numbering (taking
+% predication into account) and shoving them **BACK** into the FIFO,
+% but MODIFYING the remaining "vectorised" op, subtracting the now
+% scalar ops from it.
+
+\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}
+ \end{center}
+}
+
+
+\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}
+ \end{center}
+}
+
+
+\frame{\frametitle{Predicated 8-parallel ADD: 3 phase FIFO expansion}
+ \begin{center}
+ \includegraphics[height=2.5in]{padd9_fifo.png}\\
+ {\bf \red First cycle takes first four 1s; second takes the rest}
+ \end{center}
+}
+
+
+\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{How are SIMD Instructions Vectorised?}
+
+ \begin{itemize}
+ \item SIMD ALU(s) primarily unchanged
+ \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:
\begin{itemize}
- \item Table should be expanded out for high-speed implementations
- \item Multiple "keys" (and values) theoretically permitted
- \item RVV rules about deleting higher-indexed CSRs followed
+ \item Many SIMD ALUs possible (parallel execution)
+ \item Implementor free to choose (API remains the same)
+ \item Unused ALU units wasted, but s/w DRASTICALLY simpler
+ \item Very long SIMD ALUs could waste significant die area
\end{itemize}
}
+% With multiple SIMD ALUs at for example 32-bit wide they can be used
+% to either issue 64-bit or 128-bit or 256-bit wide SIMD operations
+% or they can be used to cover several operations on totally different
+% vectors / registers.
+
+\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}
+ \end{center}
+}
-\frame{\frametitle{Register key-value CSR store}
+\frame{\frametitle{Why are overlaps allowed in Regfiles?}
\begin{itemize}
- \item key is int regfile number or FP regfile number (1 bit)\vspace{6pt}
- \item treated as vector if referred to in op (5 bits, key)\vspace{6pt}
- \item starting register to actually be used (5 bits, value)\vspace{6pt}
- \item element bitwidth: default/8/16/32/64/rsvd (3 bits)\vspace{6pt}
- \item element type: still under consideration\vspace{6pt}
+ \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:\\
+ GREV @ VL=N,wid=32; SIMD @ VL=Nx4,wid=8)
+ \item RGB 565 (video): BEXTW plus 4x8-bit SIMD plus BDEPW\\
+ (BEXT/BDEP @ VL=N,wid=32; SIMD @ VL=Nx4,wid=8)
+ \item Same register(s) can be offset (no need for VSLIDE)\vspace{6pt}
\end{itemize}
- Notes:\vspace{10pt}
+ Note:
\begin{itemize}
- \item Same notes apply (previous slide) as for predication CSR table
- \item Level of indirection has implications for pipeline latency
+ \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/ no pred): VSPLAT
\item scalar-to-vector (w/ dest-pred): Sparse VSPLAT
\item scalar-to-vector (w/ 1-bit dest-pred): VINSERT
- \item vector-to-scalar (w/ src-pred): VEXTRACT
+ \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 Really powerful!
- \item Any other options?
+ \item Surprisingly powerful! Zero-predication even more so
+ \item Same arrangement for FCVT, FMV, FSGNJ etc.
\end{itemize}
}
+\begin{frame}[fragile]
+\frametitle{MV pseudocode with predication}
+
+\begin{semiverbatim}
+function op\_mv(rd, rs) # MV not VMV!
+ rd = int\_vec[rd].isvector ? int\_vec[rd].regidx : rd;
+ rs = int\_vec[rs].isvector ? int\_vec[rs].regidx : rs;
+ ps = get\_pred\_val(FALSE, rs); # predication on src
+ pd = get\_pred\_val(FALSE, rd); # ... AND on dest
+ for (int i = 0, int j = 0; i < VL && j < VL;):
+ if (int\_vec[rs].isvec) while (!(ps \& 1<<i)) i++;
+ if (int\_vec[rd].isvec) while (!(pd \& 1<<j)) j++;
+ ireg[rd+j] <= ireg[rs+i];
+ if (int\_vec[rs].isvec) i++;
+ if (int\_vec[rd].isvec) j++;
+\end{semiverbatim}
+
+ \begin{itemize}
+ \item elwidth != default not covered above (might be a bit hairy)
+ \item Ending early with 1-bit predication not included (VINSERT)
+ \end{itemize}
+\end{frame}
+
+
+\begin{frame}[fragile]
+\frametitle{VSELECT: stays or goes? Stays if MV.X exists...}
+
+\begin{semiverbatim}
+def op_mv_x(rd, rs): # (hypothetical) RV MX.X
+ rs = regfile[rs] # level of indirection (MV.X)
+ regfile[rd] = regfile[rs] # straight regcopy
+\end{semiverbatim}
+
+Vectorised version aka "VSELECT":
+
+\begin{semiverbatim}
+def op_mv_x(rd, rs): # SV version of MX.X
+ for i in range(VL):
+ rs1 = regfile[rs+i] # indirection
+ regfile[rd+i] = regfile[rs] # straight regcopy
+\end{semiverbatim}
+
+ \begin{itemize}
+ \item However MV.X does not exist in RV, so neither can VSELECT
+ \item \red SV is not about adding new functionality, only parallelism
+ \end{itemize}
+
+
+\end{frame}
+
+
\frame{\frametitle{Opcodes, compared to RVV}
\begin{itemize}
- \item All integer and FP opcodes all removed (no CLIP!)\vspace{8pt}
- \item VMPOP, VFIRST etc. all removed (use xBitManip)\vspace{8pt}
- \item VSLIDE removed (use regfile overlaps)\vspace{8pt}
- \item C.MV covers VEXTRACT VINSERT and VSPLAT (and more)\vspace{8pt}
- \item VSETVL, VGETVL, VSELECT stay\vspace{8pt}
- \item Issue: VCLIP is not in RV* (add with custom ext?)\vspace{8pt}
- \item Vector (or scalar-vector) use C.MV (MV is a pseudo-op)\vspace{8pt}
- \item VMERGE: twin predicated C.MVs (one inverted. macro-op'd)\vspace{8pt}
+ \item All integer and FP opcodes all removed (no CLIP, FNE)
+ \item VMPOP, VFIRST etc. all removed (use xBitManip)
+ \item VSLIDE removed (use regfile overlaps)
+ \item C.MV covers VEXTRACT VINSERT and VSPLAT (and more)
+ \item Vector (or scalar-vector) copy: use C.MV (MV is a pseudo-op)
+ \item VMERGE: twin predicated C.MVs (one inverted. macro-op'd)
+ \item VSETVL, VGETVL stay (the only ops that do!)
+ \end{itemize}
+ Issues:
+ \begin{itemize}
+ \item VSELECT stays? no MV.X, so no (add with custom ext?)
+ \item VSNE exists, but no FNE (use predication inversion?)
+ \item VCLIP is not in RV* (add with custom ext? or CSR?)
\end{itemize}
}
-\frame{\frametitle{Under consideration}
+\begin{frame}[fragile]
+\frametitle{Example c code: DAXPY}
+
+\begin{semiverbatim}
+ void daxpy(size_t n, double a,
+ const double x[], double y[])
+ \{
+ for (size_t i = 0; i < n; i++) \{
+ y[i] = a*x[i] + y[i];
+ \}
+ \}
+\end{semiverbatim}
+
+ \begin{itemize}
+ \item See "SIMD Considered Harmful" for SIMD/RVV analysis\\
+ https://sigarch.org/simd-instructions-considered-harmful/
+ \end{itemize}
+
+
+\end{frame}
+
+
+\begin{frame}[fragile]
+\frametitle{RVV DAXPY assembly (RV32V)}
+
+\begin{semiverbatim}
+# a0 is n, a1 is ptr to x[0], a2 is ptr to y[0], fa0 is a
+ li t0, 2<<25
+ vsetdcfg t0 # enable 2 64b Fl.Pt. registers
+loop:
+ setvl t0, a0 # vl = t0 = min(mvl, n)
+ vld v0, a1 # load vector x
+ slli t1, t0, 3 # t1 = vl * 8 (in bytes)
+ vld v1, a2 # load vector y
+ add a1, a1, t1 # increment pointer to x by vl*8
+ vfmadd v1, v0, fa0, v1 # v1 += v0 * fa0 (y = a * x + y)
+ sub a0, a0, t0 # n -= vl (t0)
+ vst v1, a2 # store Y
+ add a2, a2, t1 # increment pointer to y by vl*8
+ bnez a0, loop # repeat if n != 0
+\end{semiverbatim}
+\end{frame}
+
+
+\begin{frame}[fragile]
+\frametitle{SV DAXPY assembly (RV64D)}
+
+\begin{semiverbatim}
+# a0 is n, a1 is ptr to x[0], a2 is ptr to y[0], fa0 is a
+ 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(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
+ add a1, a1, t1 # increment pointer to x by vl*8
+ fmadd a7, a3, fa0, a7 # v1 += v0 * fa0 (y = a * x + y)
+ sub a0, a0, t0 # n -= vl (t0)
+ st a7, a2 # store 4 registers a7-10 to y
+ add a2, a2, t1 # increment pointer to y by vl*8
+ bnez a0, loop # repeat if n != 0
+\end{semiverbatim}
+\end{frame}
+
+
+\frame{\frametitle{Under consideration (some answers documented)}
\begin{itemize}
- \item Is C.FNE actually needed? Should it be added if it is?
- \item Element type implies polymorphism. Should it be in SV?
+ \item Should future extra bank be included now?
+ \item How many Register and Predication CSRs should there be?\\
+ (and how many in RV32E)
+ \item How many in M-Mode (for doing context-switch)?
\item Should use of registers be allowed to "wrap" (x30 x31 x1 x2)?
- \item Is detection of all-scalar ops ok (without slowing pipeline)?
- \item Can VSELECT be removed? (it's really complex)
\item Can CLIP be done as a CSR (mode, like elwidth)
\item SIMD saturation (etc.) also set as a mode?
+ \item Include src1/src2 predication on Comparison Ops?\\
+ (same arrangement as C.MV, with same flexibility/power)
\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}
}
\frame{\frametitle{What's the downside(s) of SV?}
\begin{itemize}
\item EVERY register operation is inherently parallelised\\
- (scalar ops are just vectors of length 1)\vspace{8pt}
- \item An extra pipeline phase is pretty much essential\\
- for fast low-latency implementations\vspace{8pt}
- \item Assuming an instruction FIFO, N ops could be taken off\\
- of a parallel op per cycle (avoids filling entire FIFO;\\
- also is less work per cycle: lower complexity / latency)\vspace{8pt}
+ (scalar ops are just vectors of length 1)\vspace{4pt}
+ \item Tightly coupled with the core (instruction issue)\\
+ could be disabled through MISA switch\vspace{4pt}
+ \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).
- \end{itemize}
-}
-
-
-\frame{\frametitle{Is this OK (low latency)? Detect scalar-ops (only)}
- \begin{center}
- \includegraphics[height=2.5in]{scalardetect.png}\\
- {\bf \red Detect when all registers are scalar for a given op}
- \end{center}
-}
-
-
-\frame{\frametitle{TODO (break into separate slides)}
-
- \begin{itemize}
- \item Then explain why this proposal is a good way to \\
- abstract parallelism\\
- (hopefully also explaining how \\
- a good compiler can make clever use of this increase parallelism\\
- Then explain how this can be implemented (at instruction\\
- issue time???) with\\
- implementation options, and what these "cost".\\
- Finally give examples that show simple usage that compares\\
- C code\\
- RVIC\\
- RVV\\
- RVICXsimplev
+ 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)
\end{itemize}
}
\frame{\frametitle{Summary}
\begin{itemize}
- \item Actually about parallelism, not Vectors (or SIMD) per se
+ \item Actually about parallelism, not Vectors (or SIMD) per se\\
+ and NOT about adding new ALU/logic/functionality.
+ \item Only needs 2 actual instructions (plus the CSRs).\\
+ RVV - and "standard" SIMD - require ISA duplication
\item Designed for flexibility (graded levels of complexity)
\item Huge range of implementor freedom
\item Fits RISC-V ethos: achieve more with less
\item Reduces SIMD ISA proliferation by 3-4 orders of magnitude \\
(without SIMD downsides or sacrificing speed trade-off)
\item Covers 98\% of RVV, allows RVV to fit "on top"
- \item Not designed for supercomputing (that's RVV), designed for
- in between: DSPs, RV32E, Embedded 3D GPUs etc.
- \item Not specifically designed for Vectorisation: designed to\\
- reduce code size (increase efficiency, just
- like Compressed)
- \end{itemize}
-}
-
-
-\frame{\frametitle{slide}
-
- \begin{itemize}
- \item \vspace{10pt}
- \end{itemize}
- Considerations:\vspace{10pt}
- \begin{itemize}
- \item \vspace{10pt}
+ \item Byproduct of SV is a reduction in code size, power usage
+ etc. (increase efficiency, just like Compressed)
\end{itemize}
}
\frame{
\begin{center}
- {\Huge \red The end\vspace{20pt}\\
- Thank you}
+ {\Huge The end\vspace{20pt}\\
+ Thank you\vspace{20pt}\\
+ Questions?\vspace{20pt}
+ }
\end{center}
+
+ \begin{itemize}
+ \item Discussion: ISA-DEV mailing list
+ \item http://libre-riscv.org/simple\_v\_extension/
+ \end{itemize}
}
projects / libreriscv.git / blobdiff