Flanger — HLE reconstruction and DSP bytecode

DSPHLE selector 0x04 · program image prog04_flanger · Intervention-pinned (a panel control was driven and the moving C-RAM cell identified).

swept delay + feedback; rate=0x05, feedback=0x00, depth=0x08.

This page pairs the bytecode the effect runs on the NEC µPD6383GF with the high-level reconstruction that makes it audible in MAME. The reconstruction is not the chip’s microcode — it is a textbook DSP block built from the decoded meaning of that microcode, and reading the two together is how the bytecode is understood (and, in time, driven toward a faithful low-level emulation). See also the signal-flow flowchart.

The bytecode below is the source of truth; the HLE reconstruction is not. The HLE is our best current interpretation and may contain mistakes — where the two disagree, the bytecode wins, and a better HLE should be updated here. This page is the permanent archive of the reference HLE: it is kept here even after the code is eventually removed from the MAME sources.

DSP bytecode (reverse-engineered microprogram)

The disassembled image the chip executes for this effect. Source (regenerable): dsp/disasm/prog04_flanger.dsm.

; Reverse-engineered disassembly of the KN5000 effects-DSP microcode
; (NEC uPD6383GF), recovered from the original firmware.  The microcode is the
; work of its original authors; this is a disassembly for preservation and
; interoperability, and no claim of copyright is made over the disassembled program.
; KN5000 effects-DSP program -- FLANGER
; image rep algo 4  |  slots 4  |  unit 0 (I-RAM load 84)
; family modulation  |  confidence medium  |  65 words, 18 class-A multiplies (12 named)
; role: flanger: swept all-pass chain, 0.3 feedback, wet mix
; coefficient cursor base 0x00
;
; GENERATED by dsp/tools/gen_dsp_disasm.py -- DO NOT EDIT.
; Put labels/comments in the matching dsp/sym/*.sym; analysis in dsp/algorithms/.

  w0    088013000B   ?word   0x088013000B   ; 880.1.30.00B  hi12{ESC ?7 res=080}  [external delay-DRAM READ (FORCED, adjudication-round5 sect. 3 -- addr8 bit 6 is the direction field and 0x60 is the WRITE; this REVERSES R1 F1, which bounded the read latency to one repetition when the descriptors need twenty words); this end moves with the user's DELAY (ms) knob, and the delay is READ_CELL - WRITE_CELL; addr8 0x30 also marks the FIRST DRAM access of a body, 37 of 38 distinct images (R3 sect. 6.2). external delay-DRAM access; address = DESCRIPTOR_CELL[k] + G, from the host bank behind pointer ...825 / tag 0x4C (R3, PROVEN BY CONSTRUCTION) -- the k-th class-1 escape word of a body takes the k-th cell of that body's own descriptor block (the IDENTITY map, FORCED in adjudication-round5 sect. 1), so the address is NOT in this word]
  w1    00400008BC   ?word   0x00400008BC   ; 040.0.00.8BC  hi12{?6 res=040}  [SPECULATIVE: lo12 bit-11 modifier word + pointer-mode (bit11-family); the base selector/register is OPEN]
  w2    00002051CD   ld      (p),(p)+5
  w3    000020040E   ld      acc,(p)+0
  w4    0212A011D5   mac     (p),c+,(p)+1 ; mem[p]<-acc, acc=0
        ; C-RAM[0x00] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x00] = filter section cell 0 (role filter, INFERRED)
  w5    0202A451D5   mac     (p),c+,(p)+69
        ; C-RAM[0x01] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x01] = filter section cell 1 (role filter, INFERRED)
  w6    0202AC01D5   mac     (p),c+,(p)-64
        ; C-RAM[0x02] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x02] = filter section cell 2 (role filter, INFERRED)
  w7    09001602D9   dly.w  dsc[k],p+96
  w8    0192AFB1D5   mac     (p),c+,(p)-5 ; store SUPPRESSED (bit7)
        ; C-RAM[0x03] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x03] = filter section cell 3 (role filter, INFERRED)
  w9    0182245000   mac.b   (p)0,(p)+69
  w10   000020044C   ?word   0x000020044C   ; 000.2.00.44C  hi12{-}  [SPECULATIVE (prospective, not measured): SRC 0x11 = ACCB (2nd accumulator); ACT 0x0C = delay READ]
  w11   088012040B   ?word   0x088012040B   ; 880.1.20.40B  hi12{ESC ?7 res=080}  [external delay-DRAM READ (FORCED, adjudication-round5 sect. 3 -- addr8 bit 6 is the direction field and 0x60 is the WRITE; this REVERSES R1 F1, which bounded the read latency to one repetition when the descriptors need twenty words); this end moves with the user's DELAY (ms) knob, and the delay is READ_CELL - WRITE_CELL. external delay-DRAM access; address = DESCRIPTOR_CELL[k] + G, from the host bank behind pointer ...825 / tag 0x4C (R3, PROVEN BY CONSTRUCTION) -- the k-th class-1 escape word of a body takes the k-th cell of that body's own descriptor block (the IDENTITY map, FORCED in adjudication-round5 sect. 1), so the address is NOT in this word]
  w12   0012201655   mac     ta,(p)+1 ; mem[p]<-acc, acc=0
  w13   01042001D5   post    (p),(p)+0
  w14   0A00000041   ?word   0x0A00000041   ; A00.0.00.041  hi12{ESC f98=2}  [head of a fixed 3-word template (S-5); operands SRC 0x01/ACT 0x01 dark]
  w15   08801202C7   dly.r  dsc[k],p+32
  w16   0102AB94C8   ?word   0x0102AB94C8   ; 102.A.B9.4C8  hi12{f98=1 f31=1} cur+  [gain multiply (same op in phaser all-pass and reverb diffuser)]
        ; C-RAM[0x04] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x04] = filter section cell 4 (role filter, INFERRED)
  w17   0000204407   ld.st   acc,(p)+4
  w18   00002FA407   ld.st   acc,(p)-6
  w19   0092A00200   mac.b   c,c+,(p)+0 ; store SUPPRESSED (bit7)
        ; C-RAM[0x05] (coeff, base 0x00 MEASURED)
  w20   00822001C0   mac.b   (p),(p)+0
  w21   0094A00200   wrap    acc,c+          ; acc <- datum(acc) & coef  (LFO modulus)
        ; C-RAM[0x06] (coeff, base 0x00 MEASURED)
  w22   0000200447   ?word   0x0000200447   ; 000.2.00.447  hi12{-}  [SPECULATIVE (prospective, not measured): SRC 0x11 = ACCB (2nd accumulator)]
  w23   0092200700   ?word   0x0092200700   ; 092.2.00.700  hi12{ST f31=1 ?7 res=080}  [SPECULATIVE (prospective, not measured): SRC 0x1C = control/mod source into MAC (100% MAC-consumed; LFO in mod fx, envelope/AGC in dynamics) -- NOT LFO-only: present in 19 non-LFO programs (dsp_datapath_fingerprint)]
  w24   0000A001D5   ld      (p),c+,(p)+0
        ; C-RAM[0x07] (coeff, base 0x00 MEASURED)
  w25   0182200000   mac.b   (p)0,(p)+0
  w26   0040000C63   ?word   0x0040000C63   ; 040.0.00.C63  hi12{?6 res=040}  [SPECULATIVE (prospective, not measured): SRC 0x11 = ACCB (2nd accumulator)]
  w27   00006184CD   ?word   0x00006184CD   ; 000.6.18.4CD  hi12{-}  [table-lookup idiom, class-6 addr8 = table selector (INFERRED)]
  w28   00124011CE   ?word   0x00124011CE   ; 012.4.01.1CE  hi12{ST f31=1}  [table-lookup idiom, third word (INFERRED)]
  w29   01042081CE   post    (p),(p)+8
  w30   0102200000   mac.b   (p)0,(p)+0
  w31   0000A00415   ld      acc,c+,(p)+0
        ; C-RAM[0x08] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x08] = op0x66[0] (role mix/tap, INFERRED)
  w32   0212200000   mac.b   (p)0,(p)+0 ; mem[p]<-acc, acc=0
  w33   00002F9407   ld.st   acc,(p)-7
  w34   0092A00200   mac.b   c,c+,(p)+0 ; store SUPPRESSED (bit7)
        ; C-RAM[0x09] (coeff, base 0x00 MEASURED)
  w35   00822001C0   mac.b   (p),(p)+0
  w36   0094A00200   wrap    acc,c+          ; acc <- datum(acc) & coef  (LFO modulus)
        ; C-RAM[0x0A] (coeff, base 0x00 MEASURED)
  w37   0000200447   ?word   0x0000200447   ; 000.2.00.447  hi12{-}  [SPECULATIVE (prospective, not measured): SRC 0x11 = ACCB (2nd accumulator)]
  w38   0092200700   ?word   0x0092200700   ; 092.2.00.700  hi12{ST f31=1 ?7 res=080}  [SPECULATIVE (prospective, not measured): SRC 0x1C = control/mod source into MAC (100% MAC-consumed; LFO in mod fx, envelope/AGC in dynamics) -- NOT LFO-only: present in 19 non-LFO programs (dsp_datapath_fingerprint)]
  w39   0000A001D5   ld      (p),c+,(p)+0
        ; C-RAM[0x0B] (coeff, base 0x00 MEASURED)
  w40   0182200000   mac.b   (p)0,(p)+0
  w41   0040000C63   ?word   0x0040000C63   ; 040.0.00.C63  hi12{?6 res=040}  [SPECULATIVE (prospective, not measured): SRC 0x11 = ACCB (2nd accumulator)]
  w42   00006184CD   ?word   0x00006184CD   ; 000.6.18.4CD  hi12{-}  [table-lookup idiom, class-6 addr8 = table selector (INFERRED)]
  w43   00124011CE   ?word   0x00124011CE   ; 012.4.01.1CE  hi12{ST f31=1}  [table-lookup idiom, third word (INFERRED)]
  w44   01042FE1CE   post    (p),(p)-2
  w45   0102200000   mac.b   (p)0,(p)+0
  w46   0000AFE415   ld      acc,c+,(p)-2
        ; C-RAM[0x0C] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x0C] = op0x66[1] (role mix/tap, INFERRED)
  w47   021220A1CD   mac     (p),(p)+10 ; mem[p]<-acc, acc=0
  w48   00002F81CE   ld      (p),(p)-8
  w49   0212204407   mac.st  acc,(p)+4 ; mem[p]<-acc, acc=0
  w50   0000AFF1D5   ld      (p),c+,(p)-1
        ; C-RAM[0x0D] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x0D] = filter section cell 0 (role filter, INFERRED)
  w51   0202A481D5   mac     (p),c+,(p)+72
        ; C-RAM[0x0E] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x0E] = filter section cell 1 (role filter, INFERRED)
  w52   0202AB51D5   mac     (p),c+,(p)-75
        ; C-RAM[0x0F] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x0F] = filter section cell 2 (role filter, INFERRED)
  w53   09001602D9   dly.w  dsc[k],p+96
  w54   0192A031D5   mac     (p),c+,(p)+3 ; store SUPPRESSED (bit7)
        ; C-RAM[0x10] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x10] = filter section cell 3 (role filter, INFERRED)
  w55   0182248000   mac.b   (p)0,(p)+72
  w56   000020044C   ?word   0x000020044C   ; 000.2.00.44C  hi12{-}  [SPECULATIVE (prospective, not measured): SRC 0x11 = ACCB (2nd accumulator); ACT 0x0C = delay READ]
  w57   088012040B   ?word   0x088012040B   ; 880.1.20.40B  hi12{ESC ?7 res=080}  [external delay-DRAM READ (FORCED, adjudication-round5 sect. 3 -- addr8 bit 6 is the direction field and 0x60 is the WRITE; this REVERSES R1 F1, which bounded the read latency to one repetition when the descriptors need twenty words); this end moves with the user's DELAY (ms) knob, and the delay is READ_CELL - WRITE_CELL. external delay-DRAM access; address = DESCRIPTOR_CELL[k] + G, from the host bank behind pointer ...825 / tag 0x4C (R3, PROVEN BY CONSTRUCTION) -- the k-th class-1 escape word of a body takes the k-th cell of that body's own descriptor block (the IDENTITY map, FORCED in adjudication-round5 sect. 1), so the address is NOT in this word]
  w58   0012201655   mac     ta,(p)+1 ; mem[p]<-acc, acc=0
  w59   01042001D5   post    (p),(p)+0
  w60   0A00000041   ?word   0x0A00000041   ; A00.0.00.041  hi12{ESC f98=2}  [head of a fixed 3-word template (S-5); operands SRC 0x01/ACT 0x01 dark]
  w61   08801202C7   dly.r  dsc[k],p+32
  w62   0102AB84C8   ?word   0x0102AB84C8   ; 102.A.B8.4C8  hi12{f98=1 f31=1} cur+  [gain multiply (same op in phaser all-pass and reverb diffuser)]
        ; C-RAM[0x11] (coeff, base 0x00 MEASURED)
        ; coeff C-RAM[0x11] = filter section cell 4 (role filter, INFERRED)
  w63   0880160000   dly.w  dsc[k],p+96
  w64   040010E407   endblk  #0E             ; END OF BLOCK -- the last word of a block

HLE reconstruction (MAME, kn5000_tonegen.cpp)

The decode of the bytecode above, rebuilt as audible DSP. Two parts: the parameter refresh (reads the decoded C-RAM coefficient cells once per update) and the per-sample insert (the signal processing). Default OFF, behind the DSPHLE research port. Source: src/mame/matsushita/kn5000_tonegen.cpp.

Parameter refresh (decode the C-RAM coefficients):

	const bool fl_hle = (dsphle == 4);
	double fl_inc = 0.0, fl_depth = 0.0, fl_base = 0.0, fl_fb = 0.0;
	if (fl_hle)
	{
		auto q22f = [](u32 v) -> double {
			int32_t s = (v & 0x800000) ? int32_t(v) - 0x1000000 : int32_t(v);
			return double(s) / 4194304.0; };
		const bool bit1 = (dspcfg & 2) != 0;
		const double cs = bit1 ? 1.0 : 2.0;                        // operand -> cell scale
		const double inc = double(m_dsp1->cram_read(0x05) & 0xffffff) * cs;   // phase incr (cell scale)
		const double f_hz = inc * 44100.0 / 8388608.0;
		fl_inc = std::clamp(f_hz, 0.02, 12.0) / double(STREAM_RATE);
		// DEPTH cell (0..1) scales a max LFO sweep of ~5 ms (240 samples @44.1k), -> stream rate
		const double depth01 = std::clamp(q22f(m_dsp1->cram_read(0x08)) * cs, 0.0, 1.0);
		fl_depth = depth01 * 240.0 * double(STREAM_RATE) / 44100.0;
		fl_base  = 2.0;                                            // very short base (high notches)
		fl_fb    = std::clamp(q22f(m_dsp1->cram_read(0x00)) * cs * 0.5, -0.95, 0.95);  // feedback = operand
	}

Per-sample insert (the reconstructed signal path):

		if (fl_hle)
		{
			m_fl_phase += fl_inc;
			if (m_fl_phase >= 1.0) m_fl_phase -= 1.0;
			const double d = fl_base + fl_depth * (0.5 + 0.5 * std::sin(2.0 * M_PI * m_fl_phase));
			const double sl = double(mix_l) / 32768.0, sr = double(mix_r) / 32768.0;
			const double tl = m_fl_l.read(d), tr = m_fl_r.read(d);
			m_fl_l.write(sl + fl_fb * tl);
			m_fl_r.write(sr + fl_fb * tr);
			mix_l = int32_t((0.5 * sl + 0.5 * tl) * 32768.0);
			mix_r = int32_t((0.5 * sr + 0.5 * tr) * 32768.0);
		}