Skip to content

Octobus ND-5000 CPU-presence detection - CARVE ANSWER (2026-07-18)

Answers E:\Dev\Ronny\ND5000UC\CARVER-REQUEST-OCTOBUS-CPU-PRESENCE-2026-07-18.md. Sources: ../NPL-SOURCE/NPL/MP-P2-N500.NPL (5OMBREAD 3453-3536, CON5IDENT 3614-3634), ../NPL-SOURCE/NPL/PH-P2-OPPSTART.NPL (CH5CPUPRESENT 3903-3945), and SINTRAN-OCTOBUS-MESSAGE-CATALOG.md (sections 4.3/5/6). Tags: [V] = byte/source-cited; [I] = inferred; [OPEN] = not carved.

VERDICT (one line)

Mechanism (a): an octobus multibyte ACK frame. SINTRAN's "CPU present" = the 5ALIVE bit in CPUAVAILABLE, and that bit is set only by 5OMBREAD when the ND-100 receives a multibyte message from the SAMSON station whose ETYPE high byte = MFACK (0) - the Ack to the CMSYSPAR ("WriteSysPar / alive") message. Your station consumes the CMSYSPAR but sends no such Ack, so 5ALIVE never sets and the monitor prints "No ND-500(0) CPU found." NOT a 3RMICV mailbox answer, NOT a GIVEINT, NOT an ACCP status bit. [V]

Q1 - the presence flag and who sets it

"CPU present" is decided by CPUAVAILABLE BONE 5ALIVE - set at MP-P2-N500.NPL:3470 (octal 146616), inside 5OMBREAD: 

146565  IF "LMFIELD".MOCTSTATION>=FN5DEST AND A<=LN5DEST THEN   % source is a SAMSON (70B..73B)
146575     X.ETYPE=:D SHZ -10=:CSTS ; A:=D/\377=:CMICP           % CSTS = ETYPE high byte
146607     IF CSTS=MFACK OR A=MFNACK THEN                        % Ack/Nack on WriteSysPar
146616        CPUAVAILABLE BONE 5ALIVE=:CPUAVAILABLE ; GO I5OMBR % <-- "I'm present"
[V]. CON5IDENT sends the alive message and returns via GO I5OMBR to wait for exactly this Ack (its own comment: "Send 'alive' message to the ACCP to verify that it's present ... Ack/Nack answer is handled by 5OMBREAD", 3606-3609). [V]

The J04 monitor's "No ND-500(0) CPU found" print-site is not itself carved (the message catalog section 9 notes J04's octobus paths are unanalyzed) - but the flag it reflects is 5ALIVE. The J04 monitor runs the ND-500 bring-up through SINTRAN; with 5ALIVE clear it declares the CPU absent. [V for the flag; [OPEN] for the exact J04 test address.]

Q2 - what response is expected after the @nd-500 CMSYSPAR

(a) An octobus reply frame - specifically a multibyte message that 5OMBREAD reads off the reserved OMD (5OMDNO) via OMBREAD (3462), from the SAMSON source station, carrying: - source station in MOCTSTATION in range FN5DEST..LN5DEST (the SAMSON, 70B). [V, 3463] - word at the ETYPE position with high byte = MFACK (= 0) (low byte CMICP = reporting source, e.g. 1). CSTS = ETYPE >> 8 = MFACK is the trigger. [V, 3464-3467]

That is all 5OMBREAD needs to set 5ALIVE. Not (b)/(c)/(d): - (b) 3RMICV/CPU-type is a LATER mailbox exchange, not the presence gate. [I] - (c) no GIVEINT/level-13 doorbell is tested for presence; the Ack rides the normal multibyte-message receive path (whatever OMD interrupt/ID12 already drives 5OMBREAD). [V - 5OMBREAD is the only 5ALIVE writer] - (d) no ACCP status-register bit is read for presence. [V]

Emulator action (OctobusND5000Station): after consuming the CMSYSPAR, send a multibyte message back on OMD 5OMDNO (= 3 here; see Q4) from your SAMSON station (70B) with the command/ETYPE word high byte = MFACK (0). A minimal ETYPE = 0x0000 (or 0x0001 with CMICP=1) satisfies CSTS=MFACK, MMSGLENGTH = 1. Then 5OMBREAD sets 5ALIVE and @nd-500 finds the CPU. [V for the trigger condition; the exact SOMB/EOMB byte framing is the same multibyte envelope your station already PARSES for the inbound CMSYSPAR, emitted in reverse.]

Q3 - why boot differs from @nd-500

They mark presence by two different bits: - Boot CH5CPUPRESENT (PH-P2-OPPSTART.NPL:3903-3945) probes the card (IOXT 100406, wait RFT bit 3), sends emergencies 241B (CMMACLE master-clear) + 242B (CMACONT continue-ACCP), then sets MIFLAG MUDOM + CPUAVAILABLE := SAMSON - the interface-type bit. It does not set 5ALIVE and does not send CMSYSPAR. (Contrast: the OLD-500/3022 branch sets OLD500 + 5ALIVE immediately, because a 3022 is alive the moment it answers RSTA5.) [V, catalog 6.1 / lines 178,186] - @nd-500 runs the 5PIT warm bring-up XX5CONOMD (RP-P2-N500.NPL:944-999): CON5OMD (connect receive OMD) -> MFPREPARE per MF station -> CON5IDENT per SAMSON, whose Ack via 5OMBREAD sets 5ALIVE. [V, catalog 6.3]

So boot establishes "a SAMSON interface exists" (card + emergencies), and @nd-500 performs the alive handshake that the monitor actually gates "CPU found" on. This is exactly your trace: boot = 241B/242B only; @nd-500 = the CMSYSPAR whose missing Ack is the failure. Boot does not re-verify aliveness; @nd-500 is where presence is truly confirmed for a SAMSON. [V]

Q4 - decode of CMSYSPAR body 03 07 0E 01 03 00 00 00 00

Against the CON5IDENT builder (MP-P2-N500.NPL:3622-3632): | byte | value | field | meaning | |---|---|---|---| | [0] | 03 | MOCTOMD | OMD number = OMDACCP = 3 (matches the SOMB/EOMB OMD 3 frames) | | [1] | 07 | MMSGLENGTH | message length = 7 | | [2] | 0E | MCOMMAND hi | CMSYSPAR = 016B = 0x0E ("write system parameters") | | [3] | 01 | MCOMMAND lo | N100IDENT = 1 (the ND-100's octobus ident) | | [4] | 03 | S5 hi | 5OMDNO SHZ 10 = 5OMDNO = 3 = the reply-to OMD the host listens on | | [5..8] | 00 00 00 00 | S5 lo / S6 / S7 | zero (0=:X.S6=:X.S7, 3629) |

[V]. Yes - the message carries the reply-to OMD: field S5 (byte[4]) = 5OMDNO (= 3). The ACCP must send its Ack back on that OMD. In this config OMDACCP = 5OMDNO = 3, so replying on the OMD you received it on is also correct, but the authoritative source is S5/byte[4]. [V]

(MBROADCAST = 0 at builder line 3624 is folded into the MMSGLENGTH/MOCTOMD framing, not a separate visible payload byte here.) [I]

Bottom line for the emulator

Add to OctobusND5000Station: on receiving a CMSYSPAR (016B) on OMD 3, reply with a multibyte message on OMD 5OMDNO (= S5/body[4] = 3) from station 70B whose command/ETYPE word has high byte MFACK (0). That single Ack sets CPUAVAILABLE.5ALIVE in 5OMBREAD, and @nd-500 will find CPU 0. No mailbox 3RMICV, no GIVEINT, no status bit is required for presence. [V]

Still OPEN (honest)

  • The exact J04 monitor instruction that reads 5ALIVE/prints "No ND-500(0) CPU found" (J04 octobus paths not carved; the SINTRAN-side flag is proven). [OPEN]
  • Whether the ACCP's real Ack also carries a nonzero CMICP/payload (5OMBREAD ignores it for the present/absent decision - only CSTS=MFACK matters). [I]
  • Exact on-wire SOMB/EOMB byte packing of the reply (mirror of the inbound envelope your station already parses). [I]

Status-of-record: SINTRAN-OCTOBUS-MESSAGE-CATALOG.md section 4.3/10 (reply shape now [V] - see update).