possible "enhancement" for kl10 NI20

[jeffgunter]

my tops10 hiccups with a KNIWUV stopcode during system startup.

the last two machines our company ran on were the sc40 and then the klh10. both of these machines don't actually require that the NI ucode be loaded from boot, and so we didn't load the ucode at all.

tops10, in routine INIKNI/INIKN1, the ucode version is preferred to be MINUVR, or 167.

i'm assuming that the NI20 doesn't actually execute DEC's NI20 ucode. not sure how simh's kl10 NI20 handles being "loaded", but if my tops10 reads station info, the default "unloaded" ucode version word is 0,,1717, so the ucode version # is seen as zero. if the NI20 "runs" without actually being loaded with ucode, then maybe the default "unloaded" ucode version field 3,,770000 ought to be 1,,670000 (or higher) so that tops10 will be happy with the ucode version number ? ie: something like 1,,671717 ...

my current machine has a ucode version word of 1,,721717 (ie: version 172); don't recall what the sc40 returned.

-jfg

[rcornwell]

I support tracking the microcode by saving locations 274 - 277. If your versions stores it at a different location I can expand the range.

[jeffgunter]

i interpret your comment to mean that the expectation is that valid dec ni ucode is "loaded" into the NI20, by boot or knildr, and then the simulator makes note of info in ucode adresses 274-277 and provides that back to the pdp10 code on reading the station address ... ? does this imply that the emulated NI20 really executes the code it is given, or does it just keep track of those locations solely for the purpose of returning version info etc ? i can probably dig up the real microcode and brew a version of boot that will load it.

if the emulated ni20 doesn't actually execute dec ni20 ucode, then maybe it should return version info related to the version of the ucode that it does attempt to emulate... ?

in any case, not complaining, i can live with teaching boot to load the real ucode just for that purpose. or neutering the pdp10 code in tops10 kniser that checks it.

thanx,
-jfg

[rcornwell]

I pretty much copied the microcode checking from Klh10. And no I do not execute the microcode. I am not sure if there is sufficient documentation actually emulate it. Under Tops 10 is verifies some location (274-277) to validate that the microcode loaded. I emulate versions 0172.

[jeffgunter]

no need to change anything, unless you're doing it as a voluntary enhancement. as i say, we don't have the ni ucode in the version of boot we normally run because it isn't necessary on the sc40 or klh10. the sc40 and the klh10 apparently do not need to have any dummy loads sent to them in order to return a reasonable ucode version number. not sure as to whether they actually do return the same number as given if a load is done, but they return a valid number even if a load is not done.

i'm happy, i loaded up boot with the ni ucode to make the kniwuv stopcode go away. as the title says "possible enhancement".

[rcornwell]

I assume you will load the microcode, which should be automatic. If you don't load the microcode then you will have empty machine. I don't do a SC40 so no idea about whether microcode was loaded or not.

Can we close issues that are resolved?

[jeffgunter]

sure, close any issues you like.  maybe i went overboard in reporting issues, and perhaps i should not have interrupted. we long ago took the ucodes out of boot, because they just take up room in boot, and then in "core", and in the case of the sc40 and the klh10, it is not needed. in the case of the sc40, the front end will load the channels.

…

-jfg

------ Original Message ------ Received: 02:22 PM EDT, 09/30/2023 From: Richard Cornwell ***@***.***> To: rcornwell/sims ***@***.***> Cc: jeffgunter ***@***.***>, Author ***@***.***> Subject: Re: [rcornwell/sims] possible "enhancement" for kl10 NI20 (Issue #310) I assume you will load the microcode, which should be automatic. If you don't load the microcode then you will have empty machine. I don't do a SC40 so no idea about whether microcode was loaded or not. Can we close issues that are resolved? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***> [ { ***@***.***": "http://schema.org", ***@***.***": "EmailMessage", "potentialAction": { ***@***.***": "ViewAction", "target": "#310 (comment)", "url": "#310 (comment)", "name": "View Issue" }, "description": "View this Issue on GitHub", "publisher": { ***@***.***": "Organization", "name": "GitHub", "url": "https://github.com" } } ]

[rcornwell]

I try to emulate closer to hardware. It would not be hard to force a specific version into the NIA20 if you wanted.

Also not overboard, I thank you for finding these errors. Keep going, it is very helpful.