|Re: [hatari-devel] DSP Bootstrap ROM|
[ Thread Index |
| More lists.tuxfamily.org/hatari-devel Archives
- To: hatari-devel@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [hatari-devel] DSP Bootstrap ROM
- From: Andreas Grabher <andreas_g86@xxxxxxxxxx>
- Date: Thu, 23 Sep 2021 13:18:31 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=icloud.com; s=1a1hai; t=1632395917; bh=fHDfcdPVsiu5nATD50paMQoyvYNPt4OUgEOcJU0AgDo=; h=From:Content-Type:Mime-Version:Date:Subject:Message-Id:To; b=P2uHItUcp+cQubSgR+AmL7atXUMIggoVbf+9sDiWqdtw8OsbP6/B5u0oorq5V/MYf J9PDBMyfeHj241FDwhXLy5fbKgeGIKr5YtsxyoW8ddaqkuYcGWlNRVurXo/mAiROY4 YOKX9byd5JgYH1VclIF2ShRYelBy9g7z/aZJWIEkuhn/h0B3jkmfkYtUasEptbuGmX b68M1YexO23RxxJiB7j9wiGzU7znFXpruHcXeKjqHCj1P831igqT2S/yxtAqTD4cGl fxf2kmL/7+YCnII70hOW3EutflTB7ACJ7V2MEsEhKuedGq+tluJOgdTiB3AELpwyqP x6ABvnDLTj1EA==
> Am 23.09.2021 um 09:36 schrieb Nicolas Pomarède <npomarede@xxxxxxxxxxxx>:
> Le 23/09/2021 à 06:49, Andreas Grabher a écrit :
>>> Am 22.09.2021 um 22:11 schrieb Eero Tamminen <oak@xxxxxxxxxxxxxx>:
>>> Few comments...
>>> * Code calls Statusbar_SetDspLed(), but that's missing
>> Sorry, that is an oversight. It is specific to Previous and indicates that the DSP is active.
>>> * Add defines for magic (mode / mode_wait) constants
>> I do not understand. Mode is a number between 0 and 3 (you find it in the datasheet of the 56001) and mode_wait is for the timed bootstrap ROM enable/disable.
>>> Why BOOTSTRAP_ROM ifdef is needed i.e. what downside it has?
>> At least with Previous the extension causes some loss of efficiency because the waiting loop (waiting for program to be supplied by the host) is permanently running.
> I'm not really familiar with the DSP, but doesn't it already run in a waiting loop anyway in the case of Hatari if no specific program is supplied ?
> Also, does the bootstrap mode exist on real falcon/dsp HW ? If so, I think it should be added to Hatari for better accuracy.
The actual loop is some kind of hack, that partially simulates what the bootstrap program does. While in that loop the DSP does not run. That saves lots of host CPU cycles.
I don‘t know how the DSP is connected on the Falcon. On the NeXT it is possible to select DSP mode by software via system control register.
Furthermore it would be possible to set the DSP to any mode by writing to the OMR. But I did not yet see a program doing that.