aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorH. Peter Anvin <hpa@trantor.hos.anvin.org>2010-09-18 23:22:56 -0700
committerH. Peter Anvin <hpa@trantor.hos.anvin.org>2010-09-18 23:22:56 -0700
commit60ab6cc5203a894544d7ba08e182f987e96ec851 (patch)
treed315a253fe8492b53355511e3b8110d9f85b870e
parentef31d30dad3d8660b746c47132218e6e8d7a1066 (diff)
downloadabc80-60ab6cc5203a894544d7ba08e182f987e96ec851.tar.gz
abc80-60ab6cc5203a894544d7ba08e182f987e96ec851.tar.xz
abc80-60ab6cc5203a894544d7ba08e182f987e96ec851.zip
Document the new FG handling.
-rw-r--r--CHANGES22
1 files changed, 22 insertions, 0 deletions
diff --git a/CHANGES b/CHANGES
index 30fc13f..9f7f8a6 100644
--- a/CHANGES
+++ b/CHANGES
@@ -1,3 +1,25 @@
+Changes in release DE1-18:
+--------------------------
+* Use a 6x10 character matrix as the original ABC80/800 did, instead
+ of a 8x16. To make this compatible with a VGA monitor, we use
+ double scanning (each pixel row drawn as two scanlines.)
+* Add an ABC800C/M-style "fine graphics" (FG) unit. Since there is
+ not enough on-chip memory, we store the FG data in the SRAM, which
+ is multiplexed with the CPU. This works without stalling either
+ unit, since the CPU never needs the SRAM for two consecutive CPU
+ cycles, so we can always get one access every two CPU cycles for the
+ graphics unit, which is more than enough. In fact, we probably
+ could supply quite a bit more bandwidth than this (especially since
+ the SRAM is really 16 bits wide), but this is good enough for now.
+
+ In ABC800, the FGCTL control is OUT 6; since this is the audio
+ generator on ABC80 we use OUT 158 for now. OUT 159 can be used to
+ select which 16K page in the SRAM contains the graphics; this
+ defaults to 4 (= 64K) on reset.
+
+ This is intended to be a first step toward implementing ABC800.
+
+
Changes in release DE1-17:
--------------------------
* Fix handling of "abcprintd -d directory" with a relative pathname