# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1586989491 -7200
# Node ID 62b4bf8a2511da8f2938769d4cb1d22d65d6e91f
# Parent  4f335343eec7ff0cbfd468df183e0b56c5e31b57
Added benchmark observations and a table of standard bandwidth characteristics.

diff -r 4f335343eec7 -r 62b4bf8a2511 ULA.txt
--- a/ULA.txt	Wed Apr 15 22:31:08 2020 +0200
+++ b/ULA.txt	Thu Apr 16 00:24:51 2020 +0200
@@ -251,6 +251,27 @@
 2MHz RAM Access enhancement below for bandwidth calculations that consider
 this limitation removed.
 
+A summary of the bandwidth figures is as follows (with extra timing details
+described below):
+
+                Standard ULA    % Total   Slowdown  BBC-10s BBC-34s
+MODE 0, 1, 2    9728 bytes      24%       4.11      43s     105s
+MODE 3          12288 bytes     31%       3.25      34s
+MODE 4, 5       19968 bytes     50%       2         20s
+MODE 6          19968 bytes     50%       2         20s     50s
+
+The review of the Electron in Practical Computing (October 1983) provides a
+concise overview of the RAM access limitations and gives timing comparisons
+between modes and BBC Micro performance. In the above, "BBC-10s" is the
+measured or stated time given for a program taking 10 seconds on the BBC
+Micro, whereas "BBC-34s" is the apparently measured time given for the
+"Persian" program taking 34 seconds to complete on the BBC Micro, with a
+"quick" mode presumably switching to MODE 6 using the ULA directly in order to
+reduce display bandwidth usage while the program draws to the screen.
+Evidently, the measured slowdown is slightly lower than the theoretical
+slowdown, most likely due to the running time not being entirely dominated by
+RAM access performance characteristics.
+
 Video Timing
 ------------