CX16: diskio.f_write() now uses fast MCIOUT block writes, including hiram bank boundary wrap-over

compiler/res/prog8lib/cx16/diskio.p8

@@ -312,13 +312,12 @@ close_end:
         reset_read_channel()
         list_blocks = 0     ; we reuse this variable for the total number of bytes read
 
-        ; commander X16 supports fast block-read via MACPTR() kernal call
         uword readsize
         while num_bytes {
             readsize = 255
             if num_bytes<readsize
                 readsize = num_bytes
-            readsize = cx16.MACPTR(lsb(readsize), bufferpointer, false)
+            readsize = cx16.MACPTR(lsb(readsize), bufferpointer, false)     ; fast block reads
             if_cs
                 goto byte_read_loop     ; MACPTR block read not supported, do fallback loop
             list_blocks += readsize
@@ -433,6 +432,21 @@ _end        rts
         ; -- write the given number of bytes to the currently open file
         if num_bytes!=0 {
             reset_write_channel()
+            do {
+                cx16.r0 = cx16.MCIOUT(lsb(num_bytes), bufferpointer, false)     ; fast block writes
+                if_cs
+                    goto no_mciout
+                num_bytes -= cx16.r0
+                bufferpointer += cx16.r0
+                if msb(bufferpointer) == $c0
+                    bufferpointer = mkword($a0, lsb(bufferpointer))  ; wrap over bank boundary
+                if cbm.READST()!=0
+                    return false
+            } until num_bytes==0
+            return not cbm.READST()
+
+no_mciout:
+            ; the device doesn't support MCIOUT, use a normal per-byte write loop
             repeat num_bytes {
                 cbm.CHROUT(@(bufferpointer))
                 bufferpointer++

docs/source/index.rst

@@ -21,7 +21,7 @@ while still being low level enough to create high performance programs.
 You can compile programs for various machines with this CPU:
 
 * Commodore 64
-* Commander X16 (release R42 or newer is required)
+* Commander X16
 * Commodore 128 (limited support for now)
 * Atari 800 XL (limited support for now)
 
@@ -181,7 +181,7 @@ For MacOS you can use the Homebrew system to install a recent version of OpenJDK
 Finally: an **emulator** (or a real machine of course) to test and run your programs on.
 In C64 mode, the compiler assumes the presence of the `VICE emulator <http://vice-emu.sourceforge.net/>`_.
 If you're targeting the Commander X16 instead,
-download a recent emulator version (R42 or newer) for the CommanderX16, such as `x16emu <https://cx16forum.com/forum/viewforum.php?f=30>`_
+download a recent emulator version for the CommanderX16, such as `x16emu <https://cx16forum.com/forum/viewforum.php?f=30>`_
 (preferred, this is the official emulator. If required, source code is `here <https://github.com/X16Community/x16-emulator/>`_.
 There is also `Box16 <https://github.com/indigodarkwolf/box16>`_ which has powerful debugging features.
 You can select which one you want to launch using the ``-emu`` or ``-emu2`` command line options.

examples/cx16/diskspeed.p8

@@ -6,14 +6,13 @@
 main {
 
     ubyte[256] buffer = 0 to 255
-    const ubyte REPEATS = 2
 
     sub print_speed(uword jiffies) {
         if jiffies==0 {
             txt.print("\n 0 jiffies measured, speed is extremely high\n")
             return
         }
-        float speed = 65536.0 * REPEATS / (jiffies as float / 60.0)
+        float speed = floats.floor(65536.0 / (jiffies as float / 60.0))
         txt.nl()
         txt.print_uw(jiffies)
         txt.print(" jiffies = ")
@@ -22,95 +21,56 @@ main {
     }
 
     sub start() {
-        txt.print("\n\ndisk benchmark. drive 8. repeats = ")
-        txt.print_ub(REPEATS)
+        txt.print("\n\ndisk benchmark on drive 8.\n\n")
 
         uword batchtotaltime
 
-        txt.print("\n\nwriting 64kb using save")
-        batchtotaltime = 0
-        repeat REPEATS {
-            cbm.SETTIM(0,0,0)
-            void diskio.save("@:benchmark.dat", $100, 32768)
-            void diskio.save("@:benchmark.dat", $100, 32768)
-            batchtotaltime += cbm.RDTIM16()
-            txt.chrout('.')
-        }
-        print_speed(batchtotaltime)
+        txt.print("writing 64kb using save()")
+        cbm.SETTIM(0,0,0)
+        ; save 2 times 32Kb to make it 64Kb total
+        void diskio.save("@:benchmark.dat", $100, 32768)
+        void diskio.save("@:benchmark.dat", $100, 32768)
+        print_speed(cbm.RDTIM16())
 
         txt.print("\nwriting 64kb sequentially")
-        batchtotaltime = 0
-        repeat REPEATS {
-            if diskio.f_open_w("@:benchmark.dat") {
-                cbm.SETTIM(0,0,0)
-                repeat 65536/256 {
-                    if not diskio.f_write(buffer, 256)
-                        sys.exit(1)
-                }
-                batchtotaltime += cbm.RDTIM16()
-                diskio.f_close_w()
+        if diskio.f_open_w("@:benchmark.dat") {
+            cbm.SETTIM(0,0,0)
+            repeat 65536/256 {
+                if not diskio.f_write(buffer, 256)
+                    sys.exit(1)
             }
-            txt.chrout('.')
+            diskio.f_close_w()
+            print_speed(cbm.RDTIM16())
         }
-        print_speed(batchtotaltime)
 
-        txt.print("\nreading 64kb using load into hiram")
-        batchtotaltime = 0
-        repeat REPEATS {
-            cbm.SETTIM(0,0,0)
-            cx16.rambank(4)
-            if not diskio.load("benchmark.dat", $a000)
-                sys.exit(1)
-            batchtotaltime += cbm.RDTIM16()
-            txt.chrout('.')
-        }
-        print_speed(batchtotaltime)
+        txt.print("\nreading 64kb using load() into hiram")
+        cbm.SETTIM(0,0,0)
+        cx16.rambank(4)
+        if not diskio.load("benchmark.dat", $a000)
+            sys.exit(1)
+        print_speed(cbm.RDTIM16())
 
-        txt.print("\nreading 64kb using vload into videoram")
-        batchtotaltime = 0
-        repeat REPEATS {
-            cbm.SETTIM(0,0,0)
-            if not diskio.vload("benchmark.dat", 0, $0000)
-                sys.exit(1)
-            batchtotaltime += cbm.RDTIM16()
-            txt.chrout('.')
-        }
-        print_speed(batchtotaltime)
+        txt.print("\nreading 64kb using vload() into vram")
+        cbm.SETTIM(0,0,0)
+        if not diskio.vload("benchmark.dat", 0, $0000)
+            sys.exit(1)
+        print_speed(cbm.RDTIM16())
 
         txt.print("\nreading 64kb sequentially")
-        batchtotaltime = 0
-        repeat REPEATS {
-            if diskio.f_open("benchmark.dat") {
-                cbm.SETTIM(0,0,0)
-                repeat 65536/255 {
-                    if not diskio.f_read(buffer, 255)
-                        sys.exit(1)
-                }
-                batchtotaltime += cbm.RDTIM16()
-                diskio.f_close()
+        if diskio.f_open("benchmark.dat") {
+            cbm.SETTIM(0,0,0)
+            repeat 65536/255 {
+                if not diskio.f_read(buffer, 255)
+                    sys.exit(1)
             }
-            txt.chrout('.')
+            diskio.f_close()
+            print_speed(cbm.RDTIM16())
         }
-        print_speed(batchtotaltime)
-
-        txt.print("\nreading 64kb sequentially (x16 optimized)")
-        batchtotaltime = 0
-        repeat REPEATS {
-            if diskio.f_open("benchmark.dat") {
-                cbm.SETTIM(0,0,0)
-                repeat 65536/255 {
-                    if not diskio.f_read(buffer, 255)
-                        sys.exit(1)
-                }
-                batchtotaltime += cbm.RDTIM16()
-                diskio.f_close()
-            }
-            txt.chrout('.')
-        }
-        print_speed(batchtotaltime)
 
         txt.nl()
         txt.print(diskio.status())
         txt.print("\ndone.\n")
+
+        diskio.delete("benchmark.dat")
     }
 }