1 files changed, 45 insertions, 0 deletions
diff --git a/asm/i386/00README.TXT b/asm/i386/00README.TXT
new file mode 100644
index 0000000..81b0142
--- /dev/null
+++ b/asm/i386/00README.TXT
@@ -0,0 +1,45 @@
+
+Directory overview:
+===================
+
+As writing portable assembler sources supporting different operating
+systems, compilers and assemblers has proven to be extremely painful,
+the assembler sources have been converted into a more portable 'db'
+format. Use these whenever possible.
+
+  src_gas       sources converted for portable gcc/gas syntax
+  src_masm      sources converted for portable masm/tasm/wasm syntax
+  src_nasm      sources converted for portable nasm syntax
+
+  src           assembler sources (you need the OpenSource nasm assembler)
+
+  obj           pre-assembled object files
+
+Also look 'src_XXX/all/asm_all.asm' which contains all assembler
+functions conveniently arranged into a single file.
+
+
+Notes:
+======
+
+- The assembler sources are designed for a flat 32-bit memory model
+  running in protected mode - they should work with all i386
+  32-bit compilers around.
+
+- All functions expect a 'cdecl' (C stack based) calling convention.
+  The function return value will be placed into 'eax'.
+  All other registers are preserved.
+
+- Prototypes for the assembler functions can be found in <lzo/lzo_asm.h>.
+
+- For reasons of speed all fast assembler decompressors (having '_fast'
+  in their name) can access (write to) up to 3 bytes past the end of
+  the decompressed (output) block. Data past the end of the compressed
+  (input) block is never accessed (read from).
+  [ technical note: because data is transferred in 32-bit units ]
+
+- Finally you should test if the assembler versions are actually faster
+  than the C version on your machine - some compilers can do a very good
+  optimization job, and they also can optimize the code for a specific
+  processor type.
+