diff --git a/boot/floppy.asm b/boot/floppy.asm
index 8c03b57..a0993f5 100644
--- a/boot/floppy.asm
+++ b/boot/floppy.asm
@@ -1,4 +1,4 @@
-org 0x0600
+org 0x0000
 
 fdc:		; FDC Descriptor as per ECMA-107
   jmp _startup
@@ -9,19 +9,19 @@ fdc:		; FDC Descriptor as per ECMA-107
 .ss:
   dw 0x200	; sector size
 .sc:
-  db 0		; sectors per cluster
+  db 2		; sectors per cluster
 .rsc:
-  dw 2		; reserved sector count
+  dw 1		; reserved sector count
 .fn:
-  db 0		; number of file allocation tables
+  db 2		; number of file allocation tables
 .rde:
-  dw 0		; number of root directory entries
+  dw 0x70	; number of root directory entries
 .ts:
   dw 720	; total number of sectors
 .mi:		; medium identifier
   db 0xFD	; 5.25-inch Double sided, 40 tracks per side, 9 sectors per track (360 KB)
 .sf:		; sectors per fat
-  dw 0
+  dw 2
 .spt:
   dw 9		; sectors per track
 .nos:
@@ -44,33 +44,65 @@ fdc:		; FDC Descriptor as per ECMA-107
   times (62 - ($-$$)) db " "
 
 _startup:
-  xor ax, ax
-  mov ds, ax
-  mov es, ax
+  mov ax, 0x1050
+  ; setup relocated stack area
   mov ss, ax
   xor sp, sp
-  ; relocate
+  ; backup important values
+  push bx
+  push cx
+  push dx
+  ; setup relocated code area
+  mov es, ax
+  mov di, 0x0000
+  ; setup start code area
+  xor ax, ax
+  mov ds, ax
   mov si, 0x7C00
-  mov di, 0x0600
+  ; relocate 256 words
   mov cx, 0x0100
   rep movsw
   ; adjust CS
-  jmp 0x0000:main
+  jmp 0x1050:main
+
+; non-fdc variables
+rootdir:
+.start:
+  dw 0
+.length:
+  dw 0
 
 main:
-  mov [fdc.drv], dl ; backup drive number
+  ; setup buffer area directly after the BDA
+  mov ax, 0x0050
+  mov ds, ax
+  mov es, ax
+
+  mov [cs:fdc.drv], dl ; backup drive number
 
   call fix_chs
-  call find_first_sector
+  call calc_sectors
+  call find_first_cluster
+  xor bx, bx
+  call load_cluster
 
-  call error
-  db "END OF IMPLEMENTATION", 0
+  ; restore important variables
+  xor ax, ax
+  mov bp, ax
+  mov si, ax
+  mov di, ax
+  pop dx
+  pop cx
+  pop bx
+  jmp 0x0050:0000
 
 ; Write AX, a string and bail out
 ; Return pointer is the string start
 ; String must be 0 suffixed
 error:
   call print16
+  push cs
+  pop ds
   pop si
   mov bx, 0x0000
   mov al, 0x20
@@ -90,41 +122,49 @@ fix_chs:
   int 0x13
   jc .end ; skip if function does not exist
   inc dh
-  mov [fdc.nos], dh
+  mov [cs:fdc.nos], dh
   mov ax, cx
   and ax, 0x003F
-  mov [fdc.spt], ax ; no adjustment because sectors are 1-indexed
+  mov [cs:fdc.spt], ax ; no adjustment because sectors are 1-indexed
   mov ax, cx
   xchg al, ah
   mov cl,6
   shr ah,cl
   inc ax ; convert from maximum number (0-based) to total number (1-based) of cylinders
-  mul word [fdc.nos] ; number of tracks = number of cylinders * heads
-  mul word [fdc.spt] ; number of sectors = number of tracks * sectors per track
-  mov [fdc.ts], ax
+  mul word [cs:fdc.nos] ; number of tracks = number of cylinders * heads
+  mul word [cs:fdc.spt] ; number of sectors = number of tracks * sectors per track
+  mov [cs:fdc.ts], ax
 .end:
   ret
 
-; Load the root directory [buffer]
-find_first_sector:
-  mov bx, [fdc.ss]      ; bytes per sector
+calc_sectors:
+  mov bx, [cs:fdc.ss]  ; bytes per sector
   mov cl, 5
-  shr bx, cl            ; div by 2^5, 32 bytes per directory entity
-  mov ax, [fdc.rde]     ; number of root directory entities
+  shr bx, cl           ; div by 2^5, 32 bytes per entry
+  mov ax, [cs:fdc.rde] ; number of entries
   xor dx, dx
   div bx
-  mov cx, ax            ; cx = number of rootdir sectors
-  mov ax, [fdc.sf]      ; sectors per fat
-  mul byte [fdc.fn]     ; number of fats
-  add ax, [fdc.rsc]     ; ax = starting sector of rootdir
-  mov bx, buffer
+  mov [cs:rootdir.length], ax
+  mov ax, [cs:fdc.sf]  ; sectors per fat
+  mul byte [cs:fdc.fn] ; number of fats
+  add ax, [cs:fdc.rsc] ; reserved sectors
+  mov [cs:rootdir.start], ax
+  ret
+
+; Load the root directory
+find_first_cluster:
+  mov ax, [cs:rootdir.start]
+  mov cx, [cs:rootdir.length]
+  xor bx, bx
 .blkloop:
   call loadblk
-  add bx, buffer
+  add bx, [cs:fdc.ss]
   add ax, 1
   loop .blkloop
-  mov cx, [fdc.rde]
-  mov bx, buffer
+  push cs
+  pop es
+  mov cx, [cs:fdc.rde]
+  xor bx, bx
 .dirloop:
   push cx
   mov cx, 0x000B
@@ -138,38 +178,49 @@ find_first_sector:
   call error
   db "NOT FOUND", 0
 .found:
+  push ds
+  pop es
   add bx, 26
   mov ax, [bx]
   ret
 
 ; Load a single FAT cluster into memory
 ; in    ax      cluster number
-;       bx      buffer
+;       es:bx   buffer
 ;
-loadsec:
-  call print16
-  mov ax, bx
-  call print16
+load_cluster:
+  xor ch, ch
+  mov cl, [cs:fdc.sc] ; sectors per cluster
+  xor dx, dx
+  dec ax ; somehow 2 is first cluster
+  div word cx
+  add ax, [cs:rootdir.start]
+  add ax, [cs:rootdir.length]
+.loop:
+  call loadblk
+  inc ax
+  add bx, [cs:fdc.ss]
+  loop .loop
   ret
 
 ; Load a single block into memory
 ; Does not return on error
 ; in    ax      sector number
-;       bx	buffer
+;       es:bx	buffer
 loadblk:
   push ax
   push cx
   push dx
   xor dx, dx
-  div word [fdc.spt]    ; ax:temp     = (lba / spt)
+  div word [cs:fdc.spt]    ; ax:temp     = (lba / spt)
   inc dx                ; dx:sector   = (lba % spt) + 1
   mov cl, dl ; sector number
   xor dx, dx
-  div word [fdc.nos]    ; ax:cylinder = (tmp / heads)
+  div word [cs:fdc.nos]    ; ax:cylinder = (tmp / heads)
                         ; dx:head     = (tmp % heads)
   mov ch, al ; cylinder number
   mov dh, dl ; head number
-  mov dl, [fdc.drv] ; driver number
+  mov dl, [cs:fdc.drv] ; driver number
   mov ax, 0x0201 ; ah=0x02 al=0x01
   int 0x13
   jc .error
@@ -194,5 +245,3 @@ bootfilename:
 times (0x1FE - ($-$$)) db 0
 
 dw 0xAA55
-
-buffer: