diff --git a/mandel.s b/mandel.s
index 526953a..7466cd9 100644
--- a/mandel.s
+++ b/mandel.s
@@ -31,10 +31,10 @@ chroma_offset   = $bb ; u8
 palette_ticks   = $bc ; u8
 chroma_ticks    = $bd ; u8
 count_frames    = $be ; u8
-count_pixels    = $bf ; u8
+; free space $bf
 
-total_pixels    = $c0 ; float48
-total_ms        = $c6 ; float48
+count_iters     = $c0 ; u16
+; free space c2-cb
 temp            = $cc ; u16
 temp2           = $ce ; u16
 
@@ -59,10 +59,12 @@ LBUFF  = $0580 ; result buffer for FASC routine
 ; FP ROM routine vectors
 FASC   = $D8E6 ; FLOATING POINT TO ASCII (output in INBUFF, last char has high bit set)
 IFP    = $D9AA ; INTEGER TO FLOATING POINT CONVERSION (FR0:u16 -> FR0:float48)
+FPI    = $D9D2 ; floating point to integer
 FADD   = $DA66 ; ADDITION       (FR0 += FR1)
 FSUB   = $DA60 ; SUBTRACTION    (FR0 -= FR1)
 FMUL   = $DADB ; MULTIPLICATION (FR0 *= FR1)
 FDIV   = $DB28 ; DIVISION       (FR0 /= FR1)
+ZFR0   = $DA44 ; clear FR0
 ZF1    = $DA46 ; CLEAR ZERO PAGE FLOATING POINT NUMBER (XX)
 FLD0R  = $DD89 ; LOAD FR0 WITH FLOATING POINT NUMBER (YYXX)
 FLD1R  = $DD98 ; LOAD FR1 WITH FLOATING POINT NUMBER (YYXX)
@@ -141,7 +143,7 @@ str_self:
     .byte "MANDEL-6502"
 str_self_end:
 str_speed:
-    .byte " ms/px"
+    .byte "us/iter: "
 str_speed_end:
 str_run:
     .byte " RUN"
@@ -195,14 +197,32 @@ aspect_x: ; fixed3.13 5/4
 aspect_y: ; fixed3.13 3/4
     .word 3 << (13 - 2)
 
-ms_per_frame: ; float48 16.66666667
-    .byte 64  ; exponent/sign
-    .byte $16 ; BCD digits
+sec_per_frame: ; float48 00 . 01 66 66 66 67
+    .byte 63  ; exponent/sign - -1 bytes
+    .byte $01 ; BCD digits
     .byte $66
     .byte $66
     .byte $66
     .byte $67
 
+us_per_sec: ; float48 1e9 01 00 0,0 00 . 00
+    .byte 67  ; exponent/sign +3 bytes
+    .byte $01 ; BCD digits
+    .byte $00
+    .byte $00
+    .byte $00
+    .byte $00
+
+total_iters: ; float48
+    .repeat 6
+        .byte 0
+    .endrepeat
+
+total_sec: ; float48
+    .repeat 6
+        .byte 0
+    .endrepeat
+
 display_list_start:
     ; 24 lines overscan
     .repeat 3
@@ -917,6 +937,11 @@ next:
     sta z_buffer_end
 
 loop:
+    inc count_iters
+    bne low_iters
+    inc count_iters + 1
+low_iters:
+
     ; iter++ & max-iters break
     inc iter
     bne keep_going
@@ -1220,6 +1245,7 @@ done:
     ; clobbers A, X
     .local loop
     .local done
+    .local padding
     ldx #0
 loop:
     cpx #len
@@ -1227,11 +1253,23 @@ loop:
     txa
     tay
     lda (strptr),y
+    bmi padding
     tay
     lda char_map,y
     sta textbuffer + col,x
     inx
     jmp loop
+
+padding:
+    ldy #32 ; space
+    lda char_map,y
+
+    cpx #len
+    beq done
+    sta textbuffer + col,x
+    inx
+    jmp padding
+
 done:
 .endmacro
 
@@ -1559,16 +1597,20 @@ copy_byte_loop:
     jsr SETVBV
 
 main_loop:
-    ; count_frames = 0; count_pixels = 0
+    ; count_frames = 0; count_iters = 0
     lda #0
     sta count_frames
-    sta count_pixels
+    sta count_iters
+    sta count_iters + 1
 
-    ; total_ms = 0.0; total_pixels = 0.0
-    ldx #total_ms
-    jsr ZF1
-    ldx #total_pixels
-    jsr ZF1
+    ; total_sec = 0.0; total_iters = 0.0
+    jsr ZFR0
+    ldx #.lobyte(total_sec)
+    ldy #.hibyte(total_sec)
+    jsr FST0R
+    ldx #.lobyte(total_iters)
+    ldy #.hibyte(total_iters)
+    jsr FST0R
 
     jsr clear_screen
     jsr status_bar
@@ -1640,38 +1682,32 @@ not_skipped_mask:
 
 no_key:
     ; check if we should update the counters
-    ;
-    ; count_pixels >= width? update!
-    inc count_pixels
-    lda count_pixels
-    cmp #width
-    bmi update_status
 
     ; count_frames >= 120? update!
     lda count_frames
     cmp #120 ; >= 2 seconds
-    bmi skipped
+    bpl update_status
+    jmp skipped
 
 update_status:
-    ; FR0 = (float)count_pixels & clear count_pixels
-    lda count_pixels
-    sta FR0
-    lda #0
-    sta FR0 + 1
-    sta count_pixels
+    ; FR0 = (float)count_iters & clear count_iters
+    copy16 FR0, count_iters
     jsr IFP
+    lda #0
+    sta count_iters
+    sta count_iters + 1
 
-    ; FR1 = total_pixels
-    ldx #.lobyte(total_pixels)
-    ldy #.hibyte(total_pixels)
+    ; FR1 = total_iters
+    ldx #.lobyte(total_iters)
+    ldy #.hibyte(total_iters)
     jsr FLD1R
 
     ; FR0 += FR1
     jsr FADD
 
-    ; total_pixels = FR0
-    ldx #.lobyte(total_pixels)
-    ldy #.hibyte(total_pixels)
+    ; total_iters = FR0
+    ldx #.lobyte(total_iters)
+    ldy #.hibyte(total_iters)
     jsr FST0R
 
 
@@ -1684,35 +1720,48 @@ update_status:
     sta count_frames
     jsr IFP
 
-    ; FR0 *= ms_per_frame
-    ldx #.lobyte(ms_per_frame)
-    ldy #.hibyte(ms_per_frame)
+    ; FR0 *= sec_per_frame
+    ldx #.lobyte(sec_per_frame)
+    ldy #.hibyte(sec_per_frame)
     jsr FLD1R
     jsr FMUL
 
-    ; FR0 += total_ms
-    ldx #total_ms
-    ldy #0
+    ; FR0 += total_sec
+    ldx #.lobyte(total_sec)
+    ldy #.hibyte(total_sec)
     jsr FLD1R
     jsr FADD
 
-    ; total_ms = FR0
-    ldx #total_ms
-    ldy #0
+    ; total_sec = FR0
+    ldx #.lobyte(total_sec)
+    ldy #.hibyte(total_sec)
     jsr FST0R
 
-    ; FR0 /= total_pixels
-    ldx #total_pixels
-    ldy #0
+    ; FR0 /= total_iters
+    ldx #.lobyte(total_iters)
+    ldy #.hibyte(total_iters)
     jsr FLD1R
     jsr FDIV
 
+    ; FR0 *= us_per_sec
+    ldx #.lobyte(us_per_sec)
+    ldy #.hibyte(us_per_sec)
+    jsr FLD1R
+    jsr FMUL
+
+    ; @fixme
+    ; round to integer
+    ; for some reason this gives bad results?
+    ;clc
+    ;jsr FPI
+    ;jsr IFP
+
     ; convert to ASCII in INBUFF
     jsr FASC
 
     ; print the first 6 digits
-    draw_text_indirect speed_start, speed_precision, INBUFF
-    draw_text speed_start + speed_precision, str_speed_len, str_speed
+    draw_text speed_start, str_speed_len, str_speed
+    draw_text_indirect speed_start + str_speed_len, speed_precision, INBUFF
 
 skipped: