I've done some tests on the pata_hard_disk.jal library.
I've set up TMR0 to give an interrupt each second, output the value of
bytes_read, and reset bytes_read to 0. Then some code like this was executed:
For word reads:
pata_hd_start_read(0)
forever loop
pata_hd_read_data (in_b, in_a)
bytes_read=bytes_read+2
end loop
For sector reads:
-- pata_hd_current_address=0
pata_hd_start_read(0)
forever loop
pata_hd_read_sector()
bytes_read=bytes_read+512
end loop
In pata_hd_read_sector() and pata_hd_read_sector2() I've inserted the still
missing "asm nop" discussed in another thread. The tests were run on a 18f4520
using INTOSC_PLL at 32MHz.
RESULTS:
pata_hd_read_data() yields approx. 317300 bytes per second. Code usage:3256,
RAM usage: 129
pata_hd_read_sector() yields approx. 167500 bytes per second. Code usage:3500,
RAM usage: 645
pata_hd_read_sector2() yields approx. 151000 bytes per second. Code usage:3772,
RAM usage: 659
------------------------------------------------------
After removing large_array:
pata_hd_read_sector() yields approx. 581000 bytes per second. Code usage:3380,
RAM usage: 638
Ok, the gain is not 400% but "just" 340%, but I think it's worth thinking about
not using large_array as the sector buffer.
Here's the modified procedure: (removed some comments to make it readable)
---------------------------------------------------------------------------
-- read one entire sector
---------------------------------------------------------------------------
var dword pata_hd_last_sector
;const dword LARGE_ARRAY_3_SIZE = 512 -- choose number of array
;const dword LARGE_ARRAY_3_VARIABLE_SIZE = 1 -- choose size of variables
;include large_array_3 -- include the array
;alias pata_hd_sector_buffer is large_array_3 -- rename/alias the array
var byte pata_hd_sector_buffer_lo[256] -- 256 low bytes
var byte pata_hd_sector_buffer_hi[256] -- 256 high bytes
var byte pata_hd_sector_count2 = 0
--
procedure pata_hd_read_sector() is
var word count1
pata_hd_word_count = 0
if pata_hd_sector_count == 0 then
pata_hd_go_to_sector(pata_hd_sector_select,0)
pata_hd_register_write(PATA_HD_COMMAND_REG,PATA_HD_READ_SECTORS)
end if
pata_hd_data_request(PATA_HD_WAIT_READ)
for 256 using count1 loop -- NOTE: now "using count1"
-- get the data trough the ide interface
pata_hd_iord = high ^ PATA_HD_NO_INVERTER -- start read pulse
asm nop
pata_hd_sector_buffer_hi[count1] = pata_hd_data_high -- get high data
pata_hd_sector_buffer_lo[count1] = pata_hd_data_low -- get low data
pata_hd_iord = low ^ PATA_HD_NO_INVERTER -- end read pulse
end loop
pata_hd_sector_select = pata_hd_sector_select + 1
pata_hd_sector_count = pata_hd_sector_count + 1
end procedure
Of course changing the data format of pata_hd_sector_buffer requires
modification of everything that uses it. It is however not used heavily in
jallib yet. So it would be worth thinking about this change as soon as
possible, once some libraries rely on the format of pata_hd_sector_buffer, it
will be harder to change it.
Greets,
Kiste
--
You received this message because you are subscribed to the Google Groups
"jallib" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/jallib?hl=en.
