Skip to content

E
eddr3
Commit 2ac46e21 authored by Andrey Filippov's avatar Andrey Filippov
Browse files
Options

implemented read leveling, eye measurement for DDR3 on random data

parent f6b8d427
Show whitespace changes
Inline Side-by-side
Showing with 555 additions and 85 deletions
.project
View file @ 2ac46e21
......@@ -62,72 +62,72 @@
<link>
<name>vivado_logs/VivadoBitstream.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoBitstream-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoBitstream-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoOpt.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoOpt-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoOpt-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoOptPhys.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoOptPhys-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoOptPhys-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoOptPower.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoOptPower-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoOptPower-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoPlace.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoPlace-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoPlace-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoRoute.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoRoute-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoRoute-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoSynthesis.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoSynthesis-20140611121031786.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoSynthesis-20140611174056482.log</location>
</link>
<link>
<name>vivado_logs/VivadoTimimgSummaryReportImplemented.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoTimimgSummaryReportImplemented-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoTimimgSummaryReportImplemented-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoTimimgSummaryReportSynthesis.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoTimimgSummaryReportSynthesis-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoTimimgSummaryReportSynthesis-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoTimingReportImplemented.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoTimingReportImplemented-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoTimingReportImplemented-20140611174132808.log</location>
</link>
<link>
<name>vivado_logs/VivadoTimingReportSynthesis.log</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoTimingReportSynthesis-20140611121113006.log</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_logs/VivadoTimingReportSynthesis-20140611174132808.log</location>
</link>
<link>
<name>vivado_state/eddr3-opt-phys.dcp</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_state/eddr3-opt-phys-20140611121113006.dcp</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_state/eddr3-opt-phys-20140611174132808.dcp</location>
</link>
<link>
<name>vivado_state/eddr3-place.dcp</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_state/eddr3-place-20140611121113006.dcp</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_state/eddr3-place-20140611174132808.dcp</location>
</link>
<link>
<name>vivado_state/eddr3-route.dcp</name>
<type>1</type>
<location>/data/vdt/vdt-projects/eddr3/vivado_state/eddr3-route-20140611121113006.dcp</location>
<location>/data/vdt/vdt-projects/eddr3/vivado_state/eddr3-route-20140611174132808.dcp</location>
</link>
<link>
<name>vivado_state/eddr3-synth.dcp</name>
......
phy/byte_lane.v
View file @ 2ac46e21
......@@ -20,7 +20,7 @@
*******************************************************************************/
`timescale 1ns/1ps
// minimizing total DQS in delay to match DQ (finedelay stage adds some?)
`define NOFINEDELAY_DQS 1
//`define NOFINEDELAY_DQS 1
module byte_lane #(
parameter IODELAY_GRP ="IODELAY_MEMORY",
parameter IBUF_LOW_PWR ="TRUE",
......
python/ddrtests.py
View file @ 2ac46e21
......@@ -26,6 +26,7 @@ __status__ = "Development"
import mmap
import sys
import struct
import random
DRY_MODE= False # True
MONITOR_EMIO=False #True
def write_mem (addr, data):
......@@ -46,7 +47,7 @@ def write_mem (addr, data):
mm[page_offs:page_offs+4]=packedData
# print ("0x%08x <== 0x%08x (%d)"%(addr,d,d))
mm.close()
if MONITOR_EMIO:
if MONITOR_EMIO and VEBOSE:
gpio0=read_mem (0xe000a068)
gpio1=read_mem (0xe000a06c)
print("GPIO: %04x %04x %04x %04x"%(gpio1>>16, gpio1 & 0xffff, gpio0>>16, gpio0 & 0xffff))
......@@ -172,12 +173,12 @@ if use200Mhz:
DLY_CMDA= [0x3c,0x3c,0x3c,0x3c,0x3b,0x3a,0x39,0x38,0x34,0x34,0x34,0x34,0x33,0x32,0x31,0x30,
0x00,0x2c,0x2c,0x2c,0x2b,0x2a,0x29,0x28,0x24,0x24,0x24,0x24,0x23,0x22,0x21,0x20] # odelay odt, cke, cas, ras, we, ba2,ba1,ba0, X, a14,..,a0
# alternative to set same type delays to the same value
DLY_DQ_IDELAY = 0x60
DLY_DQ_ODELAY = 0x48
DLY_DQS_IDELAY = 0xa0
DLY_DQ_IDELAY = 0x70 # 0x60
DLY_DQ_ODELAY = 0xf0 # 0x48
DLY_DQS_IDELAY = 0x20 # 0xa0
DLY_DQS_ODELAY = 0x4c # b0 for WLV
DLY_DM_ODELAY = 0x48
DLY_CMDA_ODELAY =0x30
DLY_DM_ODELAY = 0x48 # 0x48
DLY_CMDA_ODELAY =0x20 # 0x30
else:
DLY_LANE0_DQS_WLV_IDELAY = 0xe8 # idelay dqs
......@@ -197,10 +198,10 @@ else:
DLY_CMDA_ODELAY =0x50
NUM_FINE_STEPS= 5
#`endif
DLY_PHASE= 0x1c # mmcm fine phase shift, 1/4 tCK
DLY_PHASE= 0x39 # 0x1c # mmcm fine phase shift, 1/4 tCK
DQSTRI_FIRST= 0x3 # DQS tri-state control word, first when enabling output
DQSTRI_LAST= 0xc # DQS tri-state control word, first after disabling output
......@@ -218,7 +219,7 @@ READ_PATTERN_OFFSET=0x40 # read pattern to memory block sequence start address (
WRITE_BLOCK_OFFSET= 0x100 # write block sequence start address (in words) ..0x14c
READ_BLOCK_OFFSET= 0x180 # read block sequence start address (in words)
VERBOSE=True
def check_args(n,command,args):
if len(args) != n:
if n==0:
......@@ -241,48 +242,52 @@ def read_status(): # task read_status;
return axi_read_addr(BASEADDR_STATUS)
def wait_phase_shifter_ready(target_phase): # task wait_phase_shifter_ready;
global STATUS_PSHIFTER_RDY_MASK, PHASE_WIDTH
global STATUS_PSHIFTER_RDY_MASK, PHASE_WIDTH,VERBOSE
if (VERBOSE): print("wait_phase_shifter_ready(0x%x)..."%target_phase,end="")
status = read_status()
while ((status & STATUS_PSHIFTER_RDY_MASK) == 0) or ((status ^ target_phase) & ((1<<PHASE_WIDTH)-1) != 0):
status=read_status()
if (VERBOSE): print("DONE")
def wait_sequencer_ready(): # task wait_sequencer_ready;
global STATUS_SEQ_BUSY_MASK
global STATUS_SEQ_BUSY_MASK,VERBOSE
if (VERBOSE): print("wait_sequencer_ready()...",end="")
# input integer num_skip; #skip this cycles before testing ready (latency from write to busy)
# repeat (num_skip) @(posedge CLK);
status=read_status()
# repeat (8) @(posedge CLK); # latency from read command to registered_rdata. TODO: make it certain (read with the same ID)
while (status & STATUS_SEQ_BUSY_MASK) != 0:
status= read_status()
if (VERBOSE): print("DONE")
def run_sequence (channel,start_addr):
global BASEADDR_RUN_CHN
print("run_sequence(0x%x,0x%x)"%(channel,start_addr))
global BASEADDR_RUN_CHN,VERBOSE
if (VERBOSE): print("run_sequence(0x%x,0x%x)"%(channel,start_addr))
axi_write_single(BASEADDR_RUN_CHN+(channel<<2), start_addr)
def run_mrs(): # task run_mrs;
global INITIALIZE_OFFSET
print("RUN MRS")
global INITIALIZE_OFFSET, VERBOSE
if (VERBOSE): print("RUN MRS")
run_sequence(0,INITIALIZE_OFFSET)
def run_write_lev(): # task run_write_lev;
global WRITELEV_OFFSET
print("RUN WRITE LEVELING")
global WRITELEV_OFFSET, VERBOSE
if (VERBOSE): print("RUN WRITE LEVELING")
run_sequence(0,WRITELEV_OFFSET)
def run_read_pattern(): # task run_read_pattern;
global READ_PATTERN_OFFSET
print("RUN READ PATTERN")
global READ_PATTERN_OFFSET, VERBOSE
if (VERBOSE): print("RUN READ PATTERN")
run_sequence(0,READ_PATTERN_OFFSET)
def run_write_block(): # task run_write_block;
global WRITE_BLOCK_OFFSET
print("RUN WRITE BLOCK")
global WRITE_BLOCK_OFFSET, VERBOSE
if (VERBOSE): print("RUN WRITE BLOCK")
run_sequence(1,WRITE_BLOCK_OFFSET)
def run_read_block(): # task run_read_block;
global READ_BLOCK_OFFSET
print("RUN WRITE BLOCK")
global READ_BLOCK_OFFSET, VERBOSE
if (VERBOSE): print("RUN READ BLOCK")
run_sequence(0,READ_BLOCK_OFFSET)
......@@ -315,36 +320,38 @@ def enable_refresh(en):
axi_write_single(BASEADDR_REFRESH_EN, 0)
def write_block_buf():
global BASEADDR_PORT1_WR
print("WRITE BLOCK DATA")
global BASEADDR_PORT1_WR, VERBOSE
if (VERBOSE): print("WRITE BLOCK DATA")
for i in range(256):
d=i | (((i + 7) & 0xff) << 8) | (((i + 23) & 0xff) << 16) | (((i + 31) & 0xff) << 24)
axi_write_single(BASEADDR_PORT1_WR+(i<<2), d)
# print("Write block data (addr:data): 0x%x:0x%x "%(BASEADDR_PORT1_WR+(i<<2),d))
# if (VERBOSE): print("Write block data (addr:data): 0x%x:0x%x "%(BASEADDR_PORT1_WR+(i<<2),d))
def read_block_buf(
num_read ): #input integer num_read; // number of words to read (will be rounded up to multiple of 16)
global BASEADDR_PORT0_RD
global BASEADDR_PORT0_RD, VERBOSE
buf=[]
for i in range(num_read):
d=axi_read_addr(BASEADDR_PORT0_RD+(i<<2))
buf.append(d)
print("Read block data (addr:data): 0x%x:0x%x "%(BASEADDR_PORT0_RD+(i<<2),d))
if (VERBOSE): print("Read block data (addr:data): 0x%x:0x%x "%(BASEADDR_PORT0_RD+(i<<2),d))
return buf
def read_buf(
num_read ): #input integer num_read; // number of words to read (will be rounded up to multiple of 16)
global BASEADDR_PORT0_RD
global BASEADDR_PORT0_RD, VERBOSE;
buf=[]
for i in range(num_read):
d=axi_read_addr(BASEADDR_PORT0_RD+(i<<2))
buf.append(d)
if (VERBOSE):
for i in range(num_read):
addr= i<<2;
if (i%8) == 0:
print ("\n%08x: "%addr,end="")
print ("%08x "%buf[i],end="")
print ()
return buf
def encode_seq_word(
phy_addr_in, # [14:0] also provides pause length when the command is NOP
......@@ -769,7 +776,8 @@ def set_write_block(
# reg [31:0] data;
# integer i;
global BASEADDR_CMD0, WRITE_BLOCK_OFFSET
global BASEADDR_CMD0, WRITE_BLOCK_OFFSET,VERBOSE
if (VERBOSE): print("set_write_block(0x%x,0x%x,0x%x"%(ba,ra,ca))
cmd_addr = BASEADDR_CMD0 + (WRITE_BLOCK_OFFSET << 2)
# activate
data = (
......@@ -1222,7 +1230,7 @@ def set_refresh( # task set_refresh;
def set_mrs( # task set_mrs; # will also calibrate ZQ
reset_dll, # input reset_dll;
cl): #CAS latency code: 2 - 5, 4 - 6, 6 - 7
global BASEADDR_CMD0
global BASEADDR_CMD0, VERBOSE
# reg [17:0] mr0;
# reg [17:0] mr1;
# reg [17:0] mr2;
......@@ -1256,10 +1264,10 @@ def set_mrs( # task set_mrs; # will also calibrate ZQ
0, # 1'h0, # mpr; # MPR mode: 0 - normal, 1 - dataflow from MPR
0) # 2'h0) # [1:0] mpr_rf; # MPR read function: 2'b00: predefined pattern 0101...
cmd_addr = BASEADDR_CMD0;
print("mr0=0x%x"%mr0)
print("mr1=0x%x"%mr1)
print("mr2=0x%x"%mr2)
print("mr3=0x%x"%mr3)
if (VERBOSE): print("mr0=0x%x"%mr0)
if (VERBOSE): print("mr1=0x%x"%mr1)
if (VERBOSE): print("mr2=0x%x"%mr2)
if (VERBOSE): print("mr3=0x%x"%mr3)
data = encode_seq_word(
mr2 & 0x7fff, # mr2[14:0], # [14:0] phy_addr_in;
(mr2 >> 15) & 0x7, # mr2[17:15], # [ 2:0] phy_bank_in; #TODO: debug!
......@@ -1363,20 +1371,20 @@ def set_mrs( # task set_mrs; # will also calibrate ZQ
cmd_addr = cmd_addr + 4
#set tristate patterns
def axi_set_tristate_patterns(): # task axi_set_tristate_patterns;
global DQSTRI_LAST, DQSTRI_FIRST, DQTRI_LAST, DQTRI_FIRST
global DQSTRI_LAST, DQSTRI_FIRST, DQTRI_LAST, DQTRI_FIRST, VERBOSE
axi_write_single(BASEADDR_PATTERNS_TRI,
((DQSTRI_LAST & 0xff) << 12) |
((DQSTRI_FIRST & 0xff) << 8) |
((DQTRI_LAST & 0xff) << 4) |
((DQTRI_FIRST & 0xff) << 0))
print("SET TRISTATE PATTERNS")
if (VERBOSE): print("SET TRISTATE PATTERNS")
def axi_set_dqs_dqm_patterns(): # task axi_set_dqs_dqm_patterns;
global BASEADDR_PATTERNS
global BASEADDR_PATTERNS, VERBOSE
# set patterns for DM (always 0) and DQS - always the same (may try different for write lev.)
axi_write_single(BASEADDR_PATTERNS, 0x0055)
print("SET DQS+DQM PATTERNS")
if (VERBOSE): print("SET DQS+DQM PATTERNS")
# initialize delays
......@@ -1453,9 +1461,9 @@ def axi_set_dly_single( # task axi_set_dly_single;
def axi_set_dq_idelay(# task axi_set_dq_idelay;
delay): # input [7:0] delay;
global BASEADDRESS_LANE0_IDELAY,BASEADDRESS_LANE1_IDELAY
global BASEADDRESS_LANE0_IDELAY,BASEADDRESS_LANE1_IDELAY, VERBOSE
dly=delay;
print("SET DQ IDELAY=0x%x"%delay)
if (VERBOSE): print("SET DQ IDELAY=0x%x"%delay)
for i in range(8):
axi_write_single(BASEADDRESS_LANE0_IDELAY + (i<<2), dly & 0xff)
for i in range(8):
......@@ -1464,9 +1472,9 @@ def axi_set_dq_idelay(# task axi_set_dq_idelay;
def axi_set_dq_odelay(# task axi_set_dq_odelay;
delay): #input [7:0] delay;
global BASEADDRESS_LANE0_ODELAY,BASEADDRESS_LANE1_ODELAY
global BASEADDRESS_LANE0_ODELAY,BASEADDRESS_LANE1_ODELAY, VERBOSE
dly=delay;
print("SET DQ ODELAY=0x%x"%delay)
if (VERBOSE): print("SET DQ ODELAY=0x%x"%delay)
for i in range(8):
axi_write_single(BASEADDRESS_LANE0_ODELAY + (i<<2), dly & 0xff)
for i in range(8):
......@@ -1474,10 +1482,10 @@ def axi_set_dq_odelay(# task axi_set_dq_odelay;
axi_write_single(BASEADDR_DLY_SET, 0) # set all delays
def axi_set_dqs_idelay(# task axi_set_dqs_idelay;
delay): # input [7:0] delay;
global BASEADDRESS_LANE0_IDELAY,BASEADDRESS_LANE1_IDELAY
global BASEADDRESS_LANE0_IDELAY,BASEADDRESS_LANE1_IDELAY, VERBOSE
dly=delay
i=8
print("SET DQS IDELAY=0x%x"%delay)
if (VERBOSE): print("SET DQS IDELAY=0x%x"%delay)
axi_write_single(BASEADDRESS_LANE0_IDELAY + (i<<2), dly & 0xff)
axi_write_single(BASEADDRESS_LANE1_IDELAY + (i<<2), dly & 0xff)
axi_write_single(BASEADDR_DLY_SET, 0) # set all delays
......@@ -1486,25 +1494,27 @@ def axi_set_dqs_odelay(# task axi_set_dqs_odelay;
delay): # input [7:0] delay;
dly=delay
i=8
global BASEADDRESS_LANE0_ODELAY,BASEADDRESS_LANE1_ODELAY
print("SET DQS ODELAY=0x%x"%delay)
global BASEADDRESS_LANE0_ODELAY,BASEADDRESS_LANE1_ODELAY, VERBOSE
if (VERBOSE): print("SET DQS ODELAY=0x%x"%delay)
axi_write_single(BASEADDRESS_LANE0_ODELAY + (i<<2), dly & 0xff)
axi_write_single(BASEADDRESS_LANE1_ODELAY + (i<<2), dly & 0xff)
axi_write_single(BASEADDR_DLY_SET, 0) # set all delays
def axi_set_dm_odelay(# task axi_set_dm_odelay;
delay): # input [7:0] delay;
global VERBOSE
dly=delay;
i=9;
print("SET DQM IDELAY=0x%x"%delay)
if (VERBOSE): print("SET DQM IDELAY=0x%x"%delay)
axi_write_single(BASEADDRESS_LANE0_ODELAY + (i<<2), dly & 0xff)
axi_write_single(BASEADDRESS_LANE1_ODELAY + (i<<2), dly & 0xff)
axi_write_single(BASEADDR_DLY_SET, 0) # set all delays
def axi_set_cmda_odelay(# task axi_set_cmda_odelay;
delay): # input [7:0] delay;
global VERBOSE
dly=delay;
print("SET COMMAND and ADDRESS ODELAY=0x%x"%delay)
if (VERBOSE): print("SET COMMAND and ADDRESS ODELAY=0x%x"%delay)
for i in range(32):
axi_write_single(BASEADDRESS_CMDA + (i<<2), dly & 0xff)
axi_write_single(BASEADDR_DLY_SET, 0) # set all delays
......@@ -1516,7 +1526,8 @@ def axi_set_same_delays(# task set_same_delays;
dqs_odelay, # input [7:0] dqs_odelay;
dm_odelay, # input [7:0] dm_odelay;
cmda_odelay):# input [7:0] cmda_odelay;
print("SET DELAYS(0x%x,0x%x,0x%x,0x%x,0x%x,0x%x)"%(dq_idelay,dq_odelay,dqs_idelay,dqs_odelay,dm_odelay,cmda_odelay))
global VERBOSE
if (VERBOSE): print("SET DELAYS(0x%x,0x%x,0x%x,0x%x,0x%x,0x%x)"%(dq_idelay,dq_odelay,dqs_idelay,dqs_odelay,dm_odelay,cmda_odelay))
axi_set_dq_idelay(dq_idelay)
axi_set_dq_odelay(dq_odelay)
axi_set_dqs_idelay(dqs_idelay)
......@@ -1526,8 +1537,8 @@ def axi_set_same_delays(# task set_same_delays;
def axi_set_phase( # task axi_set_phase;
phase): # input [PHASE_WIDTH-1:0] phase;
global BASEADDRESS_PHASE, BASEADDR_DLY_SET, PHASE_WIDTH
print("SET CLOCK PHASE to 0x%x"%phase)
global BASEADDRESS_PHASE, BASEADDR_DLY_SET, PHASE_WIDTH, VERBOSE
if (VERBOSE): print("SET CLOCK PHASE to 0x%x"%phase)
axi_write_single(BASEADDRESS_PHASE, phase & ((1<<PHASE_WIDTH)-1))
axi_write_single(BASEADDR_DLY_SET, 0) # set all dealys
......@@ -1535,34 +1546,35 @@ def axi_set_phase( # task axi_set_phase;
def axi_set_wbuf_delay(# task axi_set_wbuf_delay;
delay): #input [3:0] delay;
global BASEADDR_WBUF_DELAY
print("SET WBUF DELAY to 0x%x"%delay);
global BASEADDR_WBUF_DELAY, VERBOSE
if (VERBOSE): print("SET WBUF DELAY to 0x%x"%delay);
axi_write_single(BASEADDR_WBUF_DELAY, delay);
def set_all_sequences(): # task set_all_sequences;
print("SET MRS")
global VERBOSE
if (VERBOSE): print("SET MRS")
set_mrs(1,4) # CL=5 (6: CL=7)
print("SET REFRESH")
if (VERBOSE): print("SET REFRESH")
set_refresh(
50, # input [ 9:0] t_rfc; # =50 for tCK=2.5ns
48) # 16) #input [ 7:0] t_refi; # 48/97 for normal, 8 - for simulation
print("SET WRITE LEVELING")
if (VERBOSE): print("SET WRITE LEVELING")
set_write_lev(16) # write leveling, 16 times (full buffer - 128)
print("SET READ PATTERN")
set_read_pattern(8,0,1) # 8x2*64 bits, 32x32 bits to read (second 0 - pattern type, only 0 defined)
print("SET WRITE BLOCK")
if (VERBOSE): print("SET READ PATTERN")
set_read_pattern(8,0,0) # 8x2*64 bits, 32x32 bits to read (second 0 - pattern type, only 0 defined)
if (VERBOSE): print("SET WRITE BLOCK")
set_write_block(
5, # 3'h5, # bank
0x1234, # 15'h1234, # row address
0x100 # 10'h100 # column address
)
print("SET READ BLOCK")
if (VERBOSE): print("SET READ BLOCK")
set_read_block(
5, # 3'h5, # bank
0x1234, # 15'h1234, # row address
0x100, # 10'h100 # column address
1 # use second clock for read commands
0 # use second clock for read commands
)
def set_up(): # task set_up;
......@@ -1582,6 +1594,445 @@ def set_up(): # task set_up;
axi_set_phase(DLY_PHASE)
axi_set_wbuf_delay(WBUF_DLY_DFLT)
def split_delay(dly):
global NUM_FINE_STEPS
dly_int=dly>>3
dly_fine=dly & 0x7
if dly_fine > (NUM_FINE_STEPS-1):
dly_fine= NUM_FINE_STEPS-1
return dly_int*NUM_FINE_STEPS+dly_fine
def combine_delay(dly):
global NUM_FINE_STEPS
return ((dly/NUM_FINE_STEPS)<<3)+(dly%NUM_FINE_STEPS)
def bad_data(buf):
for w in buf:
if (w!=0xffffffff): return False
return True
def scan_dqs(
low_delay,
high_delay,
num ):
global VERBOSE;
saved_verbose=VERBOSE;
VERBOSE=False;
set_read_pattern(num+1,0,0); # do not use first/last pair of the 32 bit words
low = split_delay(low_delay)
high = split_delay(high_delay)
results = []
for dly in range (low, high+1):
enc_dly=combine_delay(dly)
axi_set_dqs_idelay_individual(enc_dly, enc_dly)
run_read_pattern()
buf=read_buf(4*num+2)
if bad_data(buf):
results.append([])
else:
data=[0]*32 # for each bit - even, then for all - odd
for w in range (4*num):
lane=w%2
for wb in range(32):
g=(wb/8)%2
b=wb%8+lane*8+16*g
if (buf[w+2] & (1<<wb) != 0):
data[b]+=1
results.append(data)
print ("%3d (0x%02x): "%(dly,enc_dly),end="")
for i in range(32):
print("%5x"%data[i],end="")
print()
for index in range (len(results)):
dly=index+low
enc_dly=combine_delay(dly)
if (len (results[index])>0):
print ("%3d (0x%02x): "%(dly,enc_dly),end="")
for i in range(32):
print("%5x"%results[index][i],end="")
print()
# else:
# print ("%3d (0x%02x): *** BAD data *** "%(dly,enc_dly))
print()
print()
print ("Delay",end=" ")
for i in range(16):
print ("Bit%dP"%i,end=" ")
for i in range(16):
print ("Bit%dM"%i,end=" ")
print()
for index in range (len(results)):
dly=index+low
enc_dly=combine_delay(dly)
if (len (results[index])>0):
print ("%d"%(dly),end=" ")
for i in range(32):
print("%d"%results[index][i],end=" ")
print()
# else:
# print ("%3d (0x%02x): *** BAD data *** "%(dly,enc_dly))
print()
VEBOSE=saved_verbose
return results
def scan_dq_idelay(
low_delay,
high_delay,
num ):
global VERBOSE;
saved_verbose=VERBOSE;
VERBOSE=False;
set_read_pattern(num+1,0,0); # do not use first/last pair of the 32 bit words
low = split_delay(low_delay)
high = split_delay(high_delay)
results = []
for dly in range (low, high+1):
enc_dly=combine_delay(dly)
axi_set_dq_idelay(enc_dly)
run_read_pattern()
wait_sequencer_ready()
buf=read_buf(4*num+2)
if bad_data(buf):
results.append([])
else:
data=[0]*32 # for each bit - even, then for all - odd
for w in range (4*num):
lane=w%2
for wb in range(32):
g=(wb/8)%2
b=wb%8+lane*8+16*g
if (buf[w+2] & (1<<wb) != 0):
data[b]+=1
results.append(data)
print ("%3d (0x%02x): "%(dly,enc_dly),end="")
for i in range(32):
print("%5x"%data[i],end="")
print()
for index in range (len(results)):
dly=index+low
enc_dly=combine_delay(dly)
if (len (results[index])>0):
print ("%3d (0x%02x): "%(dly,enc_dly),end="")
for i in range(32):
print("%5x"%results[index][i],end="")
print()
# else:
# print ("%3d (0x%02x): *** BAD data *** "%(dly,enc_dly))
print()
print()
print ("Delay",end=" ")
for i in range(16):
print ("Bit%dP"%i,end=" ")
for i in range(16):
print ("Bit%dM"%i,end=" ")
print()
for index in range (len(results)):
dly=index+low
enc_dly=combine_delay(dly)
if (len (results[index])>0):
print ("%d"%(dly),end=" ")
for i in range(32):
print("%d"%results[index][i],end=" ")
print()
# else:
# print ("%3d (0x%02x): *** BAD data *** "%(dly,enc_dly))
print()
VEBOSE=saved_verbose
return results
def adjust_dq_idelay(
low_delay,
high_delay,
num,
falling ): # 0 - use rising as delay increases, 1 - use falling
global VERBOSE;
saved_verbose=VERBOSE;
VERBOSE=False;
low = split_delay(low_delay)
data_raw=scan_dq_idelay(low_delay,high_delay,num)
data=[]
delays=[]
for i,d in enumerate(data_raw):
if len(d)>0:
data.append(d)
delays.append(i+low)
print(delays)
best_dlys=[0]*16
best_diffs=[num*8.0]*16
for i in range (1,len(data)-1):
for j in range (16):
delta=abs(data[i][j] -data[i][j+16] + 0.5*(data[i-1][j] -data[i-1][j+16]+data[i+1][j] -data[i+1][j+16]))
sign=(data[i-1][j] -data[i-1][j+16]-data[i+1][j]+data[i+1][j+16])
if falling > 0: sign=-sign;
if (sign>0) and (delta < best_diffs[j]):
best_diffs[j]=delta
best_dlys[j]=delays[i]
for i in range (16):
print("%2d: %3d (0x%02x)"%(i,best_dlys[i],combine_delay(best_dlys[i])))
VEBOSE=saved_verbose
for i in range (8):
axi_set_dly_single(1,i,combine_delay(best_dlys[i]))
for i in range (8):
axi_set_dly_single(3,i,combine_delay(best_dlys[i+8]))
# VEBOSE=saved_verbose
def convert_mem16_to_w32(mem16):
res32=[]
for i in range(0,len(mem16),4):
res32.append(((mem16[i+3] & 0xff) << 24) |
((mem16[i+2] & 0xff) << 16) |
((mem16[i+1] & 0xff) << 8) |
((mem16[i+0] & 0xff) << 0))
res32.append((((mem16[i+3]>>8) & 0xff) << 24) |
(((mem16[i+2]>>8) & 0xff) << 16) |
(((mem16[i+1]>>8) & 0xff) << 8) |
(((mem16[i+0]>>8) & 0xff) << 0))
return res32
def convert_w32_to_mem16(w32):
mem16=[]
for i in range(0,len(w32),2):
mem16.append(((w32[i]>> 0) & 0xff) | (((w32[i+1] >> 0) & 0xff) << 8))
mem16.append(((w32[i]>> 8) & 0xff) | (((w32[i+1] >> 8) & 0xff) << 8))
mem16.append(((w32[i]>>16) & 0xff) | (((w32[i+1] >> 16) & 0xff) << 8))
mem16.append(((w32[i]>>24) & 0xff) | (((w32[i+1] >> 24) & 0xff) << 8))
return mem16
# calibratin finedelay dealy steps using everaged "eye" data, assuming that most error
# is in finedelay stage
def calibrate_finedelay(
low, # absolute delay value of start scan
avg_types, # weights of weach of the 8 bit sequences
res_avg, # averaged eye data tablle, each line has 8 elements, or [] for bad measurements
ends_dist, # do not process if one of the primary interval ends is within this from 0.0 or 1.0
min_diff): # minimal difference between primary delay steps to process
global NUM_FINE_STEPS
start_index=0;
weights=[0.0]*( NUM_FINE_STEPS-1)
corr=[0.0]*( NUM_FINE_STEPS-1)
if (low % NUM_FINE_STEPS) != 0:
start_index=NUM_FINE_STEPS-(low % NUM_FINE_STEPS)
for index in range(start_index, len(res_avg)-NUM_FINE_STEPS,NUM_FINE_STEPS):
if (len(res_avg[index])>0) and (len(res_avg[index+NUM_FINE_STEPS])>0):
for t,w in enumerate(avg_types):
if (w>0):
f=res_avg[index][t];
s=res_avg[index+NUM_FINE_STEPS][t];
# print ("index=%d t=%d f=%f s=%s"%(index,t,f,s))
if ((f>ends_dist) and (s>ends_dist) and
(f< (1-ends_dist)) and (s < (1-ends_dist)) and
(abs(s-f)>min_diff)):
diff=s-f
wd=w* diff*diff # squared? or use abs?
for j in range (1,NUM_FINE_STEPS):
if ( (len(res_avg[index+j])>0)):
v=res_avg[index+j][t];
#correction to the initila step==1
d=(v-f)/(s-f)*NUM_FINE_STEPS-j
#average
corr[j-1]+=wd*d
weights[j-1]+=wd
# print ("\n weights:")
# print(weights)
# print ("\n corr:")
# print(corr)
for i,w in enumerate(weights):
if (w>0) : corr[i]/=w
print ("\ncorr:")
print("%f"%0.0)
# print(corr)
for c in corr:
print ("%f"%c)
return corr
def scan_dq_idelay_random(
low_delay,
high_delay,
use_dq, # 0 - scan dqs, 1 - scan dq (common valuwe, post-adjustment)
ends_dist, # do not process if one of the primary interval ends is within this from 0.0 or 1.0
min_diff): # minimal difference between primary delay steps to process
global BASEADDR_PORT1_WR,VERBOSE;
saved_verbose=VERBOSE;
VERBOSE=False;
# set_read_pattern(num+1,0,0); # do not use first/last pair of the 32 bit words
low = split_delay(low_delay)
high = split_delay(high_delay)
rand16=[]
for i in range(512):
rand16.append(random.randint(0,65535))
wdata=convert_mem16_to_w32(rand16)
print("rand16:")
for i in range(len(rand16)):
if (i & 0x1f) == 0:
print("\n%03x:"%i,end=" ")
print("%04x"%rand16[i],end=" ")
print("\n")
print("wdata:")
for i in range(len(wdata)):
if (i & 0xf) == 0:
print("\n%03x:"%i,end=" ")
print("%08x"%wdata[i],end=" ")
print("\n")
bit_type=[] # does not include first and last elements
for i in range(1,511):
types=[]
for j in range(16):
types.append((((rand16[i-1]>>j) & 1)<<2) | (((rand16[i ]>>j) & 1)<<1) | (((rand16[i+1]>>j) & 1)))
bit_type.append(types)
# print ("i=%d",i)
# print(types)
# total_types=[[0]*8]*16 # number of times each type occured in the block for each DQ bit (separate for DG up/down?)
total_types=[] # number of times each type occured in the block for each DQ bit (separate for DG up/down?)
for i in range(16): total_types.append([0]*8)
for type in bit_type:
# print(type)
for j in range(16):
# total_types[j][type[j]]+=1
total_types[j][type[j]]=total_types[j][type[j]]+1
print("\ntotal_types:")
print (total_types)
avg_types=[0.0]*8
N=0
for t in total_types:
for j,n in enumerate(t):
avg_types[j]+=n
N+=n
for i in range(len(avg_types)):
avg_types[i]/=N
print("\avg_types:")
print (avg_types)
#write blobk buffer with 256x32bit data
for i in range(256):
axi_write_single(BASEADDR_PORT1_WR+(i<<2), wdata[i])
set_write_block(
5, # 3'h5, # bank
0x1234, # 15'h1234, # row address
0x100 # 10'h100 # column address
)
run_write_block()
wait_sequencer_ready()
#now scanning - first DQS, then try with DQ (post-adjustment - best fit)
results = []
for dly in range (low, high+1):
enc_dly=combine_delay(dly)
if (use_dq!=0):
axi_set_dq_idelay(enc_dly)
else:
axi_set_dqs_idelay_individual(enc_dly, enc_dly)
run_read_block()
wait_sequencer_ready()
buf32=read_buf(256)
if bad_data(buf32):
results.append([])
else:
read16=convert_w32_to_mem16(buf32) # 512x16 bit, same as DDR3 DQ over time
if VERBOSE and (dly==low):
print("buf32:")
for i in range(len(buf32)):
if (i & 0xf) == 0:
print("\n%03x:"%i,end=" ")
print("%08x"%buf32[i],end=" ")
print("\n")
print("read16:")
for i in range(len(read16)):
if (i & 0x1f) == 0:
print("\n%03x:"%i,end=" ")
print("%04x"%read16[i],end=" ")
print("\n")
# exit (0)
# data=[[0]*8]*16 # for each bit - 8 types
data=[] # number of times each type occured in the block for each DQ bit (separate for DG up/down?)
for i in range(16):
data.append([0]*8)
for i in range (1,511):
w= read16[i]
type=bit_type[i-1] # first and last words are not used, no type was calculated
for j in range(16):
if (w & (1<<j)) !=0:
data[j][type[j]]+=1
for i in range(16):
for t in range(8):
if (total_types[i][t] >0 ):
data[i][t]*=1.0/total_types[i][t]
results.append(data)
print ("%3d (0x%02x): "%(dly,enc_dly),end="")
for i in range(16):
print("[",end="")
for j in range(8):
print("%3d"%(round(100.0*data[i][j])),end=" ")
print("]",end=" ")
print()
titles=["'000","'001","'010", "'011","'100","'101","'110","'111"]
print ("delay",end=" ")
for t in range(8):
for i in range(16):
print("%02d:%s"%(i,titles[t]),end=" ")
print()
for dly in range (len(results)):
if (len(results[dly])>0):
print ("%d"%(dly+low),end=" ")
for t in range(8):
for i in range(16):
print("%.4f"%(results[dly][i][t]),end=" ")
print()
#calculate weighted averages
#TODO: for DQ scan shift individula bits for the best match
if use_dq:
print("TODO: shift individual bits for the best match before averaging")
res_avg=[]
for dly in range (len(results)):
if (len(results[dly])>0):
data=results[dly]
avg=[0.0]*8
for t in range(8):
weight=0;
d=0.0
for i in range(16):
weight+=total_types[i][t]
d+=total_types[i][t]*data[i][t]
if (weight>0):
d/=weight
avg[t] = d
res_avg.append(avg)
else:
res_avg.append([])
print ("delay",end=" ")
for t in range(8):
print(titles[t],end=" ")
print()
for dly in range (len(res_avg)):
if (len(res_avg[dly])>0):
print ("%d"%(dly+low),end=" ")
for t in range(8):
print("%.4f"%(res_avg[dly][t]),end=" ")
print()
corr_fine=calibrate_finedelay(
low, # absolute delay value of start scan
avg_types, # weights of weach of the 8 bit sequences
res_avg, # averaged eye data tablle, each line has 8 elements, or [] for bad measurements
ends_dist/256.0, # ends_dist, # do not process if one of the primary interval ends is within this from 0.0 or 1.0
min_diff/256.0) #min_diff): # minimal difference between primary delay steps to process
VEBOSE=saved_verbose
# main
if len(sys.argv)<2:
print ("Usage: %s command [hex_parameter, ...]"%sys.argv[0])
......@@ -1776,6 +2227,25 @@ elif command=="set_up":
set_up()
print("set_up() OK")
#
elif command=="scan_dqs":
check_args(3,command,args)
scan_dqs(args[0],args[1],args[2])
print("scan_dqs(0x%x,0x%x,0x%x) OK"%(args[0],args[1],args[2]))
elif command=="scan_dq_idelay":
check_args(3,command,args)
scan_dq_idelay(args[0],args[1],args[2])
print("scan_dq_idelay(0x%x,0x%x,0x%x) OK"%(args[0],args[1],args[2]))
elif command=="scan_dq_idelay_random":
check_args(5,command,args)
scan_dq_idelay_random(args[0],args[1],args[2],args[3],args[4])
print("scan_dq_idelay_random(0x%x,0x%x,0x%x,0x%x,0x%x) OK"%(args[0],args[1],args[2],args[3],args[4]))
elif command=="adjust_dq_idelay":
check_args(4,command,args)
adjust_dq_idelay(args[0],args[1],args[2],args[3])
print("adjust_dq_idelay(0x%x,0x%x,0x%x,0x%x) OK"%(args[0],args[1],args[2],args[3]))
else:
print("Invalid command: %s"%command)
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment