Commit | Line | Data |
---|---|---|
86530b38 AT |
1 | !----------------------------------------------------------------------- |
2 | ! Run 32 threads on four cores with DMA on thread 5. | |
3 | !----------------------------------------------------------------------- | |
4 | TSOTOOL.PROCESSOR niagara2.rtl | |
5 | TSOTOOL.MODE GEN | |
6 | TSOTOOL.N_THREADS 32 | |
7 | TSOTOOL.TEST_NAME fc4_prop_diag | |
8 | TSOTOOL.BATCH Y | |
9 | ! | |
10 | ! It appears that DMA generating threads are about 18% efficient. | |
11 | ! In other words, if all threads try to execute the same number | |
12 | ! of 'instructions' the non-DMA threads will finish long before | |
13 | ! the DMA threads. The sollowing weighting seems to work for an | |
14 | ! FC1 model with one thread doing DMA. | |
15 | ! | |
16 | ! 0 0 1 2 3 4 5 6 7 1 0 1 2 3 4 5 6 7 2 0 1 2 3 4 5 6 7 3 0 1 2 3 4 5 6 7 | |
17 | GEN.N_INSTR_PER_THREAD 100, 100, 100, 100, 100, 19, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100 | |
18 | GEN.AVG_LOOP_SIZE 0 | |
19 | GEN.AVG_LOOP_ITER 0 | |
20 | ! GEN.SEED 499406026 | |
21 | ADMAP.RTL.REGION_PA_SEPARATION 8M | |
22 | ADMAP.N_REGIONS 4 | |
23 | ADMAP.REGION_SIZE 64K | |
24 | ADMAP.REGION_OFFSETS 916-920-984-1012, 180-1416-2228-3228, 316-692-1268-1520-1560-2404, 2108-2132-2156-2304-2412-2544-2556 | |
25 | ADMAP.ATTRIBUTES CV=1110,CP=1110 | |
26 | ADMAP.NC_MEMMAP 0xc800002000:8G | |
27 | ADMAP.N_ALIASES 0 | |
28 | ADMAP.ALIAS_FREQUENCY 64 | |
29 | ADMAP.ALIAS_OFFSET 8388608 | |
30 | RUN.SIMULATOR no_simulator_defined | |
31 | WT.PCT_FP_INSTR 10 | |
32 | WT.PCT_LITTLE_ENDIAN 5 | |
33 | WT.PCT_LOADS_NF 0 | |
34 | WT.PCT_NFS_FAULT 0 | |
35 | WT.PCT_PREFETCH_FAULT 20 | |
36 | WT.PCT_PREFETCH_UNIMP 20 | |
37 | WT.PCT_CBRANCH 0 | |
38 | WT.PCT_SECONDARY_CTX 0 | |
39 | WT.PCT_NUCLEUS_CTX 0 | |
40 | WT.REPLACEMENT 5 0 cpu=5 | |
41 | WT.INTERRUPT 0 | |
42 | WT.LD 10 0 cpu=5 | |
43 | WT.BLD 0 5 cpu=0-4,6-31 region=0-2 | |
44 | WT.DWLD 0 1 cpu=0-4,6-31 region=0-2 | |
45 | WT.LDD 0 | |
46 | WT.QWLD 0 | |
47 | WT.AQLD 0 1 cpu=0-4,6-31 region=1 | |
48 | WT.ST 5 0 cpu=5 | |
49 | WT.BST 0 10 cpu=0-4,6-31 region=0-2 | |
50 | WT.BSTC 0 | |
51 | WT.ST_BINIT 0 | |
52 | WT.DWST_BINIT 0 | |
53 | WT.DWST 0 10 cpu=0-4,6-31 region=0-2 | |
54 | WT.QWST 0 | |
55 | WT.SWAP 0 5 cpu=0-4,6-31 region=0-2 | |
56 | WT.CAS 0 5 cpu=0-4,6-31 region=0-2 | |
57 | WT.CASX 0 5 cpu=0-4,6-31 region=0-2 | |
58 | WT.ASI_L2_FLUSH 0 | |
59 | WT.FLUSHI 0 | |
60 | WT.MEMBAR 0 | |
61 | WT.PREFETCH 10 0 cpu=5 | |
62 | WT.NOP 0 | |
63 | ADV.L2_WAYS 16 | |
64 | ADV.RESULTS_TO_MEM N | |
65 | ADV.BST_MEMBARS Y | |
66 | ADV.BLD_MEMBARS Y | |
67 | ADV.FP_FLUSH_MEMBARS Y | |
68 | ADV.CAS_IMPLICIT_MEMBARS Y | |
69 | ADV.WARMUP_ITERATIONS 0 | |
70 | ADV.TEST_REPETITIONS 1 | |
71 | ||
72 | ! PEP based DMA operation weights | |
73 | ! The weights here try to skew for mostly small | |
74 | ! and large DMAs, with the emphasis on large and | |
75 | ! skewed towards writes. | |
76 | ||
77 | WT.MACRO.ALM_DMA0_RD 0 10 cpu=5 region=0-2 | |
78 | WT.MACRO.ALM_DMA1_RD 0 10 cpu=5 region=0-2 | |
79 | WT.MACRO.ALM_DMA2_RD 0 10 cpu=5 region=0-2 | |
80 | WT.MACRO.ALM_DMA3_RD 0 10 cpu=5 region=0-2 | |
81 | WT.MACRO.DMA0_RD_0x8 0 15 cpu=5 region=0-2 | |
82 | WT.MACRO.DMA0_RD_0xC 0 4 cpu=5 region=0-2 | |
83 | WT.MACRO.DMA0_RD_0x10 0 4 cpu=5 region=0-2 | |
84 | WT.MACRO.DMA0_RD_0x14 0 4 cpu=5 region=0-2 | |
85 | WT.MACRO.DMA0_RD_0x20 0 4 cpu=5 region=0-2 | |
86 | WT.MACRO.DMA0_RD_0x30 0 4 cpu=5 region=0-2 | |
87 | WT.MACRO.DMA0_RD_0x3C 0 30 cpu=5 region=0-2 | |
88 | WT.MACRO.DMA0_RD_0x40 0 10 cpu=5 region=0-2 | |
89 | WT.MACRO.DMA1_RD_0x40 0 10 cpu=5 region=0-2 | |
90 | WT.MACRO.DMA2_RD_0x40 0 10 cpu=5 region=0-2 | |
91 | WT.MACRO.DMA3_RD_0x40 0 10 cpu=5 region=0-2 | |
92 | ||
93 | WT.MACRO.ALM_DMA0_WR 0 10 cpu=5 region=0-2 | |
94 | WT.MACRO.ALM_DMA1_WR 0 10 cpu=5 region=0-2 | |
95 | WT.MACRO.ALM_DMA2_WR 0 10 cpu=5 region=0-2 | |
96 | WT.MACRO.ALM_DMA3_WR 0 10 cpu=5 region=0-2 | |
97 | WT.MACRO.DMA0_WR_0x8 0 30 cpu=5 region=0-2 | |
98 | WT.MACRO.DMA0_WR_0xC 0 10 cpu=5 region=0-2 | |
99 | WT.MACRO.DMA0_WR_0x10 0 10 cpu=5 region=0-2 | |
100 | WT.MACRO.DMA0_WR_0x14 0 10 cpu=5 region=0-2 | |
101 | WT.MACRO.DMA0_WR_0x20 0 10 cpu=5 region=0-2 | |
102 | WT.MACRO.DMA0_WR_0x30 0 10 cpu=5 region=0-2 | |
103 | WT.MACRO.DMA0_WR_0x3C 0 80 cpu=5 region=0-2 | |
104 | WT.MACRO.DMA0_WR_0x40 0 25 cpu=5 region=0-2 | |
105 | WT.MACRO.DMA1_WR_0x40 0 25 cpu=5 region=0-2 | |
106 | WT.MACRO.DMA2_WR_0x40 0 25 cpu=5 region=0-2 | |
107 | WT.MACRO.DMA3_WR_0x40 0 25 cpu=5 region=0-2 | |
108 | ||
109 | WT.MACRO.ALM_DMA0_INT 0 40 cpu=5 region=0-2 |