This is a list of MIPS 74K's CPU's performance counter event types. Please see Programming the MIPS32 74K Core Family available from www.mips.com
Name | Description | Counters usable | Unit mask options |
CYCLES | 0-0 Cycles | all | |
INSTRUCTIONS | 1-0 Instructions graduated | all | |
PREDICTED_JR_31 | 2-0 JR $31 (return) instructions predicted including speculative instructions | 0, 2 | |
REDIRECT_STALLS | 3-0 Stall cycles due to register indirect jumps (including non-predicted JR $31), ERET/WAIT instructions, and IFU determined exception | 0, 2 | |
ITLB_ACCESSES | 4-0 Instruction micro-TLB accesses | 0, 2 | |
ICACHE_ACCESSES | 6-0 Instruction cache accesses including speculative instructions | 0, 2 | |
ICACHE_MISS_STALLS | 7-0 Instruction cache miss stall cycles | 0, 2 | |
UNCACHED_IFETCH_STALLS | 8-0 Uncached instruction fetch stall cycles | 0, 2 | |
IFU_REPLAYS | 9-0 Replays within the IFU due to full Instruction Buffer | 0, 2 | |
IFU_IDU_MISS_PRED_UPSTREAM_CYCLES | 11-0 Cycles IFU-IDU gate is closed (to prevent upstream from getting ahead) due to mispredicted branch | 0, 2 | |
IFU_IDU_CLOGED_DOWNSTREAM_CYCLES | 12-0 Cycles IFU-IDU gate is closed (waiting for downstream to unclog) due to MTC0/MFC0 sequence in pipe, EHB, or blocked DD, DR, or DS | 0, 2 | |
DDQ0_FULL_DR_STALLS | 13-0 DR stage stall cycles due to DDQ0 (ALU out-of-order dispatch queue) full | 0, 2 | |
ALCB_FULL_DR_STALLS | 14-0 DR stage stall cycles due to ALCB (ALU completion buffers) full | 0, 2 | |
CLDQ_FULL_DR_STALLS | 15-0 DR stage stall cycles due to CLDQ (data comming back from FPU) full | 0, 2 | |
ALU_EMPTY_CYCLES | 16-0 DDQ0 (ALU out-of-order dispatch queue) empty cycles | 0, 2 | |
ALU_OPERANDS_NOT_READY_CYCLES | 17-0 DDQ0 (ALU out-of-order dispatch queue) no issue cycles with valid instructions but operands not ready | 0, 2 | |
ALU_NO_ISSUES_CYCLES | 18-0 DDQ0 (ALU out-of-order dispatch queue) no issue cycles with valid instructions due to operand(s) not available, MDU busy, or CorExt resource busy | 0, 2 | |
ALU_BUBBLE_CYCLES | 19-0 DDQ0 (ALU out-of-order dispatch queue) bubbles due to MFC1 data write | 0, 2 | |
SINGLE_ISSUE_CYCLES | 20-0 Either DDQ0 (ALU out-of-order dispatch queue) or DDQ1 (AGEN out-of-order dispatch queue) valid instruction issue cycles | 0, 2 | |
OOO_ALU_ISSUE_CYCLES | 21-0 Out-of-order ALU issue cycles (issued instruction is not the oldest in the pool) | 0, 2 | |
JALR_JALR_HB_INSNS | 22-0 Graduated JALR/JALR.HB instructions | 0, 2 | |
DCACHE_LOAD_ACCESSES | 23-0 Counts all accesses to the data cache caused by load instructions | 0, 2 | |
DCACHE_WRITEBACKS | 24-0 Data cache writebacks | 0, 2 | |
JTLB_DATA_ACCESSES | 25-0 Joint TLB data (non-instruction) accesses | 0, 2 | |
LOAD_STORE_REPLAYS | 26-0 Load/store generated replays - load/store follows too closely a matching CACHEOP | 0, 2 | |
LOAD_STORE_BLOCKED_CYCLES | 27-0 Load/store graduation blocked cycles due to CP1/2 store data not ready, SYNC/SYNCI/SC/CACHEOP at the head, or FSB/LDQ/WBB/ITU FIFO full | 0, 2 | |
L2_CACHE_WRITEBACKS | 28-0 L2 Cache Writebacks | 0, 2 | |
L2_CACHE_MISSES | 29-0 L2 Cache Misses | 0, 2 | |
FSB_FULL_STALLS | 30-0 Pipe stall cycles due to FSB full | 0, 2 | |
LDQ_FULL_STALLS | 31-0 Pipe stall cycles due to LDQ full | 0, 2 | |
WBB_FULL_STALLS | 32-0 Pipe stall cycles due to WBB full | 0, 2 | |
LOAD_MISS_CONSUMER_REPLAYS | 35-0 Replays following optimistic issue of instruction dependent on load which missed, counted only when the dependent instruction graduates | 0, 2 | |
JR_NON_31_INSNS | 36-0 jr $xx (not $31) instructions graduated (at same cost as a mispredict) | 0, 2 | |
BRANCH_INSNS | 37-0 Branch instructions graduated, excluding CP1/CP2 conditional branches | 0, 2 | |
BRANCH_LIKELY_INSNS | 38-0 Branch likely instructions graduated including CP1 and CP2 branch likely instructions | 0, 2 | |
COND_BRANCH_INSNS | 39-0 Conditional branches graduated | 0, 2 | |
INTEGER_INSNS | 40-0 Integer instructions graduated including NOP, SSNOP, MOVCI, and EHB | 0, 2 | |
LOAD_INSNS | 41-0 Loads graduated including CP1 ans CP2 loads | 0, 2 | |
J_JAL_INSNS | 42-0 J/JAL graduated | 0, 2 | |
NOP_INSNS | 43-0 NOP instructions graduated - SLL 0, NOP, SSNOP, and EHB | 0, 2 | |
DSP_INSNS | 44-0 DSP instructions graduated | 0, 2 | |
DSP_BRANCH_INSNS | 45-0 DSP branch instructions graduated | 0, 2 | |
UNCACHED_LOAD_INSNS | 46-0 Uncached loads graduated | 0, 2 | |
EJTAG_INSN_TRIGGERS | 49-0 EJTAG instruction triggerpoints | 0, 2 | |
CP1_BRANCH_MISPREDICTIONS | 50-0 CP1 branches mispredicted | 0, 2 | |
SC_INSNS | 51-0 SC instructions graduated | 0, 2 | |
PREFETCH_INSNS | 52-0 Prefetch instructions graduated | 0, 2 | |
NO_INSN_CYCLES | 53-0 No instructions graduated cycles | 0, 2 | |
ONE_INSN_CYCLES | 54-0 One instruction graduated cycles | 0, 2 | |
GFIFO_BLOCKED_CYCLES | 55-0 GFIFO blocked cycles | 0, 2 | |
MISPREDICTION_STALLS | 56-0 Cycles from the time of a pipe kill due to mispredict until the first new instruction graduates | 0, 2 | |
MISPREDICTED_BRANCH_INSNS_CYCLES | 57-0 Mispredicted branch instruction graduation cycles without the delay slot | 0, 2 | |
EXCEPTIONS_TAKEN | 58-0 Exceptions taken | 0, 2 | |
COREEXTEND_EVENTS | 59-0 Implementation specific CorExtend events | 0, 2 | |
ISPRAM_EVENTS | 62-0 Implementation specific ISPRAM events | 0, 2 | |
L2_CACHE_SINGLE_BIT_ERRORS | 63-0 Single bit errors corrected in L2 | 0, 2 | |
SYSTEM_EVENT_0 | 64-0 Implementation specific system event 0 | 0, 2 | |
SYSTEM_EVENT_2 | 65-0 Implementation specific system event 2 | 0, 2 | |
SYSTEM_EVENT_4 | 66-0 Implementation specific system event 4 | 0, 2 | |
SYSTEM_EVENT_6 | 67-0 Implementation specific system event 6 | 0, 2 | |
OCP_ALL_REQUESTS | 68-0 All OCP requests accepted | 0, 2 | |
OCP_READ_REQUESTS | 69-0 OCP read requests accepted | 0, 2 | |
OCP_WRITE_REQUESTS | 70-0 OCP write requests accepted | 0, 2 | |
FSB_LESS_25_FULL | 74-0 FSB < 25% full | 0, 2 | |
LDQ_LESS_25_FULL | 75-0 LDQ < 25% full | 0, 2 | |
WBB_LESS_25_FULL | 76-0 WBB < 25% full | 0, 2 | |
JR_31_MISPREDICTIONS | 2-1 JR $31 (return) instructions mispredicted | 1, 3 | |
JR_31_NO_PREDICTIONS | 3-1 JR $31 (return) instructions not predicted | 1, 3 | |
ITLB_MISSES | 4-1 Instruction micro-TLB misses | 1, 3 | |
JTLB_INSN_MISSES | 5-1 Joint TLB instruction misses | 1, 3 | |
ICACHE_MISSES | 6-1 Instruction cache misses, includes misses from fetch-ahead and speculation | 1, 3 | |
PDTRACE_BACK_STALLS | 8-1 PDtrace back stalls | 1, 3 | |
KILLED_FETCH_SLOTS | 9-1 Valid fetch slots killed due to taken branches/jumps or stalling instructions | 1, 3 | |
IFU_IDU_NO_FETCH_CYCLES | 11-1 Cycles IFU-IDU gate open but no instructions fetched by IFU | 1, 3 | |
DDQ1_FULL_DR_STALLS | 13-1 DR stage stall cycles due to DDQ1 (AGEN out-of-order dispatch queue) full | 1, 3 | |
AGCB_FULL_DR_STALLS | 14-1 DR stage stall cycles due to AGCB (AGEN completion buffers) full | 1, 3 | |
IODQ_FULL_DR_STALLS | 15-1 DR stage stall cycles due to IODQ (data comming back from IO) full | 1, 3 | |
AGEN_EMPTY_CYCLES | 16-1 DDQ1 (AGEN out-of-order dispatch queue) empty cycles | 1, 3 | |
AGEN_OPERANDS_NOT_READY_CYCLES | 17-1 DDQ1 (AGEN out-of-order dispatch queue) no issue cycles with valid instructions but operands not ready | 1, 3 | |
AGEN_NO_ISSUES_CYCLES | 18-1 DDQ1 (AGEN out-of-order dispatch queue) no issue cycles with valid instructions due to operand(s) not available, non-issued stores blocking ready to issue loads, or non-issued CACHEOPs blocking ready to issue loads | 1, 3 | |
AGEN_BUBBLE_CYCLES | 19-1 DDQ1 (AGEN out-of-order dispatch queue) bubbles due to MFC2 data write or cache access from FSB | 1, 3 | |
DUAL_ISSUE_CYCLES | 20-1 Both DDQ0 (ALU out-of-order dispatch queue) and DDQ1 (AGEN out-of-order dispatch queue) valid instruction issue cycles | 1, 3 | |
OOO_AGEN_ISSUE_CYCLES | 21-1 Out-of-order AGEN issue cycles (issued instruction is not the oldest in the pool) | 1, 3 | |
DCACHE_LINE_REFILL_REQUESTS | 22-1 Data cache line loads (line refill requests) | 1, 3 | |
DCACHE_ACCESSES | 23-1 Data cache accesses | 1, 3 | |
DCACHE_MISSES | 24-1 Data cache misses | 1, 3 | |
JTLB_DATA_MISSES | 25-1 Joint TLB data (non-instruction) misses | 1, 3 | |
VA_TRANSALTION_CORNER_CASES | 26-1 Virtual memory address translation synonyms, homonyms, and aliases (loads/stores treated as miss in the cache) | 1, 3 | |
LOAD_STORE_NO_FILL_REQUESTS | 27-1 Load/store graduations not resulting in a bus request because misses at integer pipe graduation turn into hit or merge with outstanding fill request | 1, 3 | |
L2_CACHE_ACCESSES | 28-1 Accesses to the L2 cache | 1, 3 | |
L2_CACHE_MISS_CYCLES | 29-1 Cycles a L2 miss is outstanding, but not necessarily stalling the pipeline | 1, 3 | |
FSB_OVER_50_FULL | 30-1 FSB > 50% full | 1, 3 | |
LDQ_OVER_50_FULL | 31-1 LDQ > 50% full | 1, 3 | |
WBB_OVER_50_FULL | 32-1 WBB > 50% full | 1, 3 | |
CP1_CP2_LOAD_INSNS | 35-1 CP1/CP2 load instructions graduated | 1, 3 | |
MISPREDICTED_JR_31_INSNS | 36-1 jr $31 instructions graduated after mispredict | 1, 3 | |
CP1_CP2_COND_BRANCH_INSNS | 37-1 CP1/CP2 conditional branch instructions graduated | 1, 3 | |
MISPREDICTED_BRANCH_LIKELY_INSNS | 38-1 Mispredicted branch likely instructions graduated | 1, 3 | |
MISPREDICTED_BRANCH_INSNS | 39-1 Mispredicted branches graduated | 1, 3 | |
FPU_INSNS | 40-1 FPU instructions graduated | 1, 3 | |
STORE_INSNS | 41-1 Store instructions graduated including CP1 ans CP2 stores | 1, 3 | |
MIPS16_INSNS | 42-1 MIPS16 instructions graduated | 1, 3 | |
NT_MUL_DIV_INSNS | 43-1 Integer multiply/divide instructions graduated | 1, 3 | |
ALU_DSP_SATURATION_INSNS | 44-1 ALU-DSP graduated, result was saturated | 1, 3 | |
MDU_DSP_SATURATION_INSNS | 45-1 MDU-DSP graduated, result was saturated | 1, 3 | |
UNCACHED_STORE_INSNS | 46-1 Uncached stores graduated | 1, 3 | |
FAILED_SC_INSNS | 51-1 SC instructions failed | 1, 3 | |
CACHE_HIT_PREFETCH_INSNS | 52-1 PREFETCH instructions which did nothing, because they hit in the cache | 1, 3 | |
LOAD_MISS_INSNS | 53-1 Cacheable load instructions that miss in the cache graduated | 1, 3 | |
TWO_INSNS_CYCLES | 54-1 Two instructions graduated cycles | 1, 3 | |
CP1_CP2_STORE_INSNS | 55-1 CP1/CP2 Store graduated | 1, 3 | |
GRADUATION_REPLAYS | 58-1 Replays initiated from graduation | 1, 3 | |
DSPRAM_EVENTS | 62-1 Implementation specific events from the DSPRAM block | 1, 3 | |
SYSTEM_EVENT_1 | 64-1 Implementation specific system event 1 | 0, 2 | |
SYSTEM_EVENT_3 | 65-1 Implementation specific system event 3 | 0, 2 | |
SYSTEM_EVENT_5 | 66-1 Implementation specific system event 5 | 0, 2 | |
SYSTEM_EVENT_7 | 67-1 Implementation specific system event 7 | 0, 2 | |
OCP_ALL_CACHEABLE_REQUESTS | 68-1 All OCP cacheable requests accepted | 0, 2 | |
OCP_READ_CACHEABLE_REQUESTS | 69-1 OCP cacheable read request accepted | 0, 2 | |
OCP_WRITE_CACHEABLE_REQUESTS | 70-1 OCP cacheable write request accepted | 0, 2 | |
FSB_25_50_FULL | 74-1 FSB 25-50% full | 0, 2 | |
LDQ_25_50_FULL | 75-1 LDQ 25-50% full | 0, 2 | |
WBB_25_50_FULL | 76-1 WBB 25-50% full | 0, 2 |
It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories instead of theories to suit facts.- Sherlock Holmes