ggSSKrSSKJrJr \"SS5\"SS5\"SS5\"S S 5\"S S 5/rS rSrSrSrSr Sr Sr g)N) MetricRequestRequestedMetricsParserz'smsp__issue_active.avg.per_cycle_active issue_active(smsp__maximum_warps_avg_per_active_cycletheoretical_warpsz'smsp__warps_active.avg.per_cycle_active active_warpsz)smsp__warps_eligible.avg.per_cycle_activeeligible_warpsz%smsp__warps_active.avg.peak_sustained max_warpscg)NIssueSlotUtilizationr 8nsight-compute-2025.1.1/sections/IssueSlotUtilization.pyget_identifierr%s !rcg)NzIssue Slot Utilizationr r rrget_namer(s #rcg)Nz$Scheduler instruction issue analysisr r rrget_descriptionr+s 1rcg)NSchedulerStatsr r rrget_section_identifierr.s rcS/$)NComputer r rrget_parent_rules_identifiersr1s ;rcUSR5nSU- nSnX@;aSX- n[X55n[RRnUS-nXg4$)Nrmax_throughput_normalizedd)valueminNvRules IFrontendSpeedupType_LOCAL)parent_weightsmetricsrimprovement_localthroughput_name upper_bound speedup_typeimprovement_percents rget_estimated_speedupr+5si>*002LL(1O(.99  1?$$66L+c1  ,,rc[R"U5nURS5RS5nUR 5n[ X5R [5nURS5nUSR5nUSR5nUSR5nUSR5n USR5n Sn Xk:GaS RS U- 5n U S R[U 5U5- n US :aU S - n O1U S RU 5- n X- S:a U S- n U S- n O U S- n U S- n UR[RRU 5n [XT5upUR!XU5 0UESUS- 0EnUR#XSR%5U[RR&S5 UR)U5 g)Nrrrrrr r g333333?zEvery scheduler is capable of issuing one instruction per cycle, but for this workload each scheduler only issues an instruction every {:.1f} cycles. This might leave hardware resources underutilized and may lead to less optimal performance.g?zw Out of the maximum of {} warps per scheduler, this workload allocates an average of {:.2f} active warps per scheduler,zH which already limits the scheduler to less than a warp per instruction.a but only an average of {:.2f} warps were eligible per cycle. Eligible warps are the subset of active warps that are ready to issue their next instruction. Every cycle with no eligible warp results in no instruction being issued and the issue slot remains unused.g?z To increase the number of eligible warps, reduce the time the active warps are stalled by inspecting the top stall reasons on the ze@section:WarpStateStats:Warp State Statistics@ and @section:SourceCounters:Source Counters@ sections.z To increase the number of eligible warps, avoid possible load imbalances due to highly different execution durations per warp.z Reducing stalls indicated on the @section:WarpStateStats:Warp State Statistics@ and @section:SourceCounters:Source Counters@ sections can help, too."issue_slot_util_speedup_normalizedrz*002L 34::<>*002L-.446N $**,I'FMMNPQ]N]^MTTUXYbUceqr r #  a aG ahhiwx xG/#5aaCC]]ssG--FFP&;N&T#  67  0-#2E   7 < < > gN_N_NyNy|z {^,r) r!RequestedMetricsrrr3rrrrrr+rEr rrrGss2B;^L<>QR;^L=?OP9;G "$2 -,-r