Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/KuangjuX/ncu-cli
/ functions
Functions
172 in github.com/KuangjuX/ncu-cli
⨍
Functions
172
◇
Types & classes
24
↓ 45 callers
Function
get_f64
Extract a numeric value, stripping sampling-count suffixes like ` {929}`.
src/parser.rs:130
↓ 23 callers
Function
cli_binary
()
tests/integration_test.rs:3
↓ 9 callers
Method
analyze
(&self, data: &KernelData)
src/analyzer/memory.rs:20
↓ 9 callers
Function
make_data
(sectors: f64, requests: f64, l1_hit: f64, l2_hit: f64)
src/analyzer/memory.rs:171
↓ 8 callers
Method
analyze
(&self, data: &KernelData)
src/analyzer/instruction.rs:20
↓ 8 callers
Function
make_data
(name: &str, hmma_pct: f64)
src/analyzer/instruction.rs:123
↓ 7 callers
Function
load_kernels
(path: &std::path::Path)
src/main.rs:171
↓ 7 callers
Function
make_data
(sm: f64, mem: f64)
src/analyzer/roofline.rs:191
↓ 7 callers
Function
open_output
(path: Option<&std::path::Path>)
src/main.rs:160
↓ 5 callers
Function
classify
(data: &KernelData)
src/analyzer/roofline.rs:26
↓ 4 callers
Function
all_analyzers
Return all registered analyzers (skills).
src/analyzer/mod.rs:20
↓ 4 callers
Method
analyze
(&self, data: &KernelData)
src/analyzer/occupancy.rs:20
↓ 4 callers
Method
analyze
(&self, data: &KernelData)
src/analyzer/warp_stall.rs:24
↓ 4 callers
Function
arch_display_name
(sm: u32)
src/analyzer/arch.rs:25
↓ 4 callers
Method
as_str
(&self)
src/severity.rs:26
↓ 4 callers
Function
get_str
(map: &HashMap<String, String>, key: &str)
src/parser.rs:125
↓ 4 callers
Function
make_data
(local_sectors: f64, warps_pct: f64)
src/analyzer/occupancy.rs:107
↓ 4 callers
Function
make_stall_data
(long_sb: f64, short_sb: f64, wait: f64, eligible: f64)
src/analyzer/warp_stall.rs:130
↓ 3 callers
Method
analyze
(&self, data: &KernelData)
src/analyzer/launch_config.rs:20
↓ 3 callers
Method
analyze
(&self, data: &KernelData)
src/analyzer/roofline.rs:61
↓ 3 callers
Method
category
(&self)
src/analyzer/arch.rs:43
↓ 3 callers
Function
make_launch_data
( regs: f64, shmem_kb: f64, limit_regs: f64, limit_shmem: f64, limit_w
src/analyzer/launch_config.rs:165
↓ 3 callers
Method
name
(&self)
src/analyzer/arch.rs:35
↓ 2 callers
Method
analyze
(&self, data: &KernelData)
src/analyzer/arch.rs:47
↓ 2 callers
Function
arch_family
(sm: u32)
src/analyzer/arch.rs:16
↓ 2 callers
Function
format_analyze
( w: &mut dyn Write, data: &KernelData, findings: &[Finding], format: OutputFormat, )
src/formatter.rs:35
↓ 2 callers
Method
occupancy_limiter
The minimum occupancy limiter (blocks per SM) and its name.
src/metrics.rs:243
↓ 2 callers
Function
run_command
(cmd: Commands)
src/main.rs:139
↓ 2 callers
Function
select_kernel
( kernels: &'a [metrics::KernelData], filter: Option<&str>, )
src/main.rs:176
↓ 2 callers
Method
stall_breakdown
Returns stall reasons sorted by count descending as (name, count, pct).
src/metrics.rs:209
↓ 2 callers
Function
write_report
(w: &mut dyn Write, data: &KernelData, findings: &[Finding])
src/report.rs:17
↓ 1 callers
Function
analyze_ampere
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/arch.rs:81
↓ 1 callers
Function
analyze_blackwell
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/arch.rs:125
↓ 1 callers
Function
analyze_hopper
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/arch.rs:103
↓ 1 callers
Function
analyze_occupancy_gap
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/launch_config.rs:93
↓ 1 callers
Function
analyze_occupancy_limiter
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/launch_config.rs:31
↓ 1 callers
Function
analyze_register_pressure
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/launch_config.rs:129
↓ 1 callers
Function
build_kernel_data
(map: &HashMap<String, String>)
src/parser.rs:58
↓ 1 callers
Function
check_bank_conflicts
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/memory.rs:108
↓ 1 callers
Function
check_dram_bandwidth
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/memory.rs:133
↓ 1 callers
Function
check_instruction_mix
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/instruction.rs:65
↓ 1 callers
Function
check_l1_hit_rate
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/memory.rs:74
↓ 1 callers
Function
check_l2_hit_rate
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/memory.rs:91
↓ 1 callers
Function
check_load_coalescing
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/memory.rs:34
↓ 1 callers
Function
check_store_coalescing
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/memory.rs:54
↓ 1 callers
Function
check_tensor_core_utilization
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/instruction.rs:31
↓ 1 callers
Function
check_thread_divergence
(data: &KernelData, findings: &mut Vec<Finding>)
src/analyzer/instruction.rs:91
↓ 1 callers
Function
classify_memory_sublevel
(data: &KernelData)
src/analyzer/roofline.rs:165
↓ 1 callers
Function
cmd_analyze
( path: &std::path::Path, kernel: Option<&str>, format: OutputFormat, output: Option<&std::pat
src/main.rs:196
↓ 1 callers
Function
cmd_diff
( before_path: &std::path::Path, after_path: &std::path::Path, format: OutputFormat, output: O
src/main.rs:273
↓ 1 callers
Function
cmd_export
( path: &std::path::Path, format: OutputFormat, output: Option<&std::path::Path>, )
src/main.rs:288
↓ 1 callers
Function
cmd_info
( path: &std::path::Path, format: OutputFormat, output: Option<&std::path::Path>, )
src/main.rs:211
↓ 1 callers
Function
cmd_skill
(action: SkillAction)
src/main.rs:235
↓ 1 callers
Function
cmd_summary
( path: &std::path::Path, format: OutputFormat, output: Option<&std::path::Path>, )
src/main.rs:223
↓ 1 callers
Method
description
(&self)
src/analyzer/arch.rs:39
↓ 1 callers
Function
diff_profiles
Compare two sets of kernel profiles and produce a diff report. Kernels are matched by name. When a kernel appears in both profiles, duration deltas a
src/diff.rs:30
↓ 1 callers
Method
divergence_pct
Thread divergence ratio: 0.0 = no divergence, higher = more divergence.
src/metrics.rs:235
↓ 1 callers
Function
format_diff
( w: &mut dyn Write, report: &DiffReport, format: OutputFormat, )
src/formatter.rs:341
↓ 1 callers
Function
format_info
( w: &mut dyn Write, kernels: &[KernelData], format: OutputFormat, )
src/formatter.rs:142
↓ 1 callers
Function
format_skill_list
( w: &mut dyn Write, format: OutputFormat, )
src/formatter.rs:529
↓ 1 callers
Function
format_summary
( w: &mut dyn Write, kernels: &[KernelData], format: OutputFormat, )
src/formatter.rs:234
↓ 1 callers
Function
get_analyzer
Look up an analyzer by name (case-insensitive prefix match).
src/analyzer/mod.rs:33
↓ 1 callers
Function
parse_arch_sm
NCU `device__attribute_architecture` encodes SM version as `version * 10 * 4`. e.g. 384 -> 384/4 = 96 -> SM 9.0 (Hopper), 800 -> SM 8.0 (Ampere). We r
src/parser.rs:148
↓ 1 callers
Function
parse_ncu_csv
Parse an NCU CSV export into one or more `KernelData` entries. NCU CSV is a vertical key-value format (`metric_name,value`). When multiple kernels ar
src/parser.rs:13
↓ 1 callers
Function
run_all
Run all analyzers in sequence and collect findings.
src/analyzer/mod.rs:41
↓ 1 callers
Function
run_analyzers
Run a specific set of analyzers and collect findings.
src/analyzer/mod.rs:46
↓ 1 callers
Function
stall_action
(reason: &str, pct: f64)
src/analyzer/warp_stall.rs:68
↓ 1 callers
Function
strip_sample_suffix
`"5280946840 {929}"` -> `"5280946840"`
src/parser.rs:138
↓ 1 callers
Method
total_stall_samples
Total warp stall samples across all stall reasons.
src/metrics.rs:194
↓ 1 callers
Function
write_analyze_csv
( w: &mut dyn Write, data: &KernelData, findings: &[Finding], )
src/formatter.rs:103
↓ 1 callers
Function
write_analyze_markdown
( w: &mut dyn Write, data: &KernelData, findings: &[Finding], )
src/formatter.rs:57
↓ 1 callers
Function
write_diff_csv
(w: &mut dyn Write, report: &DiffReport)
src/formatter.rs:500
↓ 1 callers
Function
write_diff_markdown
(w: &mut dyn Write, report: &DiffReport)
src/formatter.rs:445
↓ 1 callers
Function
write_diff_terminal
(w: &mut dyn Write, report: &DiffReport)
src/formatter.rs:357
↓ 1 callers
Function
write_findings
(w: &mut dyn Write, findings: &[Finding])
src/report.rs:258
↓ 1 callers
Function
write_header
(w: &mut dyn Write, data: &KernelData)
src/report.rs:30
↓ 1 callers
Function
write_info_csv
(w: &mut dyn Write, kernels: &[KernelData])
src/formatter.rs:211
↓ 1 callers
Function
write_info_markdown
(w: &mut dyn Write, kernels: &[KernelData])
src/formatter.rs:193
↓ 1 callers
Function
write_info_terminal
(w: &mut dyn Write, kernels: &[KernelData])
src/formatter.rs:168
↓ 1 callers
Function
write_metrics_table
(w: &mut dyn Write, data: &KernelData)
src/report.rs:54
↓ 1 callers
Function
write_optimization_priorities
(w: &mut dyn Write, findings: &[Finding])
src/report.rs:216
↓ 1 callers
Function
write_stall_summary
(w: &mut dyn Write, data: &KernelData)
src/report.rs:180
↓ 1 callers
Function
write_summary_csv
(w: &mut dyn Write, entries: &[SummaryEntry])
src/formatter.rs:322
↓ 1 callers
Function
write_summary_markdown
(w: &mut dyn Write, entries: &[SummaryEntry])
src/formatter.rs:306
↓ 1 callers
Function
write_summary_terminal
(w: &mut dyn Write, entries: &[SummaryEntry])
src/formatter.rs:264
Method
category
(&self)
src/analyzer/launch_config.rs:16
Method
category
(&self)
src/analyzer/instruction.rs:16
Method
category
(&self)
src/analyzer/occupancy.rs:16
Method
category
(&self)
src/analyzer/warp_stall.rs:20
Method
category
(&self)
src/analyzer/memory.rs:16
Method
category
(&self)
src/analyzer/roofline.rs:57
Method
default
()
src/formatter.rs:19
Method
description
(&self)
src/analyzer/launch_config.rs:12
Method
description
(&self)
src/analyzer/instruction.rs:12
Method
description
(&self)
src/analyzer/occupancy.rs:12
Method
description
(&self)
src/analyzer/warp_stall.rs:16
Method
description
(&self)
src/analyzer/memory.rs:12
Method
description
(&self)
src/analyzer/roofline.rs:53
Method
fmt
(&self, f: &mut fmt::Formatter<'_>)
src/severity.rs:14
Method
fmt
(&self, f: &mut std::fmt::Formatter<'_>)
src/analyzer/roofline.rs:16
next →
1–100 of 172, ranked by callers