AI/rocm-systems
2
0
Çatalla 0
Kod Konular Değişiklik İstekleri İşlemler Paketler Projeler Sürüm Viki Aktivite

correct small negatives in remote read/write traffic due to running in different passes

Kaynağa Gözat 
Signed-off-by: Nicholas Curtis <nicurtis@amd.com>
Bu işleme şunda yer alıyor:
Nicholas Curtis
2024-02-27 14:38:09 -05:00
işlemeyi yapan: Cole Ramos
ebeveyn 2c3c5fba35
işleme 20d93703fc
9 değiştirilmiş dosya ile 166 ekleme ve 219 silme
Yama Dosyasını İndir Diff Dosyasını İndir Tüm dosyaları genişlet Tüm dosyaları daralt
src/omniperf_soc/analysis_configs/gfx908/1800_L2_cache_per_channel.yaml
+16 -16
Dosyayı Görüntüle
@@ -10,7 +10,7 @@ Panel Config:
data source:
- metric_table:
id: 1801
title: Aggregate Stats (All 32 channels)
title: Aggregate Stats (All channels)
header:
metric: Metric
avg: Avg
@@ -171,7 +171,7 @@ Panel Config:
- metric_table:
id: 1802
title: L2 Cache Hit Rate (%)
title: L2 Cache Hit Rate (pct)
header:
metric: Metric
expr: Expression
@@ -186,7 +186,7 @@ Panel Config:
- metric_table:
id: 1803
title: L2 Requests (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
expr: Expression
@@ -199,12 +199,12 @@ Panel Config:
- metric_table:
id: 1804
title: L2 Access (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
read req: L2 Read Req
write req: L2 Write Req
atomic req: L2 Atomic Req
read req: L2 Read
write req: L2 Write
atomic req: L2 Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_READ[::_1]) / $denom))
@@ -216,12 +216,12 @@ Panel Config:
- metric_table:
id: 1805
title: L2 - Fabric Access (Requests)
title: L2-Fabric Requests (per normUnit)
header:
metric: Metric
read req: L2 - Fabric Read Req
write req: L2 - Fabric Write and Atomic Req
atomic req: L2 - Fabric Atomic Req
read req: L2-Fabric Read
write req: L2-Fabric Write and Atomic
atomic req: L2-Fabric Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_EA_RDREQ[::_1]) / $denom))
@@ -256,7 +256,7 @@ Panel Config:
- metric_table:
id: 1806
title: L2 - Fabric Read Latency (Cycles)
title: L2-Fabric Read Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -271,7 +271,7 @@ Panel Config:
- metric_table:
id: 1807
title: L2 - Fabric Write Latency (Cycles)
title: L2-Fabric Write and Atomic Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -286,7 +286,7 @@ Panel Config:
- metric_table:
id: 1808
title: L2 - Fabric Atomic Latency (Cycles)
title: L2-Fabric Atomic Latency (Cycles per normUnit)
header:
metric: Metric
expr: Expression
@@ -300,7 +300,7 @@ Panel Config:
- metric_table:
id: 1809
title: L2 - Fabric Read Stall (Cycles per normUnit)
title: L2-Fabric Read Stall (Cycles per normUnit)
header:
metric: Metric
ea read stall - pcie: L2 - Fabric Read Stall (PCIe)
@@ -317,7 +317,7 @@ Panel Config:
- metric_table:
id: 1810
title: L2 - Fabric Write Stall (Cycles per normUnit)
title: L2-Fabric Write and Atomic Stall (Cycles per normUnit)
header:
metric: Metric
ea write stall - pcie: L2 - Fabric Write Stall (PCIe)
src/omniperf_soc/analysis_configs/gfx90a/1800_L2_cache_per_channel.yaml
+24 -24
Dosyayı Görüntüle
@@ -6,11 +6,11 @@ Metric Description:
# Define the panel properties and properties of each metric in the panel.
Panel Config:
id: 1800
title: L2 Cache (per Channel)
title: L2 Cache (per Metric)
data source:
- metric_table:
id: 1801
title: Aggregate Stats (All 32 channels)
title: Aggregate Stats (All channels)
header:
metric: Metric
avg: Avg
@@ -171,7 +171,7 @@ Panel Config:
- metric_table:
id: 1802
title: L2 Cache Hit Rate (%)
title: L2 Cache Hit Rate (pct)
header:
metric: Metric
expr: Expression
@@ -186,7 +186,7 @@ Panel Config:
- metric_table:
id: 1803
title: L2 Requests (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
expr: Expression
@@ -199,12 +199,12 @@ Panel Config:
- metric_table:
id: 1804
title: L2 Access (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
read req: L2 Read Req
write req: L2 Write Req
atomic req: L2 Atomic Req
read req: L2 Read
write req: L2 Write
atomic req: L2 Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_READ[::_1]) / $denom))
@@ -216,12 +216,12 @@ Panel Config:
- metric_table:
id: 1805
title: L2 - Fabric Access (Requests)
title: L2-Fabric Requests (per normUnit)
header:
metric: Metric
read req: L2 - Fabric Read Req
write req: L2 - Fabric Write and Atomic Req
atomic req: L2 - Fabric Atomic Req
read req: L2-Fabric Read
write req: L2-Fabric Write and Atomic
atomic req: L2-Fabric Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_EA_RDREQ[::_1]) / $denom))
@@ -256,7 +256,7 @@ Panel Config:
- metric_table:
id: 1806
title: L2 - Fabric Read Latency (Cycles)
title: L2-Fabric Read Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -271,7 +271,7 @@ Panel Config:
- metric_table:
id: 1807
title: L2 - Fabric Write Latency (Cycles)
title: L2-Fabric Write and Atomic Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -286,7 +286,7 @@ Panel Config:
- metric_table:
id: 1808
title: L2 - Fabric Atomic Latency (Cycles)
title: L2-Fabric Atomic Latency (Cycles per normUnit)
header:
metric: Metric
expr: Expression
@@ -300,12 +300,12 @@ Panel Config:
- metric_table:
id: 1809
title: L2 - Fabric Read Stall (Cycles per normUnit)
title: L2-Fabric Read Stall (Cycles per normUnit)
header:
metric: Metric
ea read stall - pcie: L2 - Fabric Read Stall (PCIe)
ea read stall - if: L2 - Fabric Read Stall (Infinity Fabric™)
ea read stall - hbm: L2 - Fabric Read Stall (HBM)
ea read stall - pcie: L2-Fabric Read Stall (PCIe)
ea read stall - if: L2-Fabric Read Stall (Infinity Fabric™)
ea read stall - hbm: L2-Fabric Read Stall (HBM)
metric:
"::_1":
ea read stall - pcie: None # Missing perfmon
@@ -317,13 +317,13 @@ Panel Config:
- metric_table:
id: 1810
title: L2 - Fabric Write Stall (Cycles per normUnit)
title: L2-Fabric Write and Atomic Stall (Cycles per normUnit)
header:
metric: Metric
ea write stall - pcie: L2 - Fabric Write Stall (PCIe)
ea write stall - if: L2 - Fabric Write Stall (Infinity Fabric™)
ea write stall - hbm: L2 - Fabric Write Stall (HBM)
ea write stall - starve: L2 - Fabric Write Starve
ea write stall - pcie: L2-Fabric Write Stall (PCIe)
ea write stall - if: L2-Fabric Write Stall (Infinity Fabric™)
ea write stall - hbm: L2-Fabric Write Stall (HBM)
ea write stall - starve: L2-Fabric Write Starve
metric:
"::_1":
ea write stall - pcie: None # Missing perfmon
src/omniperf_soc/analysis_configs/gfx940/1700_L2_cache.yaml
+6 -6
Dosyayı Görüntüle
@@ -68,9 +68,9 @@ Panel Config:
unit: pct
tips:
Remote Read Traffic:
avg: AVG((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
min: MIN((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
max: MAX((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
avg: AVG((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum), 0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
min: MIN((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum), 0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
max: MAX((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum), 0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
unit: pct
tips:
Uncached Read Traffic:
@@ -98,9 +98,9 @@ Panel Config:
unit: pct
tips:
Remote Write and Atomic Traffic:
avg: AVG((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
min: MIN((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
max: MAX((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
avg: AVG((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
min: MIN((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
max: MAX((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
unit: pct
tips:
Atomic Traffic:
src/omniperf_soc/analysis_configs/gfx940/1800_L2_cache_per_channel.yaml
+29 -56
Dosyayı Görüntüle
@@ -6,11 +6,11 @@ Metric Description:
# Define the panel properties and properties of each metric in the panel.
Panel Config:
id: 1800
title: L2 Cache (per Channel)
title: L2 Cache (per Metric)
data source:
- metric_table:
id: 1801
title: Aggregate Stats (All 32 channels)
title: Aggregate Stats (All channels)
header:
metric: Metric
avg: Avg
@@ -107,7 +107,7 @@ Panel Config:
- metric_table:
id: 1802
title: L2 Cache Hit Rate (%)
title: L2 Cache Hit Rate (pct)
header:
metric: Metric
expr: Expression
@@ -122,7 +122,7 @@ Panel Config:
- metric_table:
id: 1803
title: Requests (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
expr: Expression
@@ -135,12 +135,12 @@ Panel Config:
- metric_table:
id: 1804
title: L1-L2 Access (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
read req: L1-L2 Read
write req: L1-L2 Write
atomic req: L1-L2 Atomic
read req: L2 Read
write req: L2 Write
atomic req: L2 Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_READ[::_1]) / $denom))
@@ -152,12 +152,12 @@ Panel Config:
- metric_table:
id: 1805
title: L2-EA Access (Requests)
title: L2-Fabric Requests (per normUnit)
header:
metric: Metric
read req: L2-EA Read
write req: L2-EA Write
atomic req: L2-EA Atomic
read req: L2-Fabric Read
write req: L2-Fabric Write and Atomic
atomic req: L2-Fabric Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_EA0_RDREQ[::_1]) / $denom))
@@ -169,12 +169,12 @@ Panel Config:
# - metric_table:
# id: 1806
# title: L2-EA Latency (Cycles)
# title: L2-Fabric Latency (Cycles)
# header:
# metric: Metric
# read lat: L2-EA Read
# write lat: L2-EA Write
# atomic lat: L2-EA Atomic
# read lat: L2-Fabric Read
# write lat: L2-Fabric Write
# atomic lat: L2-Fabric Atomic
# metric:
# "::_1":
# read lat:
@@ -192,7 +192,7 @@ Panel Config:
- metric_table:
id: 1806
title: L2-EA Read Latency (Cycles)
title: L2-Fabric Read Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -207,7 +207,7 @@ Panel Config:
- metric_table:
id: 1807
title: L2-EA Write Latency (Cycles)
title: L2-Fabric Write and Atomic Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -222,7 +222,7 @@ Panel Config:
- metric_table:
id: 1808
title: L2-EA Atomic Latency (Cycles)
title: L2-Fabric Atomic Latency (Cycles per normUnit)
header:
metric: Metric
expr: Expression
@@ -236,12 +236,12 @@ Panel Config:
- metric_table:
id: 1809
title: L2-EA Read Stall (Cycles per normUnit)
title: L2-Fabric Read Stall (Cycles per normUnit)
header:
metric: Metric
ea read stall - io: L2-EA Read Stall - IO
ea read stall - gmi: L2-EA Read Stall - GMI
ea read stall - dram: L2-EA Read Stall - DRAM
ea read stall - io: L2-Fabric Read Stall - IO
ea read stall - gmi: L2-Fabric Read Stall - GMI
ea read stall - dram: L2-Fabric Read Stall - DRAM
metric:
"::_1":
ea read stall - io: None # Missing perfmon
@@ -253,46 +253,19 @@ Panel Config:
- metric_table:
id: 1810
title: L2-EA Write Stall (Cycles per normUnit)
title: L2-Fabric Write and Atomic Stall (Cycles per normUnit)
header:
metric: Metric
ea write stall - io: L2-EA Write Stall - IO
ea write stall - gmi: L2-EA Write Stall - GMI
ea write stall - dram: L2-EA Write Stall - DRAM
ea write stall - starve: L2-EA Write Stall - Starve
ea write stall - io: L2-Fabric Write Stall - IO
ea write stall - gmi: L2-Fabric Write Stall - GMI
ea write stall - dram: L2-Fabric Write Stall - DRAM
ea write stall - starve: L2-Fabric Write Stall - Starve
metric:
"::_1":
ea write stall - io: None # Missing perfmon
ea write stall - gmi: None # Missing perfmon
ea write stall - dram: None # Missing perfmon
ea write stall - starve: AVG((TO_INT(TCC_TOO_MANY_EA_WRREQS_STALL[::_1]) / $denom))
ea write stall - starve: None # Missing perfmon
placeholder_range:
"::_1": $total_l2_chan
cli_style: simple_multiple_bar
- metric_table:
id: 1811
title: L2 Tag Stall (cycles)
header:
metric: Metric
expr: Expression
metric:
"::_1":
expr: TCC_TAG_STALL[::_1]
placeholder_range:
"::_1": $total_l2_chan
cli_style: simple_box
- metric_table:
id: 1812
title: L2 Bubble (128B request)
header:
metric: Metric
expr: Expression
metric:
"::_1":
expr: TCC_BUBBLE[::_1]
placeholder_range:
"::_1": $total_l2_chan
# tips: Number of 128-byte read requests sent to EA
cli_style: simple_box
src/omniperf_soc/analysis_configs/gfx941/1700_L2_cache.yaml
+6 -6
Dosyayı Görüntüle
@@ -68,9 +68,9 @@ Panel Config:
unit: pct
tips:
Remote Read Traffic:
avg: AVG((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
min: MIN((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
max: MAX((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
avg: AVG((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum),0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
min: MIN((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum),0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
max: MAX((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum),0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
unit: pct
tips:
Uncached Read Traffic:
@@ -98,9 +98,9 @@ Panel Config:
unit: pct
tips:
Remote Write and Atomic Traffic:
avg: AVG((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
min: MIN((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
max: MAX((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
avg: AVG((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
min: MIN((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
max: MAX((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
unit: pct
tips:
Atomic Traffic:
src/omniperf_soc/analysis_configs/gfx941/1800_L2_cache_per_channel.yaml
+29 -56
Dosyayı Görüntüle
@@ -6,11 +6,11 @@ Metric Description:
# Define the panel properties and properties of each metric in the panel.
Panel Config:
id: 1800
title: L2 Cache (per Channel)
title: L2 Cache (per Metric)
data source:
- metric_table:
id: 1801
title: Aggregate Stats (All 32 channels)
title: Aggregate Stats (All channels)
header:
metric: Metric
avg: Avg
@@ -107,7 +107,7 @@ Panel Config:
- metric_table:
id: 1802
title: L2 Cache Hit Rate (%)
title: L2 Cache Hit Rate (pct)
header:
metric: Metric
expr: Expression
@@ -122,7 +122,7 @@ Panel Config:
- metric_table:
id: 1803
title: Requests (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
expr: Expression
@@ -135,12 +135,12 @@ Panel Config:
- metric_table:
id: 1804
title: L1-L2 Access (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
read req: L1-L2 Read
write req: L1-L2 Write
atomic req: L1-L2 Atomic
read req: L2 Read
write req: L2 Write
atomic req: L2 Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_READ[::_1]) / $denom))
@@ -152,12 +152,12 @@ Panel Config:
- metric_table:
id: 1805
title: L2-EA Access (Requests)
title: L2-Fabric Requests (per normUnit)
header:
metric: Metric
read req: L2-EA Read
write req: L2-EA Write
atomic req: L2-EA Atomic
read req: L2-Fabric Read
write req: L2-Fabric Write and Atomic
atomic req: L2-Fabric Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_EA0_RDREQ[::_1]) / $denom))
@@ -169,12 +169,12 @@ Panel Config:
# - metric_table:
# id: 1806
# title: L2-EA Latency (Cycles)
# title: L2-Fabric Latency (Cycles)
# header:
# metric: Metric
# read lat: L2-EA Read
# write lat: L2-EA Write
# atomic lat: L2-EA Atomic
# read lat: L2-Fabric Read
# write lat: L2-Fabric Write
# atomic lat: L2-Fabric Atomic
# metric:
# "::_1":
# read lat:
@@ -192,7 +192,7 @@ Panel Config:
- metric_table:
id: 1806
title: L2-EA Read Latency (Cycles)
title: L2-Fabric Read Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -207,7 +207,7 @@ Panel Config:
- metric_table:
id: 1807
title: L2-EA Write Latency (Cycles)
title: L2-Fabric Write and Atomic Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -222,7 +222,7 @@ Panel Config:
- metric_table:
id: 1808
title: L2-EA Atomic Latency (Cycles)
title: L2-Fabric Atomic Latency (Cycles per normUnit)
header:
metric: Metric
expr: Expression
@@ -236,12 +236,12 @@ Panel Config:
- metric_table:
id: 1809
title: L2-EA Read Stall (Cycles per normUnit)
title: L2-Fabric Read Stall (Cycles per normUnit)
header:
metric: Metric
ea read stall - io: L2-EA Read Stall - IO
ea read stall - gmi: L2-EA Read Stall - GMI
ea read stall - dram: L2-EA Read Stall - DRAM
ea read stall - io: L2-Fabric Read Stall - IO
ea read stall - gmi: L2-Fabric Read Stall - GMI
ea read stall - dram: L2-Fabric Read Stall - DRAM
metric:
"::_1":
ea read stall - io: None # Missing perfmon
@@ -253,46 +253,19 @@ Panel Config:
- metric_table:
id: 1810
title: L2-EA Write Stall (Cycles per normUnit)
title: L2-Fabric Write and Atomic Stall (Cycles per normUnit)
header:
metric: Metric
ea write stall - io: L2-EA Write Stall - IO
ea write stall - gmi: L2-EA Write Stall - GMI
ea write stall - dram: L2-EA Write Stall - DRAM
ea write stall - starve: L2-EA Write Stall - Starve
ea write stall - io: L2-Fabric Write Stall - IO
ea write stall - gmi: L2-Fabric Write Stall - GMI
ea write stall - dram: L2-Fabric Write Stall - DRAM
ea write stall - starve: L2-Fabric Write Stall - Starve
metric:
"::_1":
ea write stall - io: None # Missing perfmon
ea write stall - gmi: None # Missing perfmon
ea write stall - dram: None # Missing perfmon
ea write stall - starve: TCC_TOO_MANY_EA_WRREQS_STALL
ea write stall - starve: None # Missing perfmon
placeholder_range:
"::_1": $total_l2_chan
cli_style: simple_multiple_bar
- metric_table:
id: 1811
title: L2 Tag Stall (cycles)
header:
metric: Metric
expr: Expression
metric:
"::_1":
expr: TCC_TAG_STALL[::_1]
placeholder_range:
"::_1": $total_l2_chan
cli_style: simple_box
- metric_table:
id: 1812
title: L2 Bubble (128B request)
header:
metric: Metric
expr: Expression
metric:
"::_1":
expr: TCC_BUBBLE[::_1]
placeholder_range:
"::_1": $total_l2_chan
# tips: Number of 128-byte read requests sent to EA
cli_style: simple_box
src/omniperf_soc/analysis_configs/gfx942/1700_L2_cache.yaml
+6 -6
Dosyayı Görüntüle
@@ -72,9 +72,9 @@ Panel Config:
unit: pct
tips:
Remote Read Traffic:
avg: AVG((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
min: MIN((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
max: MAX((100 * ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
avg: AVG((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum),0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
min: MIN((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum),0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
max: MAX((100 * (MAX((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_DRAM_sum),0) / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum != 0) else None))
unit: pct
tips:
Uncached Read Traffic:
@@ -102,9 +102,9 @@ Panel Config:
unit: pct
tips:
Remote Write and Atomic Traffic:
avg: AVG((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
min: MIN((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
max: MAX((100 * ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
avg: AVG((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
min: MIN((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
max: MAX((100 * (MAX((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_DRAM_sum),0) / TCC_EA0_WRREQ_sum) if (TCC_EA0_WRREQ_sum != 0) else None))
unit: pct
tips:
Atomic Traffic:
src/omniperf_soc/analysis_configs/gfx942/1800_L2_cache_per_channel.yaml
+43 -43
Dosyayı Görüntüle
@@ -6,11 +6,11 @@ Metric Description:
# Define the panel properties and properties of each metric in the panel.
Panel Config:
id: 1800
title: L2 Cache (per Channel)
title: L2 Cache (per Metric)
data source:
- metric_table:
id: 1801
title: Aggregate Stats (All 32 channels)
title: Aggregate Stats (All channels)
header:
metric: Metric
avg: Avg
@@ -107,7 +107,7 @@ Panel Config:
- metric_table:
id: 1802
title: L2 Cache Hit Rate (%)
title: L2 Cache Hit Rate (pct)
header:
metric: Metric
expr: Expression
@@ -122,7 +122,7 @@ Panel Config:
- metric_table:
id: 1803
title: Requests (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
expr: Expression
@@ -135,12 +135,12 @@ Panel Config:
- metric_table:
id: 1804
title: L1-L2 Access (Requests)
title: L2 Requests (per normUnit)
header:
metric: Metric
read req: L1-L2 Read
write req: L1-L2 Write
atomic req: L1-L2 Atomic
read req: L2 Read
write req: L2 Write
atomic req: L2 Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_READ[::_1]) / $denom))
@@ -152,12 +152,12 @@ Panel Config:
- metric_table:
id: 1805
title: L2-EA Access (Requests)
title: L2-Fabric Requests (per normUnit)
header:
metric: Metric
read req: L2-EA Read
write req: L2-EA Write
atomic req: L2-EA Atomic
read req: L2-Fabric Read
write req: L2-Fabric Write and Atomic
atomic req: L2-Fabric Atomic
metric:
"::_1":
read req: AVG((TO_INT(TCC_EA0_RDREQ[::_1]) / $denom))
@@ -169,12 +169,12 @@ Panel Config:
# - metric_table:
# id: 1806
# title: L2-EA Latency (Cycles)
# title: L2-Fabric Latency (Cycles)
# header:
# metric: Metric
# read lat: L2-EA Read
# write lat: L2-EA Write
# atomic lat: L2-EA Atomic
# read lat: L2-Fabric Read
# write lat: L2-Fabric Write
# atomic lat: L2-Fabric Atomic
# metric:
# "::_1":
# read lat:
@@ -192,7 +192,7 @@ Panel Config:
- metric_table:
id: 1806
title: L2-EA Read Latency (Cycles)
title: L2-Fabric Read Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -207,7 +207,7 @@ Panel Config:
- metric_table:
id: 1807
title: L2-EA Write Latency (Cycles)
title: L2-Fabric Write and Atomic Latency (Cycles)
header:
metric: Metric
expr: Expression
@@ -222,7 +222,7 @@ Panel Config:
- metric_table:
id: 1808
title: L2-EA Atomic Latency (Cycles)
title: L2-Fabric Atomic Latency (Cycles per normUnit)
header:
metric: Metric
expr: Expression
@@ -236,12 +236,12 @@ Panel Config:
- metric_table:
id: 1809
title: L2-EA Read Stall (Cycles per normUnit)
title: L2-Fabric Read Stall (Cycles per normUnit)
header:
metric: Metric
ea read stall - io: L2-EA Read Stall - IO
ea read stall - gmi: L2-EA Read Stall - GMI
ea read stall - dram: L2-EA Read Stall - DRAM
ea read stall - io: L2-Fabric Read Stall - IO
ea read stall - gmi: L2-Fabric Read Stall - GMI
ea read stall - dram: L2-Fabric Read Stall - DRAM
metric:
"::_1":
ea read stall - io: None # Missing perfmon
@@ -253,45 +253,45 @@ Panel Config:
- metric_table:
id: 1810
title: L2-EA Write Stall (Cycles per normUnit)
title: L2-Fabric Write and Atomic Stall (Cycles per normUnit)
header:
metric: Metric
ea write stall - io: L2-EA Write Stall - IO
ea write stall - gmi: L2-EA Write Stall - GMI
ea write stall - dram: L2-EA Write Stall - DRAM
ea write stall - starve: L2-EA Write Stall - Starve
ea write stall - io: L2-Fabric Write Stall - IO
ea write stall - gmi: L2-Fabric Write Stall - GMI
ea write stall - dram: L2-Fabric Write Stall - DRAM
ea write stall - starve: L2-Fabric Write Stall - Starve
metric:
"::_1":
ea write stall - io: None # Missing perfmon
ea write stall - gmi: None # Missing perfmon
ea write stall - dram: None # Missing perfmon
ea write stall - starve: TCC_TOO_MANY_EA_WRREQS_STALL
ea write stall - starve: None # Missing perfmon
placeholder_range:
"::_1": $total_l2_chan
cli_style: simple_multiple_bar
- metric_table:
id: 1811
title: L2 Tag Stall (cycles)
header:
metric: Metric
expr: Expression
metric:
"::_1":
expr: TCC_TAG_STALL[::_1]
placeholder_range:
"::_1": $total_l2_chan
cli_style: simple_box
# - metric_table:
# id: 1811
# title: L2 Tag Stall (cycles)
# header:
# metric: Metric
# expr: Expression
# metric:
# "::_1":
# expr: TCC_TAG_STALL[::_1]
# placeholder_range:
# "::_1": $total_l2_chan
# cli_style: simple_box
- metric_table:
id: 1812
title: L2 Bubble (128B request)
title: L2-Fabric (128B read requests per normUnit)
header:
metric: Metric
expr: Expression
metric:
"::_1":
expr: TCC_BUBBLE[::_1]
expr: (TO_INT(TCC_BUBBLE[::_1]) / $denom)
placeholder_range:
"::_1": $total_l2_chan
# tips: Number of 128-byte read requests sent to EA
src/utils/schema.py
+7 -6
Dosyayı Görüntüle
@@ -91,12 +91,13 @@ supported_field = [
"Channel",
"L2 Cache Hit Rate",
"Requests",
"L1-L2 Read",
"L1-L2 Write",
"L1-L2 Atomic",
"L2-EA Read",
"L2-EA Write",
"L2-EA Atomic",
"L2 Read",
"L2 Write",
"L2 Atomic",
"L2-Fabric Requests",
"L2-Fabric Read",
"L2-Fabric Write and Atomic",
"L2-Fabric Atomic",
"L2 Read Req",
"L2 Write Req",
"L2 Atomic Req",