TRAIL Compute Status Dashboard

⚠ Outage detected — some cluster data is stale.

Apollo [129.97.163.137] — last successful update 62d 16h ago
Fir [fir.alliancecan.ca] — last successful update 5d 17h ago
Killarney-L40S [killarney.alliancecan.ca] — last successful update 2d 0h ago
Killarney-H100 [killarney.alliancecan.ca] — last successful update 2d 0h ago
Narval [narval.alliancecan.ca] — last successful update 1d 13h ago
Trillium [trillium-gpu.scinet.utoronto.ca] — last successful update 1d 17h ago
UM1 [192.168.42.153] — last successful update 176d 22h ago
UM2 [192.168.42.134] — last successful update 176d 22h ago

Member Usage

Member Usage Allocated Cluster - Narval, Trillium - GPU util only on Trillium - Inactive = cpus/gpu<1 & mem<10GB

Member	14D GPU HRS	14D GPU Util	14D Inactive GPU Hrs	14D CLUSTERS
nisarbar	276	84% (276 GPU hrs)	0	Trillium
lavoiem9	240	81% (231 GPU hrs)	2	Trillium
carril34	126	69% (28 GPU hrs)	0	Narval, Trillium
ikeefe12	125	31% (125 GPU hrs)	0	Trillium
lla	26	25% (4 GPU hrs)	0	Narval

Member Usage Other Cluster - Killarney, TamIA, Fir, Rorqual - GPU util only on Fir - Inactive = cpus/gpu<1 & mem<10GB

Member	14D GPU HRS	14D GPU Util	14D Inactive GPU Hrs	14D CLUSTERS
carril34	169	-	3	Rorqual

Member Usage UTIAS Server - DGX, Apollo, Turing, Lovelace, UMS, UM1, UM2, UM3

⚠ Only showing data from 7 days temporarily due to recent tracking changes.

Member	14D GPU HRS	14D Servers	14D DAYS USED
galcohen	384	Lovelace	7
trossen	334	DGX	6
lock_gpu	161	DGX	8
tony	36	UM3	3
mlavoie	28	Lovelace, Turing	4
hyin	16	UM3	3
rafid	4	Turing	1

Cluster Usage

Allocated Cluster Usage

Cluster	Node Type	GPU NODES	LEVELFS	14D GPU HRS USED	14D GPU HRS TARGET	14D QUEUE TIME	14D USERS
Narval (stale · 1d 13h ago)	4xA100-80GB	174	1.021	123	1331	52.8h	2 (carril34, lla)
Trillium (stale · 1d 17h ago)	4xH100-80GB	61	0.658	670	665	1.9h	4 (carril34, ikeefe12, lavoiem9, nisarbar)

UTIAS Server Usage

⚠ Only showing data from 7 days temporarily due to recent tracking changes.

Cluster	Node Type	GPU UTIL	Users	14D GPU UTIL	14D USERS
DGX	4xA100-SXM4-40GB	4/4	—	54%	2 (lock_gpu, trossen)
Apollo (stale · 62d 16h ago)	8xV100-32GB	0/8	0	—	—
Turing	4xRTX6000-Ada-48GB	4/4	0	8%	2 (mlavoie, rafid)
Lovelace	4xRTX6000-Ada-48GB	0/4	2 (galcohen +1)	62%	2 (galcohen, mlavoie)
UMS	1xRTX4090-24GB	0/1	1 (trailbot)	0%	1 (trailbot)
UM1 (stale · 176d 22h ago)	1xRTX4090-24GB	0/1	1 (trail)	—	—
UM2 (stale · 176d 22h ago)	1xRTX4090-24GB	0/1	0	—	—
UM3	1xRTX4090-24GB	0/1	1 (tony)	30%	3 (hyin, tony, yyamamoto)

Other Cluster Usage

Cluster	Node Type	GPU NODES	LEVELFS	14D GPU HRS USED	14D GPU HRS TARGET	14D QUEUE TIME	Location
Killarney-L40S (stale · 2d 0h ago)	4xL40S-48GB	168	18.027	0	—	—	Vector, Toronto
Killarney-H100 (stale · 2d 0h ago)	8xH100-48GB	10	18.027	0	—	—	Vector, Toronto
TamIA	4xH100-HGX-80GB	53	876785.268	0	—	—	Mila, Montreal
Fir (stale · 5d 17h ago)	4xH100-SXM5-80GB	160 (60 MIG)	4912.904	0	—	—	SFU, Vancouver
Rorqual	4xH100-SXM5-80GB	93 (27 MIG)	0.723	169	—	2.8h	ETS, Montreal

UTIAS SERVERS historical data

GPU utilisation over time · last 14 days, % of GPUs in use, sampled each refresh

External clusters historical data

Queue time, last observed vs 14d median · external clusters · ↖ most recent job (or longest current pending) · ↘ median over last 14 days

Combined queue time heatmap for external clusters

Historical Trends - 14d, queue times only shown for top 5 ...

External clusters live data

Last updated 2026-10-25 06:02:04 · Source and refresh procedure: github.com/TRAILab/trail-cluster-status

Numbers come from SLURM on each cluster, read over SSH when the page is refreshed. A cell marked stale means that cluster was unreachable and the last known value is shown instead.