-
749fc8a94d
Unify NVIDIA GPU recovery paths
mchus
2026-04-23 20:31:41 +03:00
-
6112094d45
fix(grub): fix bitmap error and menu rendering
v8.39
mchusandClaude Sonnet 4.6
2026-04-22 22:05:16 +03:00
-
e9a2bc9f9d
update submodule
mchus
2026-04-22 20:39:27 +03:00
-
7a8f884664
fix(boot): remove advanced options submenu
v8.38
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-22 19:01:50 +03:00
-
8bf8dfa45b
fix(boot): default to KMS + pci=realloc, drop nomodeset from main entries
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-22 19:00:04 +03:00
-
6a22199aff
chore(bible): bump ascii-safe-text contract
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-22 18:52:10 +03:00
-
ddb2bb5d1c
fix(grub): replace em-dash with ASCII -- in all menu entry titles
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-22 18:52:04 +03:00
-
aa284ae754
fix(iso): avoid grub logo scaling error
v8.37
Mikhail Chusavitin
2026-04-20 14:06:32 +03:00
-
8512098174
fix(iso): restore bootappend-live in canonical boot menu
v8.36
Mikhail Chusavitin
2026-04-20 13:39:05 +03:00
-
6b5d22c194
chore(git): ignore local audit binary
Mikhail Chusavitin
2026-04-20 13:21:35 +03:00
-
a35e90a93e
fix(iso): clear stale bootloader templates in workdir
v8.35
Mikhail Chusavitin
2026-04-20 13:19:50 +03:00
-
1ced81707f
fix(iso): validate live boot entries in final ISO
Mikhail Chusavitin
2026-04-20 13:12:24 +03:00
-
679aeb9947
Run NVIDIA DCGM diag tests on all selected GPUs simultaneously
v8.34
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-20 11:53:25 +03:00
-
647e99b697
Fix post-sync live-build ISO rebuild
v8.33
Mikhail Chusavitin
2026-04-20 11:01:15 +03:00
-
4af997f436
Update audit bee binary
Mikhail Chusavitin
2026-04-20 10:55:42 +03:00
-
6caace0cc0
Make power benchmark report phase-averaged
Mikhail Chusavitin
2026-04-20 10:53:53 +03:00
-
5f0103635b
Update power benchmark GPU reset flow
v8.32
Mikhail Chusavitin
2026-04-20 09:46:00 +03:00
-
84a2551dc0
Fix NVIDIA self-heal recovery flow
Mikhail Chusavitin
2026-04-20 09:43:22 +03:00
-
1cfabc9230
Reset GPUs before power benchmark
Mikhail Chusavitin
2026-04-20 09:42:19 +03:00
-
5dc711de23
Start power calibration from full GPU TDP
Mikhail Chusavitin
2026-04-20 09:28:58 +03:00
-
ab802719f8
Use real NVIDIA power-limit bounds in benchmark
Mikhail Chusavitin
2026-04-20 09:26:56 +03:00
-
a94e8007f8
Ignore power throttling in benchmark calibration
Mikhail Chusavitin
2026-04-20 09:26:29 +03:00
-
c69bf07b27
Commit remaining workspace changes
v8.31
mchus
2026-04-20 07:02:31 +03:00
-
b3cf8e3893
Globalize autotuned system power source
mchus
2026-04-20 07:02:12 +03:00
-
17118298bd
audit: switch power benchmark load to dcgmproftester
mchus
2026-04-20 06:57:14 +03:00
-
65bcc9ce81
refactor(webui): split pages into task modules
mchus
2026-04-20 06:56:52 +03:00
-
0cdfbc5875
fix(iso): restore boot UX and boot logs
mchus
2026-04-19 23:08:09 +03:00
-
cf9b54b600
Use last ramp-step SDR snapshot for PSU loaded power; add deploy script
mchusandClaude Sonnet 4.6
2026-04-19 21:26:44 +03:00
-
0bfb3fe954
Use PSU SDR sum for system power chart when available
mchusandClaude Sonnet 4.6
2026-04-19 19:10:01 +03:00
-
3053cb0710
Fix PSU slot regex: match MSI underscore format PSU1_POWER_IN
mchusandClaude Sonnet 4.6
2026-04-19 19:03:02 +03:00
-
2038489961
Remove MemoryMax=3G from bee-web.service to fix OOM kill during GPU tests
mchusandClaude Sonnet 4.6
2026-04-19 18:52:41 +03:00
-
e35484013e
Use SDR PSU AC input for single-card calibration server power
mchusandClaude Sonnet 4.6
2026-04-19 18:44:13 +03:00
-
2cdf034bb0
Use SDR PSU AC input for per-step server power in power ramp
mchusandClaude Sonnet 4.6
2026-04-19 18:43:36 +03:00
-
b89580c24d
Fix PSU power chart: use name-based SDR matching instead of entity ID
mchusandClaude Sonnet 4.6
2026-04-19 18:39:21 +03:00
-
df1385d3d6
Fix dcgmproftester parallel mode: use staggered script for all multi-GPU runs
mchusandClaude Sonnet 4.6
2026-04-19 18:31:34 +03:00
-
f8cd9a7376
Rework Power Fit report: 90 min stability, aligned tables, PSU/fan sections
v8.30
mchusandClaude Sonnet 4.6
2026-04-19 18:04:12 +03:00
-
d52ec67f8f
Stability hardening, build script fixes, GRUB bee logo
mchusandClaude Sonnet 4.6
2026-04-19 13:08:31 +03:00
-
61c7abaa80
Add multi-source PSU power triangulation and per-slot distribution table
mchusandClaude Sonnet 4.6
2026-04-19 13:07:48 +03:00
-
d60f7758ba
Fix grub-pc directory missing before writing grub.cfg
mchusandClaude Sonnet 4.6
2026-04-19 08:42:17 +03:00
-
52c3a24b76
Compact metrics DB in background to prevent CPU spin under load
mchusandClaude Sonnet 4.6
2026-04-18 15:28:05 +03:00
-
028bb30333
Detect PSU faults during perf and power benchmarks
mchusandClaude Sonnet 4.6
2026-04-18 15:08:41 +03:00
-
7d64e5d215
Fix two stale failing tests
mchusandClaude Sonnet 4.6
2026-04-18 15:07:27 +03:00
-
51b721aeb3
Add real-data duration estimates to benchmark and burn pages
v8.29
mchusandClaude Sonnet 4.6
2026-04-18 10:54:50 +03:00
-
bac89bb6e5
Add real-data duration estimates to validate tab profiles
v8.28
mchusandClaude Sonnet 4.6
2026-04-18 10:51:15 +03:00
-
7a618da1f9
Redesign system power chart as stacked per-PSU area chart
v8.27
mchusandClaude Sonnet 4.6
2026-04-18 10:42:00 +03:00
-
64ae1c0ff0
Sync GRUB and isolinux boot entries; document sync rule
v8.26
mchusandClaude Sonnet 4.6
2026-04-18 10:32:16 +03:00
-
49050ca717
Fix GRUB bootloader config dir: grub-pc → grub-efi
v8.25
mchusandClaude Sonnet 4.6
2026-04-18 10:30:11 +03:00
-
5ba72ab315
Add rsync to initramfs for toram progress output
v8.24
mchusandClaude Sonnet 4.6
2026-04-17 23:52:47 +03:00
-
63363e9629
Add toram boot entry and Install to RAM resume support
mchusandClaude Sonnet 4.6
2026-04-17 23:48:56 +03:00
-
5285c0d101
Capture per-run IPMI power and GPU telemetry in power benchmark
v8.23
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-17 17:59:58 +03:00
-
dca4afb8d0
Seed power ramp with single-card TDP limits
v8.22
Mikhail Chusavitin
2026-04-16 11:43:01 +03:00
-
b4280941f5
Move NCCL and NVBandwidth into validate mode
Mikhail Chusavitin
2026-04-16 11:02:30 +03:00
-
f74976ec4c
Use static overlay wallpaper in ISO build
Mikhail Chusavitin
2026-04-16 10:54:03 +03:00
-
18e24a9aa5
Estimate fan duty from observed RPM maxima
v8.21
v8.20
Mikhail Chusavitin
2026-04-16 10:10:18 +03:00
-
e306250da7
Disable fp64/fp4 in mixed gpu burn
v8.19
v8.18
Mikhail Chusavitin
2026-04-16 10:00:03 +03:00
-
c5b2081ac9
Disable unstable fp4/fp64 benchmark phases
Mikhail Chusavitin
2026-04-16 09:58:02 +03:00
-
434528083e
Power bench: compare GPU-reported TDP vs IPMI server power delta
mchusandClaude Sonnet 4.6
2026-04-16 07:21:02 +03:00
-
30aa30cd67
LiveCD: set Baby Bee wallpaper centered on black background
v8.17
mchusandClaude Sonnet 4.6
2026-04-16 06:57:23 +03:00
-
4f76e1de21
Dashboard: per-device status chips with hover tooltips
mchusandClaude Sonnet 4.6
2026-04-16 06:54:13 +03:00
-
3732e64a4a
Add slowdown temperature exceedance detector to benchmark
mchusandClaude Sonnet 4.6
2026-04-16 06:46:45 +03:00
-
0d925299ff
Use per-GPU temperature limits from nvidia-smi -q for headroom calculation
mchusandClaude Sonnet 4.6
2026-04-16 06:45:15 +03:00
-
a8d5e019a5
Translate report to English; add power anomaly detector
mchusandClaude Sonnet 4.6
2026-04-16 06:42:00 +03:00
-
72ec086568
Restructure benchmark report as balanced scorecard (5 perspectives)
mchusandClaude Sonnet 4.6
2026-04-16 06:40:06 +03:00
-
7a0b0934df
Separate compute score from server quality score
mchusandClaude Sonnet 4.6
2026-04-16 00:45:55 +03:00
-
d8ca0dca2c
Redesign scoring metrics: variance-based sustain scores, throttle stability
mchusandClaude Sonnet 4.6
2026-04-16 00:39:59 +03:00
-
d90250f80a
Fix DCGM cleanup and shorten memory validate
mchus
2026-04-16 00:39:37 +03:00
-
8d6eaef5de
Update perf benchmark report methodology to reflect new design
mchusandClaude Sonnet 4.6
2026-04-16 00:31:58 +03:00
-
732bf4cbab
Redesign power and performance benchmarks with new methodology
mchusandClaude Sonnet 4.6
2026-04-16 00:30:50 +03:00
-
fa6d905a10
Tune bee-gpu-burn single-precision benchmark phases
mchus
2026-04-16 00:05:47 +03:00
-
5c1862ce4c
Use lb clean --all to clear bootstrap cache on every build
v8.16
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 17:37:08 +03:00
-
b65ef2ea1d
Fix: use --debootstrap-options to include ca-certificates in bootstrap
v8.15
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 17:26:01 +03:00
-
533d703c97
Bootstrap ca-certificates so NVIDIA CUDA HTTPS source is trusted
v8.14
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 17:24:20 +03:00
-
04eb4b5a6d
Revert "Pre-download DCGM/fabricmanager debs on host to bypass chroot apt"
v8.13
Mikhail Chusavitin
2026-04-15 17:19:53 +03:00
-
4110dbf8a6
Pre-download DCGM/fabricmanager debs on host to bypass chroot apt
v8.12
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 17:10:23 +03:00
-
7237e4d3e4
Add fabric manager boot and support diagnostics
v8.11
Mikhail Chusavitin
2026-04-15 16:14:26 +03:00
-
ab3ad77cd6
Fix Go module: upgrade modernc.org/libc v1.70.0 → v1.72.0
v8.10
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 14:32:04 +03:00
-
cd9e2cbe13
Fix ramp-up power bench: one task instead of N redundant tasks
v8.9
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 12:29:11 +03:00
-
0317dc58fd
Fix memtest hook: grub.cfg/live.cfg missing during binary hooks is expected
v8.8.3
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 10:33:22 +03:00
-
1c5cb45698
Fix memtest hook: bad ver_arg format in apt-get download
v8.8.2
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 10:15:01 +03:00
-
090b92ca73
Re-enable security repo: kernel 6.1.0-44 is in bookworm-security only
v8.8.1
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 10:02:52 +03:00
-
2dccbc010c
Use MEPHI mirror, disable security repo, fix memtest in ISO build
v8.8
Mikhail ChusavitinandClaude Sonnet 4.6
2026-04-15 09:57:29 +03:00
-
e84c69d360
Fix optional step log dir missing after memtest recovery
mchusandClaude Sonnet 4.6
2026-04-15 07:28:36 +03:00
-
c80a39e7ac
Add power results table, fix benchmark results refresh, bound memtester
v8.7
mchusandClaude Sonnet 4.6
2026-04-15 07:16:18 +03:00
-
a5e0261ff2
Refactor power ramp to use true single-card baselines
mchusandClaude Sonnet 4.6
2026-04-14 23:47:57 +03:00
-
ee422ede3c
Revert "Add raster Easy Bee branding assets"
mchus
2026-04-14 23:00:15 +03:00
-
d560b2fead
Add raster Easy Bee branding assets
mchus
2026-04-14 22:39:25 +03:00
-
3cf2e9c9dc
Run power calibration for all GPUs simultaneously
v8.6
mchusandClaude Sonnet 4.6
2026-04-14 22:25:05 +03:00
-
19dbabd71d
Simplify power calibration: pure binary search, no telemetry guessing
v8.5
mchusandClaude Sonnet 4.6
2026-04-14 22:12:45 +03:00
-
a6a07f2626
Replace linear power derate with binary search + telemetry-guided jump
v8.4
mchusandClaude Sonnet 4.6
2026-04-14 22:05:23 +03:00
-
f87461ee4a
Detect thermal throttle with fans below 100% as cooling misconfiguration
v8.2
mchusandClaude Sonnet 4.6
2026-04-14 21:44:57 +03:00
-
a636146dbd
Fix power calibration failing due to DCGM resource contention
v8.1
mchusandClaude Sonnet 4.6
2026-04-14 20:41:17 +03:00
-
303de2df04
Add slot-aware ramp sequence to bee-bench power
v8.0
Mikhail Chusavitin
2026-04-14 17:47:40 +03:00
-
95124d228f
Split bee-bench into perf and power workflows
Mikhail Chusavitin
2026-04-14 17:33:13 +03:00
-
54338dbae5
Unify live RAM runtime state
Mikhail Chusavitin
2026-04-14 16:18:33 +03:00
-
2be7ae6d28
Refine NVIDIA benchmark phase timing
Mikhail Chusavitin
2026-04-14 14:12:06 +03:00
-
b1a5035edd
Normalize task queue priorities by workflow
v7.21
v7.20
Mikhail Chusavitin
2026-04-14 11:13:54 +03:00
-
8fc986c933
Add benchmark fan duty cycle summary to report
Mikhail Chusavitin
2026-04-14 10:24:02 +03:00
-
88b5e0edf2
Harden IPMI power probe timeout
v7.19
Mikhail Chusavitin
2026-04-14 10:18:23 +03:00
-
82fe1f6d26
Disable precision fallback and pin cuBLAS 13.1
Mikhail Chusavitin
2026-04-14 10:17:44 +03:00
-
81e7c921f8
дебаг при сборке
mchus
2026-04-14 07:02:37 +03:00