starred/nix-config
1
0 Fork 0
mirror of https://github.com/ryan4yin/nix-config.git synced 2026-11-13 12:13:46 +01:00
Code Issues Packages Projects Releases 10 Wiki Activity

docs(idols-ai): record the mixed-memory freeze and current state (#433)

Browse Source 
Assisted-by: DeepSeek Harness (gpt-6.1-sol, deepseek-flash)
This commit is contained in:
ryan4yin | 二花 authored and GitHub committed 2026-10-11 20:13:06 +08:00
1 parent c347bfe6a9
commit 5c5c11d646
1 file changed
+58 -47
Download Patch File Download Diff File Expand all files Collapse all files
hosts/idols-ai/MIXED-MEMORY.md
+58 -47
View File
@@ -2,24 +2,26 @@
128G from JUHOR 2×48G + GLOWY 2×16G. All four DIMMs are needed because 96G is insufficient. 128G from JUHOR 2×48G + GLOWY 2×16G. All four DIMMs are needed because 96G is insufficient.
## Target configuration ## Current configuration
This is the settled configuration for normal use. It passed a **6-hour 100G stress test with 0 CPU VDD2 is set to **1.17 V** (BIOS reads **1.168 V**). This is the best configuration tested so
errors** (highest sampled memory temperature: **56.8°C**) and then **29.6 hours of daily use over far, but it still froze during daily use on 2026-10-11. Keep it for normal use for now; higher
two nights and one workday**, displays on and off, LLM serving throughout. Keep these settings tested VDD2 settings have not provided a stable alternative.
rather than increasing voltage further.
| Field | Target | | Field | Setting |
| ----------------------- | ------------------------- | | ----------------------- | -------------------------------------------- |
| XMP | Disabled | | XMP | Disabled |
| Gear mode | Gear2 | | Gear mode | Gear2 |
| Memory speed | 4800 MT/s | | Memory speed | 4800 MT/s |
| CPU VDD2 | **1.17 V fixed** | | CPU VDD2 | **1.17 V fixed (BIOS reads 1.168 V)** |
| VCCSA | Auto (BIOS reads 1.288 V) | | VCCSA | Auto (BIOS reads 1.288 V) |
| DRAM VDD | **1.28 V fixed** | | VDD2_MV (BIOS label) | Auto (BIOS reads 1.280 V) |
| DRAM VDDQ | **1.28 V fixed** | | DRAM VDD | **1.28 V fixed** |
| DRAM VPP | Auto (1.8 V) | | DRAM VDDQ | **1.28 V fixed** |
| tCL / tRCD / tRP / tRAS | 40 / 40 / 40 / 77 | | DRAM VPP | Auto (1.8 V) |
| tCL / tRCD / tRP / tRAS | 40 / 40 / 40 / 77 (last photographed values) |
Voltage settings and readings come from BIOS and user reports, not OS measurements.
## Hardware ## Hardware
@@ -41,47 +43,56 @@ Earlier attempts starting with the 48G pair or all four DIMMs did not reach BIOS
2. Enter BIOS and disable XMP. 2. Enter BIOS and disable XMP.
3. Power off and add the 2×48G in DIMM0. 3. Power off and add the 2×48G in DIMM0.
4. Allow 1 to 3 minutes for retraining, possibly with several reboots. 4. Allow 1 to 3 minutes for retraining, possibly with several reboots.
5. Apply the target settings above. 5. Apply the current settings above.
If failed training prevents BIOS entry, repeat this order; clear CMOS first if necessary. Failed If failed training prevents BIOS entry, repeat this order; clear CMOS first if necessary. Failed
training can silently restore defaults; re-check settings in BIOS afterwards. training can silently restore defaults; re-check settings in BIOS afterwards.
## Test results ## Test results
Earlier DRAM tests used CPU VDD2 Auto (BIOS reading 1.104 V). **Bold values identify the rails CPU VDD2 entries below are BIOS settings: Auto read 1.104 V in the earlier DRAM tests; the 1.17 V
changed in each test.** setting reads 1.168 V. These BIOS readings are not measurements during the OS tests.
| CPU VDD2 | DRAM VDD | DRAM VDDQ | Result | Highest sampled memory temperature | | CPU VDD2 | DRAM VDD | DRAM VDDQ | Result | Highest sampled DIMM temperature |
| ---------- | ------------------------ | ------------------------ | ------------------------------------------------------------ | ---------------------------------- | | ---------- | -------------------- | -------------------- | ------------------------------------------------------------------------------------------------------ | ---------------------------------------------- |
| Auto | **1.10 V** | **1.10 V** | Immediate freezes during memory tests, even 4G. | Not recorded | | Auto | 1.10 V | 1.10 V | Immediate freezes during memory tests, even 4G. | Not recorded |
| Auto | **1.25 V** | **1.25 V** | 8G / 32G / 100G short tests passed; llama.cpp load froze. | Not recorded | | Auto | 1.25 V | 1.25 V | 8G / 32G / 100G short tests passed; llama.cpp load froze. | Not recorded |
| Auto | **1.28 V** | **1.28 V** | 100G short test passed; daily use froze twice on 2026-10-08. | Not recorded | | Auto | 1.28 V | 1.28 V | 100G short test passed; daily use froze on 2026-10-08. | Not recorded |
| Auto | **1.282 V** | **1.282 V** | Display corruption in BIOS. | Not recorded | | Auto | 1.282 V | 1.282 V | Display corruption in BIOS. | Not recorded |
| Auto | **1.29 / 1.30 / 1.32 V** | **1.29 / 1.30 / 1.32 V** | Each setting failed training. | Not recorded | | Auto | 1.29 / 1.30 / 1.32 V | 1.29 / 1.30 / 1.32 V | Each setting failed training. | Not recorded |
| **1.13 V** | 1.28 V | 1.28 V | BIOS normal; no long test. | Not recorded | | 1.13 V | 1.28 V | 1.28 V | BIOS normal; no long test. | Not recorded |
| **1.15 V** | 1.28 V | 1.28 V | 100G, 15 minutes: PASS. | 45.5°C* | | 1.15 V | 1.28 V | 1.28 V | 100G, 15 minutes: PASS. | 45.5°C* |
| **1.15 V** | 1.28 V | 1.28 V | Long test froze; last saved duration: 21 minutes 17 seconds. | 56.5°C | | 1.15 V | 1.28 V | 1.28 V | Long test froze; last saved duration 21 min 17 s. | 56.5°C |
| **1.17 V** | 1.28 V | 1.28 V | 100G, **6 hours: PASS**. | **56.8°C** | | **1.17 V** | **1.28 V** | **1.28 V** | **100G, 6 hours: PASS, 0 errors or hardware incidents.** | **56.8°C** |
| **1.17 V** | 1.28 V | 1.28 V | **2 nights + 1 workday: no freeze**, displays on and off. | Not recorded | | 1.17 V | 1.28 V | 1.28 V | About 56 hours across three boots ending cleanly; next boot froze during LLM serving (2026-10-11). | 47.2°C in the last available monitoring sample |
| **1.20 V** | 1.28 V | 1.28 V | BIOS display corruption and freeze within 10 to 20 seconds. | Not recorded | | 1.20 V | 1.28 V | 1.28 V | Earlier attempt: BIOS display corruption and freeze within 10 to 20 s. | Not recorded |
| 1.20 V | 1.28 V | 1.28 V | 100G test froze with screen corruption; last saved duration 7 min 23 s (2026-10-11). | 55.25°C |
| 1.18 V | 1.28 V | 1.28 V | 100G test froze with screen corruption; last saved duration 14 min 36 s (2026-10-11). | 59.5°C |
| 1.28 V | 1.28 V | 1.28 V | Reached the desktop, then froze; journal spans 16:30:16 to 16:31:20 (64 s). No recorded pressure test. | Not recorded |
| 1.30 V | 1.28 V | 1.28 V | Set in BIOS only; rolled back without booting or testing. | Not tested |
| **1.17 V** | **1.28 V** | **1.28 V** | **100G, 15 minutes: PASS, 0 errors or hardware incidents (2026-10-11 17:29–17:45).** | **59.25°C** |
*The 15-minute test used sparse temperature samples; its true peak is unknown. Long tests sampled *The 1.15 V short test used sparse temperature samples; its true peak is unknown. The monitored long
about every 5 seconds and used a 60°C stop threshold. tests and the 2026-10-11 100G tests sampled about every 5 seconds with a 60°C stop threshold. Saved
durations are lower bounds on time before interruption, not exact freeze times.
Both recent long tests were on 2026-10-09: The passing 1.17 V test ran on 2026-10-09 from 01:29 to 07:30 without swap use. Subsequent daily use
included displays on and off with LLM serving. Sleep/wake is untested.
- At 1.15 V, the last record was at 01:02, with machine uptime of 1 hour 53 seconds. The saved The 1.17 V freeze on 2026-10-11 ended logging at 13:55:46, about 4 h 47 min after boot. Logs stopped
durations are lower bounds, not exact freeze times. during generation without a shutdown sequence or a recorded machine check, BUG or Oops.
- At 1.17 V, the test ran from 01:29 to 07:30, with 0 errors or hardware incidents and no swap use.
Recorded temperatures stayed below 60°C.
## Daily use ## BIOS limitations and conclusion
One unbroken boot, 2026-10-09 01:24 to 2026-10-10 07:02: the 6-hour test and a workday with the Attempts to apply tRFC 1180 and primary timings 42 / 42 / 42 / 82 reverted automatically after
displays on, then a night with them off. Load stayed high — 100G stress test, then continuous LLM saving, according to the user's BIOS checks. Those relaxed settings never became a tested
serving with the KV cache streaming from RAM — and occupancy stayed around 70% or more. configuration. The memory frequency control is also locked at 4800 MT/s in the current BIOS setup.
The table above retains the last photographed timings; a later readback has not been recorded.
No machine check, hardware error, BUG, Oops, or GPU fall-off; no OOM or hung task; no log gap over Keep CPU VDD2 at 1.17 V and DRAM VDD/VDDQ at 1.28 V for now. The passing stress test and daily-use
60 seconds. Sleep/wake is untested — displays off only blanks them. Occupancy and temperature were record make this the best tested option, but the later freeze means stability remains unresolved.
not logged and the board has no EDAC counters, so an error would show up as a freeze or corruption,
not a count. The user reports a possible ambient temperature rise of about 10°C. Temperature may affect the
remaining margin, but neither temperature nor an IMC, DIMM or motherboard fault has been isolated.
Manual timing relaxation and frequency reduction are currently unavailable; further tuning is
paused. Revisit only if freezes continue or those BIOS controls become usable.