new error raise RuntimeError("the engine exited before it was ready" + (f" (see {log})" if log else "") +

  1) swift-1.5-iq3_xxs
  2) qwen3.8-flash-next-coder-iq1_m
  3) install another model / change settings
Which one? [1]: 1
  [ok] GPU: GPU 0 (NVIDIA GeForce RTX 4090 Laptop GPU, 16 GB)

  ----------------------------------------------------------------------------------------------------
  Starting swift-1.5-iq3_xxs: it loads about 40 GB into RAM and locks part of it for the GPU.
  While it does, YOUR PC CAN BE SLOW OR STOP RESPONDING FOR 1-3 MINUTES (longer the first time after a
  restart). That is normal: please wait and don't close this window - the browser opens when it is ready.
  Later, closing this window stops the model.
  ----------------------------------------------------------------------------------------------------
loading the model (the first start takes a minute or two) ...
[strata] starting the engine: reading the model's weights ...
[strata] loading the experts into RAM (about 40 GB) and locking part of them for the GPU.
         YOUR PC CAN BE SLOW OR STOP RESPONDING FOR 1-3 MINUTES NOW - this is normal.
         Please wait and don't close this window; the browser opens when it is ready.
[strata] still starting (26 s) - please wait ...
[strata] experts loaded: 39.97 GiB at 1.39 GiB/s (36 s so far)
[strata] filling the GPU's expert cache (5235 experts, 8.49 GiB of VRAM) ...
[strata] almost ready ...
Traceback (most recent call last):
  File "E:\Strata\serve\server.py", line 2703, in <module>
    sys.exit(main())
             ~~~~^^
  File "E:\Strata\serve\server.py", line 2580, in main
    engine = StrataEngine(exe, engine_args(cfg), cwd=cfg.get("cwd"), log=cfg.get("log"), env=env, lazy=lazy)
  File "E:\Strata\serve\server.py", line 291, in __init__
    raise RuntimeError("the engine exited before it was ready" + (f" (see {log})" if log else "") +
                       start_failure_hint(log, log_start))
RuntimeError: the engine exited before it was ready (see E:\Strata\strata-swift-iq3_xxs.log)
Press any key to continue . . .

hadn't updated this on laptop for a few days was working fine, latest update as of now, seems broken, 16gb vram, and onboard intel gpu, it would be nice if it could just be set to use a 12gb vram, not sure why it wasa picking 8.49gb, nothing else was running other than desktop and dedicated gpu had plenty of vram to available ie more than 14gb was free to use

the log file

strata generate: native pack: E:\Strata\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding type 21 in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: PLE on, table 320001536 rows of E:\Strata\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata generate: expert arena: locked 9477 MiB via working-set minimum + VirtualLock; cudaHostRegister of the whole arena FAILED (out of memory); 38 slices pinned (30 GiB); large pages refused (GetLargePageMinimum=2097152, VirtualAlloc error 87 - needs SeLockMemoryPrivilege); using 4 KB pages
strata generate: loaded 39.97 GiB at 0.98 GiB/s
strata mtp: draft layer loaded, 949 MiB of VRAM (experts 675, dense 111)
strata generate: profile E:\Strata\data\expert-profile.bin: 8000 ranked pairs, built for 8000 slots
strata generate: expert cache auto: 8.55 GiB free, 700 MiB reserved -> 3624 slots
strata generate: expert cache 4845 slots, 7.86 GiB of VRAM; policy is
strata generate: *** WARNING: --expert-cache is enabled and the GPU hit path is NOT
                 CORRECT. The generated tokens diverge from a cache-off run (measured:
                 first difference at token 40 at 2.97% hits, token 0 at 54.4%). Any
                 timing from this run is real; any OUTPUT from it is not. ***
                 PROFILE, ranked by routing frequency, no eviction.
strata generate: pre-filled 4845 of 4845 slots from the profile; slot 0 verified
strata generate: R4 hit path ON - resident experts are computed on the GPU
strata generate: 23 expert-pool workers + the host thread
strata generate: session is up; locating the head
strata generate: experimental native Q5_K head, 437043200 bytes
strata generate: token graph hit path: 4845 resident experts, decided on the device
strata serve: the prompt path borrows 962 cache slots (1.53 GiB)
strata verify: window up to 4 tokens, 58.3 MiB of device buffers
strata mtp: draft head over 40525 tokens (68.0 MiB)
strata serve: 136 MiB of VRAM free with everything loaded
strata generate: this CPU has no AVX-512: the expert kernels run on AVX2
strata generate: native pack: E:\Strata-data\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding type 21 in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata-data\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: PLE on, table 320001536 rows of E:\Strata-data\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata generate: expert arena: locked 9477 MiB via working-set minimum + VirtualLock; cudaHostRegister of the whole arena FAILED (out of memory); 38 slices pinned (30 GiB); large pages refused (GetLargePageMinimum=2097152, VirtualAlloc error 87 - needs SeLockMemoryPrivilege); using 4 KB pages
strata generate: loaded 39.97 GiB at 0.98 GiB/s
strata mtp: draft layer loaded, 949 MiB of VRAM (experts 675, dense 111)
strata generate: profile E:\Strata\data\expert-profile.bin: 24576 ranked pairs, built for 24576 slots
strata generate: expert cache auto: 8.55 GiB free, 700 MiB reserved -> 3624 slots
strata generate: expert cache 4845 slots, 7.86 GiB of VRAM; policy is
strata generate: *** WARNING: --expert-cache is enabled and the GPU hit path is NOT
                 CORRECT. The generated tokens diverge from a cache-off run (measured:
                 first difference at token 40 at 2.97% hits, token 0 at 54.4%). Any
                 timing from this run is real; any OUTPUT from it is not. ***
                 PROFILE, ranked by routing frequency, no eviction.
strata generate: pre-filled 4845 of 4845 slots from the profile; slot 0 verified
strata generate: R4 hit path ON - resident experts are computed on the GPU
strata generate: 23 expert-pool workers + the host thread
strata generate: session is up; locating the head
strata generate: experimental native Q5_K head, 437043200 bytes
strata generate: token graph hit path: 4845 resident experts, decided on the device
strata serve: the prompt path borrows 962 cache slots (1.53 GiB)
strata verify: window up to 4 tokens, 58.3 MiB of device buffers
strata mtp: draft head over 40525 tokens (68.0 MiB)
strata serve: 136 MiB of VRAM free with everything loaded
strata generate: this CPU has no AVX-512: the expert kernels run on AVX2
strata generate: native pack: E:\Strata-data\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding type 21 in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata-data\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: PLE on, table 320001536 rows of E:\Strata-data\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata generate: PCIe probe: 23.9 GB/s host->device -> pcie_frac 0.55 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: E:\Strata-data\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding IQ3_S in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata-data\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: the SSD is kept awake while rows are read: one page of the table after 100 ms without a read, until 60 s after the last request (STRATA_SSD_KEEPALIVE=0 turns it off)
strata generate: profile E:\Strata\data\expert-profile.bin: 24576 ranked pairs, built for 24576 slots
strata generate: PLE on, table 320001536 rows of E:\Strata-data\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata mtp: draft layer loaded, 950 MiB of VRAM (experts 675, dense 111), files read in 0.74 s (1062 MiB/s)
strata generate: GPU 0: NVIDIA GeForce RTX 4090 Laptop GPU, compute capability 8.9
strata generate: expert arena: VirtualLock stopped at 0 of 9477 MiB (error 87); cudaHostRegister of the whole arena FAILED (out of memory); 38 slices pinned (30 GiB); large pages (2097152 B)
strata generate: loaded 39.97 GiB at 1.37 GiB/s
strata generate: native head upload: out of memory
strata generate: PCIe probe: 24.0 GB/s host->device -> pcie_frac 0.55 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: E:\Strata-data\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding IQ3_S in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata-data\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: the SSD is kept awake while rows are read: one page of the table after 100 ms without a read, until 60 s after the last request (STRATA_SSD_KEEPALIVE=0 turns it off)
strata generate: profile E:\Strata\data\expert-profile.bin: 24576 ranked pairs, built for 24576 slots
strata generate: KV streaming: 32768 of 131072 cells per QSA layer in VRAM, the K/V in 1.55 GiB of pinned RAM
strata generate: PLE on, table 320001536 rows of E:\Strata-data\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata mtp: draft layer loaded, 835 MiB of VRAM (experts 675, dense 111), files read in 0.71 s (1102 MiB/s)
strata generate: GPU 0: NVIDIA GeForce RTX 4090 Laptop GPU, compute capability 8.9
strata generate: expert arena: VirtualLock stopped at 0 of 11452 MiB (error 87); cudaHostRegister of the whole arena FAILED (out of memory); 36 slices pinned (28 GiB); large pages (2097152 B)
strata generate: loaded 39.97 GiB at 1.37 GiB/s
strata generate: native head upload: out of memory
strata generate: PCIe probe: 23.9 GB/s host->device -> pcie_frac 0.55 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: E:\Strata-data\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding IQ3_S in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata-data\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: the SSD is kept awake while rows are read: one page of the table after 100 ms without a read, until 60 s after the last request (STRATA_SSD_KEEPALIVE=0 turns it off)
strata generate: profile E:\Strata\data\expert-profile.bin: 24576 ranked pairs, built for 24576 slots
strata generate: KV streaming: 32768 of 131072 cells per QSA layer in VRAM, the K/V in 1.55 GiB of pinned RAM
strata generate: PLE on, table 320001536 rows of E:\Strata-data\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata mtp: draft layer loaded, 835 MiB of VRAM (experts 675, dense 111), files read in 0.71 s (1105 MiB/s)
strata generate: GPU 0: NVIDIA GeForce RTX 4090 Laptop GPU, compute capability 8.9
strata generate: expert arena: VirtualLock stopped at 0 of 11452 MiB (error 87); cudaHostRegister of the whole arena FAILED (out of memory); 36 slices pinned (28 GiB); large pages (2097152 B)
strata generate: loaded 39.97 GiB at 1.38 GiB/s
strata generate: experimental native Q5_K head, 437043200 bytes
strata generate: expert cache auto: 9.36 GiB free, 700 MiB reserved (+184 MiB for the draft head) -> 3915 slots
strata generate: ExpertCache: cudaMalloc(8.49 GiB) failed: out of memory (Windows has 7.4 GiB of commit left: RAM + page file); trying a smaller expert cache: 3906 slots
strata generate: expert cache: 3906 slots (6.37 GiB) after 1 smaller tries - a bigger page file lets it use more of the free VRAM
strata generate: expert cache 3906 slots, 6.37 GiB of VRAM; policy is
strata generate: the GPU computes the experts in the cache; it rounds differently from the CPU,
                 so a reply can differ slightly from a run without the cache (same quality:
                 bench/results/2026-09-27-cache-parity).
                 PROFILE, ranked by routing frequency, no eviction.
strata generate: pre-filled 3906 of 3906 slots from the profile; slot 0 verified
strata generate: R4 hit path ON - resident experts are computed on the GPU
strata generate: 23 expert-pool workers + the host thread
strata generate: session is up (engine 0.1.36)
strata generate: token graph hit path: 3906 resident experts, decided on the device
strata serve: prompt chunk auto: 8192 tokens
strata serve: the prompt path borrows 2351 CUDA0 cache slots (3.81 GiB)
strata serve: prefill: host buffers or events for a chunk of 8192 tokens
strata generate: PCIe probe: 23.8 GB/s host->device -> pcie_frac 0.55 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: E:\Strata-data\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding IQ3_S in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata-data\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: the SSD is kept awake while rows are read: one page of the table after 100 ms without a read, until 60 s after the last request (STRATA_SSD_KEEPALIVE=0 turns it off)
strata generate: profile E:\Strata\data\expert-profile.bin: 24576 ranked pairs, built for 24576 slots
strata generate: KV streaming: 32768 of 131072 cells per QSA layer in VRAM, the K/V in 1.55 GiB of pinned RAM
strata generate: PLE on, table 320001536 rows of E:\Strata-data\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata mtp: draft layer loaded, 835 MiB of VRAM (experts 675, dense 111), files read in 0.76 s (1038 MiB/s)
strata generate: GPU 0: NVIDIA GeForce RTX 4090 Laptop GPU, compute capability 8.9
strata generate: expert arena: VirtualLock stopped at 0 of 11452 MiB (error 87); cudaHostRegister of the whole arena FAILED (out of memory); 36 slices pinned (28 GiB); large pages (2097152 B)
strata generate: loaded 39.97 GiB at 1.38 GiB/s
strata generate: experimental native Q5_K head, 437043200 bytes
strata generate: expert cache auto: 9.36 GiB free, 700 MiB reserved (+184 MiB for the draft head) -> 3915 slots
strata generate: expert cache 5235 slots, 8.49 GiB of VRAM; policy is
strata generate: the GPU computes the experts in the cache; it rounds differently from the CPU,
                 so a reply can differ slightly from a run without the cache (same quality:
                 bench/results/2026-09-27-cache-parity).
                 PROFILE, ranked by routing frequency, no eviction.
strata generate: pre-filled 5235 of 5235 slots from the profile; slot 0 verified
strata generate: R4 hit path ON - resident experts are computed on the GPU
strata generate: 23 expert-pool workers + the host thread
strata generate: session is up (engine 0.1.36)
strata generate: token graph hit path: 5235 resident experts, decided on the device
strata serve: prompt chunk auto: 8192 tokens
strata serve: the prompt path borrows 2374 CUDA0 cache slots (3.81 GiB)
strata hc: CUDA0: the hyper-connection read runs as staged (the norm per token and stream, the down projection's activations staged ahead by cp.async); checked bit for bit against the plain read on this card (STRATA_HC_SPLIT=1 or 0 for the earlier ones)
strata verify: window up to 6 tokens, 69.6 MiB of device buffers
strata mtp: draft head over 106299 tokens (178.4 MiB)
strata serve: 444 MiB of VRAM free with everything loaded
strata verify: captured the 6-token window (upload no error, sync no error)
strata verify: captured the 3-token window (upload no error, sync no error)
strata generate: PCIe probe: 23.3 GB/s host->device -> pcie_frac 0.55 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: E:\Strata-data\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding IQ3_S in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata-data\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: the SSD is kept awake while rows are read: one page of the table after 100 ms without a read, until 60 s after the last request (STRATA_SSD_KEEPALIVE=0 turns it off)
strata generate: profile E:\Strata\data\expert-profile.bin: 24576 ranked pairs, built for 24576 slots
strata generate: KV streaming: 32768 of 131072 cells per QSA layer in VRAM, the K/V in 1.55 GiB of pinned RAM
strata generate: PLE on, table 320001536 rows of E:\Strata-data\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata mtp: draft layer loaded, 835 MiB of VRAM (experts 675, dense 111), files read in 0.71 s (1099 MiB/s)
strata generate: GPU 0: NVIDIA GeForce RTX 4090 Laptop GPU, compute capability 8.9
strata generate: expert arena: VirtualLock stopped at 0 of 11452 MiB (error 87); cudaHostRegister of the whole arena FAILED (out of memory); 36 slices pinned (28 GiB); large pages (2097152 B)
strata generate: loaded 39.97 GiB at 1.38 GiB/s
strata generate: experimental native Q5_K head, 437043200 bytes
strata generate: expert cache auto: 9.36 GiB free, 700 MiB reserved (+184 MiB for the draft head) -> 3915 slots
strata generate: expert cache 5235 slots, 8.49 GiB of VRAM; policy is
strata generate: the GPU computes the experts in the cache; it rounds differently from the CPU,
                 so a reply can differ slightly from a run without the cache (same quality:
                 bench/results/2026-09-27-cache-parity).
                 PROFILE, ranked by routing frequency, no eviction.
strata generate: pre-filled 5235 of 5235 slots from the profile; slot 0 verified
strata generate: R4 hit path ON - resident experts are computed on the GPU
strata generate: 23 expert-pool workers + the host thread
strata generate: session is up (engine 0.1.36)
strata generate: token graph hit path: 5235 resident experts, decided on the device
strata serve: prompt chunk auto: 8192 tokens
strata serve: the prompt path borrows 2374 CUDA0 cache slots (3.81 GiB)
strata hc: CUDA0: the hyper-connection read runs as staged (the norm per token and stream, the down projection's activations staged ahead by cp.async); checked bit for bit against the plain read on this card (STRATA_HC_SPLIT=1 or 0 for the earlier ones)
strata verify: window up to 6 tokens, 69.6 MiB of device buffers
strata mtp: draft head over 106299 tokens (178.4 MiB)
strata serve: 444 MiB of VRAM free with everything loaded
strata generate: PCIe probe: 23.9 GB/s host->device -> pcie_frac 0.55 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: E:\Strata-data\packs\swift-iq3_xxs experts (largest blob 2.33 MB), token embedding IQ3_S in mapped host memory (260 MiB)
strata generate: 1466 MiB of weights loaded from E:\Strata-data\packs\swift-iq3_xxs (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1781.11 MiB of weights
strata generate: the SSD is kept awake while rows are read: one page of the table after 100 ms without a read, until 60 s after the last request (STRATA_SSD_KEEPALIVE=0 turns it off)
strata generate: profile E:\Strata\data\expert-profile.bin: 24576 ranked pairs, built for 24576 slots
strata generate: KV streaming: 32768 of 131072 cells per QSA layer in VRAM, the K/V in 1.55 GiB of pinned RAM
strata generate: PLE on, table 320001536 rows of E:\Strata-data\models\swift-IQ3_XXS\Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf
strata mtp: draft layer loaded, 835 MiB of VRAM (experts 675, dense 111), files read in 0.76 s (1033 MiB/s)
strata generate: GPU 0: NVIDIA GeForce RTX 4090 Laptop GPU, compute capability 8.9
strata generate: expert arena: VirtualLock stopped at 0 of 11452 MiB (error 87); cudaHostRegister of the whole arena FAILED (out of memory); 36 slices pinned (28 GiB); large pages (2097152 B)
strata generate: loaded 39.97 GiB at 1.39 GiB/s
strata generate: experimental native Q5_K head, 437043200 bytes
strata generate: expert cache auto: 9.36 GiB free, 700 MiB reserved (+184 MiB for the draft head) -> 3915 slots
strata generate: expert cache 5235 slots, 8.49 GiB of VRAM; policy is
strata generate: the GPU computes the experts in the cache; it rounds differently from the CPU,
                 so a reply can differ slightly from a run without the cache (same quality:
                 bench/results/2026-09-27-cache-parity).
                 PROFILE, ranked by routing frequency, no eviction.
strata generate: pre-filled 5235 of 5235 slots from the profile; slot 0 verified
strata generate: R4 hit path ON - resident experts are computed on the GPU
strata generate: 23 expert-pool workers + the host thread
strata generate: session is up (engine 0.1.36)
strata generate: token graph hit path: 5235 resident experts, decided on the device
strata serve: prompt chunk auto: 8192 tokens
strata serve: the prompt path borrows 2374 CUDA0 cache slots (3.81 GiB)
strata serve: prefill: host buffers or events for a chunk of 8192 tokens

using https://github.com/Niko1221/Strata/releases/tag/v0.1.36

need to make it easier to rollback to a previous version + engine downgrade, not much good if newer stuff is breaking what was working before :(

After a restart it seems to be getting further, I think previously I had run the earlier engine then did the git pull and closed the terminal, maybe left something in memory, anyway it did more tests now, 50tks about right, but it seems totally broken when running properly and can't even run firefox without it crashing like it has no memory left for it despite 50gb/64gb

so compared to the earlier version I did not have these problems or with firefox running

Thanks. The start failure after the update fits what you found: an engine from before the update was probably still holding memory, and after the restart it starts.

About memory: Swift IQ3_XXS keeps ~40 GB of experts in RAM, which leaves about 24 GB on a 64 GB laptop for Windows and everything else. Things that help:

START-HERE.bat --setup --low-ram on: the experts the GPU already holds aren't copied into RAM too, which saves about what the GPU holds (it can be a little slower);
0.1.38 reads the experts past the Windows file cache when RAM is short (Start: read the expert arena unbuffered when the file cache cannot help (Windows; part 1 of #285) #357 File tier: read the GGUF in place unbuffered when the file cache cannot keep it beside the RAM budget (Windows; #286 ported into the existing tier) #362), so less of it fills up.

About the VRAM: 8.49 GB wasn't a cap. It's what was left for the expert cache after the K/V cache, the prompt buffers and the draft layer. Since 0.1.37, START-HERE.bat --vram-reserve-mib N keeps N MiB free for other programs.

An easier way back to an earlier version is a fair request; noted. If Firefox still crashes on 0.1.38, please post the engine log and how much RAM is free while the model runs.

Thanks for the details. A stale old engine process held the memory and it ran fine after a reboot, so this is not a bug in the update itself. docs/TROUBLESHOOTING.md has the rollback and low-RAM notes. Closing it; please open a new issue with the start log if it comes back.

Navigation Menu

Repository navigation

new error `raise RuntimeError("the engine exited before it was ready" + (f" (see {log})" if log else "") +` #517

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

new error raise RuntimeError("the engine exited before it was ready" + (f" (see {log})" if log else "") + #517

Description

Activity

XeonG commented on Oct 2, 2026

XeonG commented on Oct 2, 2026

XeonG commented on Oct 2, 2026

XeonG commented on Oct 2, 2026

Niko1221 commented on Oct 3, 2026

Niko1221 commented on Oct 7, 2026

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions

new error `raise RuntimeError("the engine exited before it was ready" + (f" (see {log})" if log else "") +` #517