[GH-ISSUE #5621] [BUG]: Query: LanceDB as Vector Database incompatible with Mac Silicon due to lack of AVX2 support? #5198

Closed
opened 2026-06-05 14:52:35 -04:00 by yindo · 1 comment
Owner

Originally created by @dm5n on GitHub (May 12, 2026).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5621

How are you running AnythingLLM?

AnythingLLM desktop app

What happened?

Persistent Error msg shortly after seemingly successful embedding. System Log Crash Report:


Translated Report (Full Report Below)

Process: AnythingLLM Helper [66657]
Path: /Applications/AnythingLLM.app/Contents/Frameworks/AnythingLLM Helper.app/Contents/MacOS/AnythingLLM Helper
Identifier: com.anythingllm.helper
Version: 1.12.1
Code Type: ARM-64 (Native)
Role: Foreground
Parent Process: AnythingLLM [66634]
Coalition: com.anythingllm [5447]
Responsible Process: AnythingLLM [66634]

Date/Time: 2026-05-12 19:13:56.9193 +0700
Launch Time: 2026-05-12 19:13:45.2383 +0700
Hardware Model: Mac16,7
OS Version: macOS 26.4.1 (25E253)
Release Type: User

System Integrity Protection: enabled

Triggered by Thread: 0 node.CrUtilityMain, Dispatch Queue: com.apple.main-thread

Exception Type: EXC_BREAKPOINT (SIGTRAP)
Exception Codes: 0x0000000000000001, 0x000000010cc87250

Termination Reason: Namespace SIGNAL, Code 5, Trace/BPT trap: 5
Terminating Process: exc handler [66657]

Thread 0 Crashed:: node.CrUtilityMain Dispatch queue: com.apple.main-thread
0 Electron Framework 0x10cc87250 std::process::id::hb9fbed0f3f7b0c03 + 290704
1 Electron Framework 0x10cc87268 std::process::id::hb9fbed0f3f7b0c03 + 290728
2 Electron Framework 0x10cc87284 std::process::id::hb9fbed0f3f7b0c03 + 290756
3 Electron Framework 0x10cc89f5c std::process::id::hb9fbed0f3f7b0c03 + 302236
4 Electron Framework 0x10cc89950 std::process::id::hb9fbed0f3f7b0c03 + 300688
5 Electron Framework 0x10cc99360 std::process::id::hb9fbed0f3f7b0c03 + 364704
6 Electron Framework 0x10cc9f610 std::process::id::hb9fbed0f3f7b0c03 + 389968
7 libsystem_malloc.dylib 0x1855d8874 _malloc_zone_memalign + 300
8 libsystem_malloc.dylib 0x1855d8f5c _posix_memalign + 52
9 libonnxruntime.1.14.0.dylib 0x169afaddc onnxruntime::CPUAllocator::Alloc(unsigned long) + 60
10 libonnxruntime.1.14.0.dylib 0x169afe8e8 onnxruntime::BFCArena::Extend(unsigned long)::$_0::operator()(unsigned long) const + 60
11 libonnxruntime.1.14.0.dylib 0x169afe148 onnxruntime::BFCArena::Extend(unsigned long) + 300
12 libonnxruntime.1.14.0.dylib 0x169aff484 onnxruntime::BFCArena::AllocateRawInternal(unsigned long, bool, onnxruntime::Stream*, bool, std::__1::function<void (onnxruntime::Stream&, onnxruntime::synchronize::Notification&)>) + 916
13 libonnxruntime.1.14.0.dylib 0x169aff064 onnxruntime::BFCArena::Alloc(unsigned long) + 60
14 libonnxruntime.1.14.0.dylib 0x169b97f20 onnxruntime::Tensor::Tensor(onnxruntime::DataTypeImpl const*, onnxruntime::TensorShape const&, std::__1::shared_ptronnxruntime::IAllocator, gsl::span<long long const, 18446744073709551615ul>) + 148
15 libonnxruntime.1.14.0.dylib 0x169b9849c onnxruntime::Tensor::InitOrtValue(onnxruntime::DataTypeImpl const*, onnxruntime::TensorShape const&, std::__1::shared_ptronnxruntime::IAllocator, OrtValue&, gsl::span<long long const, 18446744073709551615ul>) + 116
16 libonnxruntime.1.14.0.dylib 0x169b2e6d8 onnxruntime::ExecutionFrame::AllocateMLValueTensorSelfOwnBufferHelper(OrtValue&, int, onnxruntime::DataTypeImpl const*, OrtMemoryInfo const&, onnxruntime::TensorShape const&) + 644
17 libonnxruntime.1.14.0.dylib 0x169b2fa94 onnxruntime::ExecutionFrame::AllocateAsPerAllocationPlan(OrtValue&, int, onnxruntime::TensorShape const*) + 708
18 libonnxruntime.1.14.0.dylib 0x169b2b9f4 onnxruntime::IExecutionFrame::GetOrCreateNodeOutputMLValue(int, int, onnxruntime::TensorShape const*, OrtValue*&, onnxruntime::Node const&) + 464
19 libonnxruntime.1.14.0.dylib 0x169b56bc0 onnxruntime::OpKernelContext::OutputMLValue(int, onnxruntime::TensorShape const&) + 132
20 libonnxruntime.1.14.0.dylib 0x169b568cc onnxruntime::OpKernelContext::Output(int, onnxruntime::TensorShape const&) + 20
21 libonnxruntime.1.14.0.dylib 0x16977e0dc onnxruntime::MatMul::Compute(onnxruntime::OpKernelContext*) const + 444

[... Threads 1-101 omitted for brevity ...]

Binary Images:
0x10036c000 - 0x10037ffff com.anythingllm.helper () /Applications/AnythingLLM.app/Contents/Frameworks/AnythingLLM Helper.app/Contents/MacOS/AnythingLLM Helper
0x141800000 - 0x142603fff libquery_engine-darwin-arm64.dylib.node (
) /Users//Library/Application Support/anythingllm-desktop//libquery_engine-darwin-arm64.dylib.node
0x340000000 - 0x3437cbfff lancedb.darwin-arm64.node (
) /Applications/AnythingLLM.app/Contents/Resources/backend/node_modules/@lancedb/lancedb-darwin-arm64/lancedb.darwin-arm64.node
0x169434000 - 0x16a0c7fff libonnxruntime.1.14.0.dylib (*) /Applications/AnythingLLM.app/Contents/Resources/backend/node_modules/onnxruntime-node/bin/napi-v3/darwin/arm64/libonnxruntime.1.14.0.dylib

Are there known steps to reproduce?

I believe the issue is using LanceDB on Mac Silicon because of incompatibility with AVX2 as also described in issue #1331

if this is the case, may I suggest that the default not be set as LanceDB for mac users or at the least - the Install and Setup Documentation indicate a different Vector DB required for mac users. In my case, having to start over vectorising 1000+ documents was PITA. I'm trying Qdrant as alternative.

LLM Provider & Model (if applicable)

OpenRouter: inclusionAl: Ring-2.6-1T

Embedder Provider & Model (if applicable)

Ollama - bge-m3:latest

Originally created by @dm5n on GitHub (May 12, 2026). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5621 ### How are you running AnythingLLM? AnythingLLM desktop app ### What happened? Persistent Error msg shortly after seemingly successful embedding. System Log Crash Report: ------------------------------------- Translated Report (Full Report Below) ------------------------------------- Process: AnythingLLM Helper [66657] Path: /Applications/AnythingLLM.app/Contents/Frameworks/AnythingLLM Helper.app/Contents/MacOS/AnythingLLM Helper Identifier: com.anythingllm.helper Version: 1.12.1 Code Type: ARM-64 (Native) Role: Foreground Parent Process: AnythingLLM [66634] Coalition: com.anythingllm [5447] Responsible Process: AnythingLLM [66634] Date/Time: 2026-05-12 19:13:56.9193 +0700 Launch Time: 2026-05-12 19:13:45.2383 +0700 Hardware Model: Mac16,7 OS Version: macOS 26.4.1 (25E253) Release Type: User System Integrity Protection: enabled Triggered by Thread: 0 node.CrUtilityMain, Dispatch Queue: com.apple.main-thread Exception Type: EXC_BREAKPOINT (SIGTRAP) Exception Codes: 0x0000000000000001, 0x000000010cc87250 Termination Reason: Namespace SIGNAL, Code 5, Trace/BPT trap: 5 Terminating Process: exc handler [66657] Thread 0 Crashed:: node.CrUtilityMain Dispatch queue: com.apple.main-thread 0 Electron Framework 0x10cc87250 std::process::id::hb9fbed0f3f7b0c03 + 290704 1 Electron Framework 0x10cc87268 std::process::id::hb9fbed0f3f7b0c03 + 290728 2 Electron Framework 0x10cc87284 std::process::id::hb9fbed0f3f7b0c03 + 290756 3 Electron Framework 0x10cc89f5c std::process::id::hb9fbed0f3f7b0c03 + 302236 4 Electron Framework 0x10cc89950 std::process::id::hb9fbed0f3f7b0c03 + 300688 5 Electron Framework 0x10cc99360 std::process::id::hb9fbed0f3f7b0c03 + 364704 6 Electron Framework 0x10cc9f610 std::process::id::hb9fbed0f3f7b0c03 + 389968 7 libsystem_malloc.dylib 0x1855d8874 _malloc_zone_memalign + 300 8 libsystem_malloc.dylib 0x1855d8f5c _posix_memalign + 52 9 libonnxruntime.1.14.0.dylib 0x169afaddc onnxruntime::CPUAllocator::Alloc(unsigned long) + 60 10 libonnxruntime.1.14.0.dylib 0x169afe8e8 onnxruntime::BFCArena::Extend(unsigned long)::$_0::operator()(unsigned long) const + 60 11 libonnxruntime.1.14.0.dylib 0x169afe148 onnxruntime::BFCArena::Extend(unsigned long) + 300 12 libonnxruntime.1.14.0.dylib 0x169aff484 onnxruntime::BFCArena::AllocateRawInternal(unsigned long, bool, onnxruntime::Stream*, bool, std::__1::function<void (onnxruntime::Stream&, onnxruntime::synchronize::Notification&)>) + 916 13 libonnxruntime.1.14.0.dylib 0x169aff064 onnxruntime::BFCArena::Alloc(unsigned long) + 60 14 libonnxruntime.1.14.0.dylib 0x169b97f20 onnxruntime::Tensor::Tensor(onnxruntime::DataTypeImpl const*, onnxruntime::TensorShape const&, std::__1::shared_ptr<onnxruntime::IAllocator>, gsl::span<long long const, 18446744073709551615ul>) + 148 15 libonnxruntime.1.14.0.dylib 0x169b9849c onnxruntime::Tensor::InitOrtValue(onnxruntime::DataTypeImpl const*, onnxruntime::TensorShape const&, std::__1::shared_ptr<onnxruntime::IAllocator>, OrtValue&, gsl::span<long long const, 18446744073709551615ul>) + 116 16 libonnxruntime.1.14.0.dylib 0x169b2e6d8 onnxruntime::ExecutionFrame::AllocateMLValueTensorSelfOwnBufferHelper(OrtValue&, int, onnxruntime::DataTypeImpl const*, OrtMemoryInfo const&, onnxruntime::TensorShape const&) + 644 17 libonnxruntime.1.14.0.dylib 0x169b2fa94 onnxruntime::ExecutionFrame::AllocateAsPerAllocationPlan(OrtValue&, int, onnxruntime::TensorShape const*) + 708 18 libonnxruntime.1.14.0.dylib 0x169b2b9f4 onnxruntime::IExecutionFrame::GetOrCreateNodeOutputMLValue(int, int, onnxruntime::TensorShape const*, OrtValue*&, onnxruntime::Node const&) + 464 19 libonnxruntime.1.14.0.dylib 0x169b56bc0 onnxruntime::OpKernelContext::OutputMLValue(int, onnxruntime::TensorShape const&) + 132 20 libonnxruntime.1.14.0.dylib 0x169b568cc onnxruntime::OpKernelContext::Output(int, onnxruntime::TensorShape const&) + 20 21 libonnxruntime.1.14.0.dylib 0x16977e0dc onnxruntime::MatMul<float>::Compute(onnxruntime::OpKernelContext*) const + 444 [... Threads 1-101 omitted for brevity ...] Binary Images: 0x10036c000 - 0x10037ffff com.anythingllm.helper (*) /Applications/AnythingLLM.app/Contents/Frameworks/AnythingLLM Helper.app/Contents/MacOS/AnythingLLM Helper 0x141800000 - 0x142603fff libquery_engine-darwin-arm64.dylib.node (*) /Users/<REDACTED>/Library/Application Support/anythingllm-desktop/*/libquery_engine-darwin-arm64.dylib.node 0x340000000 - 0x3437cbfff lancedb.darwin-arm64.node (*) /Applications/AnythingLLM.app/Contents/Resources/backend/node_modules/@lancedb/lancedb-darwin-arm64/lancedb.darwin-arm64.node 0x169434000 - 0x16a0c7fff libonnxruntime.1.14.0.dylib (*) /Applications/AnythingLLM.app/Contents/Resources/backend/node_modules/onnxruntime-node/bin/napi-v3/darwin/arm64/libonnxruntime.1.14.0.dylib ### Are there known steps to reproduce? I believe the issue is using LanceDB on Mac Silicon because of incompatibility with AVX2 as also described in issue [#1331](https://github.com/Mintplex-Labs/anything-llm/issues/1331) if this is the case, may I suggest that the default not be set as LanceDB for mac users or at the least - the Install and Setup Documentation indicate a different Vector DB required for mac users. In my case, having to start over vectorising 1000+ documents was PITA. I'm trying Qdrant as alternative. ### LLM Provider & Model (if applicable) OpenRouter: inclusionAl: Ring-2.6-1T ### Embedder Provider & Model (if applicable) Ollama - bge-m3:latest
yindo added the possible bugneeds info / can't replicate labels 2026-06-05 14:52:35 -04:00
yindo closed this issue 2026-06-05 14:52:35 -04:00
Author
Owner

@timothycarambat commented on GitHub (May 13, 2026):

The whole team is on mac silicon devices (M1-M4). Can you tell me more about your setup because if this was the case it would be endemic. This cannot be related to AVX2 - that's x86-only and doesn't apply to ARM Macs.

The crash is in ONNX Runtime during matrix operations, not in LanceDB itself (so could be during vectorization of strings?)

To me, this seems like reranking, not embedding - can you tell me more about your system hardware?

<!-- gh-comment-id:4445777079 --> @timothycarambat commented on GitHub (May 13, 2026): The whole team is on mac silicon devices (M1-M4). Can you tell me more about your setup because if this was the case it would be endemic. This cannot be related to AVX2 - that's x86-only and doesn't apply to ARM Macs. The crash is in ONNX Runtime during matrix operations, not in LanceDB itself (so could be during vectorization of strings?) - We use a default embedder with onnxruntime (which you have for ollama) - We also use onnxruntime for reranking - if you have that enabled? - This comes in two forms: [Search preference](https://docs.anythingllm.com/chatting-with-documents/introduction#vector-database-settings--search-preference-reranking) & [Intelligent Tool Selection](https://docs.anythingllm.com/agent/intelligent-tool-selection) To me, this seems like reranking, not embedding - can you tell me more about your system hardware?
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#5198