Skip to content

adding compile time flags in getting started #75

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
rajesh33411 wants to merge 1 commit into
base: master
Choose a base branch
from
Open

adding compile time flags in getting started #75

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 16 additions & 0 deletions docs/getting-started.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -332,6 +332,22 @@ Example with debug enabled:
cmake -DUSE_NEON=ON -DND_DEBUG=ON ..
```

**All compile-time flags reference:**

| Flag | Default | Default behaviour | What passing `-DFLAG=ON` changes |
|---|---|---|---|
| `-DDEBUG=ON` | `OFF` | Release build compiled with `-O3 -ffast-math` for maximum performance | Switches to debug build: disables all optimisations (`-O0`), adds full debug symbols (`-g3`). Use when debugging crashes or stepping through code with a debugger. |
| `-DND_DEBUG=ON` | `OFF` | No internal logging — Endee runs silently | Enables Endee's internal debug logging and operation timing output to stdout. Useful for tracing what the server is doing without a full debug build. |
| `-DND_SPARSE_INSTRUMENT=ON` | `OFF` | No timing data collected for sparse index operations | Adds fine-grained timing instrumentation around sparse index operations. Use to profile and diagnose sparse search performance. |
| `-DND_MDBX_INSTRUMENT=ON` | `OFF` | No timing data collected for MDBX (the embedded key-value store) | Adds timing instrumentation around MDBX read/write calls. Use to profile storage layer bottlenecks. |
| `-DNDD_INV_IDX_STORE_FLOATS=ON` | `OFF` | Sparse index stores values as quantized `fp16` — smaller memory footprint | Stores raw `float32` values in the sparse index instead. Higher numerical precision but doubles the memory used by the sparse index. |
| `-DUSE_AVX2=ON` | `OFF` | Must pick exactly one SIMD flag or the build fails | Compiles with AVX2, FMA, F16C instructions. Target: modern Intel/AMD x86_64 desktops and servers. |
| `-DUSE_AVX512=ON` | `OFF` | Must pick exactly one SIMD flag or the build fails | Compiles with AVX512F, BW, VNNI, FP16, VPOPCNTDQ instructions. Target: server-grade Intel Xeon / AMD EPYC. The binary exits at startup if the CPU does not support all required extensions. |
| `-DUSE_NEON=ON` | `OFF` | Must pick exactly one SIMD flag or the build fails | Compiles with NEON, FP16, DotProd. On Apple Silicon uses `-mcpu=native`; on other ARM uses `-march=armv8.2-a+fp16+dotprod`. |
| `-DUSE_SVE2=ON` | `OFF` | Must pick exactly one SIMD flag or the build fails | Compiles with SVE2, FP16, DotProd (`-march=armv8.6-a+sve2+fp16+dotprod`). Target: ARMv9 servers. |

> **Note:** Exactly one SIMD flag must be set — the build fails if none is provided.

### Step 3: Compile

```bash
Expand Down