From 6b704713cc6e23056770b0769c8e1399f4c5c41c Mon Sep 17 00:00:00 2001
From: Yash Raj Pandey <55940078+devYRPauli@users.noreply.github.com>
Date: Fri, 2 Oct 2026 10:37:35 -0400
Subject: [PATCH] ggml-quants : avoid invalid rounding in qkx3 scale search
 (llama/29817)

* ggml-quants : avoid invalid rounding in qkx3 scale search

The imatrix scale search can produce an infinite, NaN, or otherwise out-of-range value when the fitted minimum collapses to the maximum or makes the range extremely small. That value is then passed to nearest_int and can trip its assertion in Debug builds.

Clamp the quantization level to [0, nmax] before rounding so valid in-range values behave the same as before while invalid scale-search results no longer reach nearest_int.

Add regression coverage for degenerate imatrix groups across q2_K, q4_K, q5_K, q4_1, and q5_1.

Fixes #29804.

Assisted-by: Claude Opus 5.5

* tests: print degenerate imatrix quant types
---
 ggml/src/ggml-quants.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/ggml/src/ggml-quants.c b/ggml/src/ggml-quants.c
index 55db802c0..7750a72ce 100644
--- a/ggml/src/ggml-quants.c
+++ b/ggml/src/ggml-quants.c
@@ -1036,8 +1036,9 @@ static float make_qkx3_quants(int n, int nmax, const float * GGML_RESTRICT x, co
         iscale = (rmin + rdelta*is + nmax)/(max - min);
         float sum_l = 0, sum_l2 = 0, sum_xl = 0;
         for (int i = 0; i < n; ++i) {
-            int l = nearest_int(iscale*(x[i] - min));
-            l = MAX(0, MIN(nmax, l));
+            // min is the best fit so far and can be at or near max, so v can be inf, nan or out of range for nearest_int
+            const float v = iscale*(x[i] - min);
+            const int l = v > 0 ? nearest_int(MIN(v, nmax)) : 0;
             Laux[i] = l;
             float w = weights ? weights[i] : x[i]*x[i];
             sum_l  += w*l;