url/colibri
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

glm.c: remove the LMHEAD_EXACT probe accidentally merged with #152 — functional no-op strip (#197)

Browse Source 
#152 carried leftover debug scaffolding from the lm_head measurement posted on that
PR: a `g_lmhead_exact` global, an undocumented `LMHEAD_EXACT` getenv, and 7 lm_head
call sites routed through matmul_qt_ex(..., !g_lmhead_exact). None of it was in the
PR description; it rode in because `git checkout -B` carries uncommitted working-tree
edits forward.

It should not stay:
  - it does nothing worth having. Pinning lm_head off IDOT measures -0.03% perplexity
    across 5 corpora (in-distribution, isolated) — nothing.
  - LMHEAD_EXACT=1 would silently change lm_head numerics, undocumented and untested.
  - lm_head is fmt=1 (int8), so the IDOT gate applies to it unconditionally on every
    platform anyway; there is no threshold story here to expose.

The flag defaulted to 0 and `!g_lmhead_exact == 1 == matmul_qt`'s own allow_idot, so
this removal is a pure no-op. Verified rather than assumed — summed log-lik over 1023
tokens, in-distribution, CPU (deterministic), dev-with-probe vs dev-minus-probe:

  prose     -2295.245383  ==  -2295.245383
  markdown  -3272.146403  ==  -3272.146403

Bit-identical. make check green (6 C suites + 60 python tests), no new warnings.

The actual #152 changes are untouched: the three attention input projections and the
DSA indexer's ix_wk stay batched and pinned to the exact int4 kernel via
matmul_qt_ex(..., 0).
This commit is contained in:
Dennis Paul
2026-07-14 18:01:04 +02:00
committed by GitHub
parent f1fa5bf3c8
commit 4a0435cbb3
1 changed files with 7 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
c/glm.c
+7 -9
View File
@@ -477,7 +477,6 @@ static void matmul_i2(float *y, const float *x, const uint8_t *q2, const float *
#else
#define IDOT_KERNEL "scalar"
#endif
static int g_lmhead_exact=0; /* PROBE: LMHEAD_EXACT=1 -> pin lm_head to the exact int8 kernel */
static int g_idot=1;
#if defined(__ARM_NEON) && defined(__ARM_FEATURE_DOTPROD)
static int g_i4s=1; /* SDOT presente: int4 IDOT conviene anche a S=1 (decode). Misurato
@@ -2537,7 +2536,7 @@ static float *step(Model *m, const int *ids, int S, int pos_base){
if(m->has_mtp && S>=2 && g_draft>0) mtp_absorb(m, ids+1, x, S-1, pos_base);
float *last=falloc(D); rmsnorm(last, x+(int64_t)(S-1)*D, m->final_norm, D, c->eps);
double th0=now_s();
float *logit=falloc(c->vocab); matmul_qt_ex(logit,last,&m->lm_head,1,!g_lmhead_exact);
float *logit=falloc(c->vocab); matmul_qt(logit,last,&m->lm_head,1);
m->t_head += now_s()-th0;
free(x); free(last); return logit;
}
@@ -2552,7 +2551,7 @@ static float *step_all(Model *m, const int *ids, int S, int pos_base){
if(m->hlast) memcpy(m->hlast, x+(int64_t)(S-1)*D, D*sizeof(float));
float *lo=falloc((int64_t)S*c->vocab), *row=falloc(D);
for(int s=0;s<S;s++){ rmsnorm(row, x+(int64_t)s*D, m->final_norm, D, c->eps);
matmul_qt_ex(lo+(int64_t)s*c->vocab, row, &m->lm_head, 1, !g_lmhead_exact); }
matmul_qt(lo+(int64_t)s*c->vocab, row, &m->lm_head, 1); }
free(x); free(row); return lo;
}
@@ -2586,7 +2585,7 @@ static float *step_decode_batch(Model *m, const DecodeRow *rows, int S){
rmsnorm(norm+(int64_t)s*D,x+(int64_t)s*D,m->final_norm,D,c->eps);
double th0=now_s();
float *logit=falloc((int64_t)S*c->vocab);
matmul_qt_ex(logit,norm,&m->lm_head,S,!g_lmhead_exact);
matmul_qt(logit,norm,&m->lm_head,S);
m->t_head+=now_s()-th0;
free(x); free(norm);
return logit;
@@ -2635,12 +2634,12 @@ static int mtp_draft(Model *m, int next_tok, int kv, int G, int *draft){
double n_eh=0; for(int d=0;d<D;d++) n_eh+=hx[d]*hx[d];
int dbg = getenv("MTP_DEBUG") && atoi(getenv("MTP_DEBUG"))>=2;
int t_pre=-1;
if(dbg){ rmsnorm(row, hx, m->mtp_norm, D, c->eps); matmul_qt_ex(logit, row, &m->lm_head, 1, !g_lmhead_exact);
if(dbg){ rmsnorm(row, hx, m->mtp_norm, D, c->eps); matmul_qt(logit, row, &m->lm_head, 1);
t_pre=mtp_argmax(logit, c->vocab); }
layer_forward(m, &m->mtpL, li, hx, 1, pos, nrm, tmp);
double n_post=0; for(int d=0;d<D;d++) n_post+=hx[d]*hx[d];
rmsnorm(row, hx, m->mtp_norm, D, c->eps);
matmul_qt_ex(logit, row, &m->lm_head, 1, !g_lmhead_exact);
matmul_qt(logit, row, &m->lm_head, 1);
int t2=mtp_argmax(logit, c->vocab);
if(dbg) fprintf(stderr,"[mtp2] pos=%d in_tok=%d ||eh||=%.1f ||post||=%.1f pre_blk=%d post_blk=%d\n",
pos, tok, sqrt(n_eh), sqrt(n_post), t_pre, t2);
@@ -2905,7 +2904,7 @@ static void forward_all(Model *m, const int *ids, int S, int *pred){
float *row=falloc(D);
for(int s=0;s<S;s++){
rmsnorm(row, x+(int64_t)s*D, m->final_norm, D, c->eps); /* heap row (#183) */
matmul_qt_ex(lo, row, &m->lm_head, 1, !g_lmhead_exact); /* exact-mode knob (#152) */
matmul_qt(lo, row, &m->lm_head, 1);
int best=0; float bv=lo[0]; for(int i=1;i<c->vocab;i++) if(lo[i]>bv){bv=lo[i];best=i;}
pred[s]=best;
}
@@ -2942,7 +2941,7 @@ static void run_score(Model *m, const char *path){
double lp=0; int greedy=1;
for(int pos=ctxlen-1; pos<T-1; pos++){
rmsnorm(row, x+(int64_t)pos*D, m->final_norm, D, c->eps);
matmul_qt_ex(lo,row,&m->lm_head,1,!g_lmhead_exact);
matmul_qt(lo,row,&m->lm_head,1);
int am; lp += logprob_target(lo,c->vocab,ids[pos+1],&am); if(!am) greedy=0;
}
printf("%.6f %d %d\n", lp, contlen, greedy); fflush(stdout);
@@ -4169,7 +4168,6 @@ int main(int argc, char **argv){
g_repin = getenv("REPIN")?atoi(getenv("REPIN")):0; /* RFC: re-pin ogni n token emessi (0=off) / live re-pin every n emitted tokens (0=off) */
g_absorb = getenv("ABSORB")?atoi(getenv("ABSORB")):-1; /* -1 auto: assorbita per S<=4 */
g_dsa_force = getenv("DSA_FORCE")?atoi(getenv("DSA_FORCE")):0;
g_lmhead_exact = getenv("LMHEAD_EXACT")?atoi(getenv("LMHEAD_EXACT")):0;
/* matmul_qt documenta la soglia int4-IDOT come "configurabile con I4S" ma il getenv non
* c'era: la variabile non aveva alcun effetto. I4S=<n> -> IDOT int4 solo per S>=n.
* EN: matmul_qt documents the int4 IDOT threshold as "configurable via I4S", but the