AMD · 16 GB GDDR6 · 624 GB/s bandwidth ·
Google · 1.0B · model ctx up to 32K
Yes — the RX 7800 XT 16GB (16 GB GDDR6) can run gemma-3-1b-it. At Q4 with the default 8K context it needs ≈ 1.75 GB VRAM (weights 0.55 GB + KV cache 0.2 GB + 1 GB runtime overhead), which fits within 16 GB. Decode at Q4/8K: ≈ 397.1 tok/s (estimated, batch 1). FP16 (≈ 3.2 GB) also fits, at ≈ 109.2 tok/s (estimated). Estimates come from memory-bandwidth math; rows tagged measured override estimates. Relative ranking is more reliable than absolute tok/s. Methodology.
Fit & speed by quant and context
Quant
Context
VRAM needed
Fits?
tok/s (decode)
Q4
4K
1.65 GB
Yes
397.1est.
Q4
8K default
1.75 GB
Yes
397.1est.
Q4
32K
2.35 GB
Yes
397.1est.
FP16
4K
3.1 GB
Yes
109.2est.
FP16
8K default
3.2 GB
Yes
109.2est.
FP16
32K
3.8 GB
Yes
109.2est.
Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured
anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s
means the model is far beyond this card (offload-only territory).