LLM Lab
Instinct MI300X, 192 GB HBM3
Connecting
57 rows
v1.5.0

Analysis / measured comparisons

Quantization analysis

Compare memory and generation behavior by precision using only the measured MI300X benchmark rows.

Empirical only

Measured INT4

0

Source rows

INT4 average VRAM

n/aGB

Measured peak memory

INT4 average speed

n/a

Measured generation rate

Precision categories

0

Across 0 benchmark rows

Peak VRAM by precision

Mean measured peak memory. Different models/backends contribute to each precision group.

Loading benchmark data…

Generation throughput by precision

Mean measured generation throughput; no synthetic points are included.

Loading benchmark data…

Measured precision groups

Aggregates are descriptive, not causal: model sizes and backend mix vary by precision.

Loading benchmark data…
PrecisionRowsModelsMean VRAMMean tok/sComparison

Coverage is limited to observed benchmark combinations. Values are measured aggregates from amd_llm_lab_master.csv.