Home / Multi-GPU

llama.cpp · homogeneous 2–4 GPUs · not a PC builder

Can this model run across multiple GPUs?

Phase 8 answers memory feasibility for identical cards under llama.cpp. Default split is layer (pipeline parallel: each GPU owns a slice of layers; KV stays with those layers). This is not 2 × VRAM = pool, and it is not a complete PC build.

2× NVIDIA RTX 3050 6 GB

Does not fit on this configuration

Model
Gemma 4 31B Instruct · Q4 · 8K
Runtime / split
llama.cpp · layer · calculator 1.0.0
Single-GPU status
DOES_NOT_FIT · required 28.1 GB vs usable 5.4 GB
Per card advertised / usable
6 GB / 5.4 GB
Aggregate advertised
12.0 GB — capacity, not effective model capacity
Nominal aggregate usable
10.8 GB (N × advertised × 0.90)
Effective per-GPU peak
14.4 GB vs usable 5.4 GB
Per-GPU split (approx.)
weights 9.2 GB · KV 3.8 GB · runtime 1.0 GB · safety 0.5 GB
Why
Even a 2K context exceeds per-GPU usable VRAM after partition and per-device reserves.
Topology
Requires space for 2 discrete GPUs. Exact cooler slot width is not in the RigForAI graph. layer split is pipeline parallel and can run over PCIe. KV stays with the layers on each GPU. NVLink is not treated as a 1× aggregate VRAM pool.
Performance
Phase 5 predictions are single-GPU. Multi-GPU tok/s is not published.
GPU-only current cost
$419.98 = 2 × $209.99 at the current per-card matched price. This does not mean 2 units are in stock, and it is not a complete build cost.
GIGABYTE GeForce RTX 3050 WINDFORCE OC V2 6G NVIDIA 6 GB GDDR6 · ASIN B0DQK4XSC7
Amazon
Formula
per_gpu_peak = ceil(weight_bytes/N) + ceil(kv_bytes/N) + runtime_base + safety_base + weight_fractions×ceil(weight_bytes/N). Compare to usable = advertised×0.90. Not (single_gpu_required × N) and not advertised_vram × N as effective capacity.

NVIDIA RTX 3050 6 GB · Single-GPU can-run · Find a single GPU · Build a machine for this GPU setup

Query combinations are not indexed. Source: llama.cpp multi-GPU documentation.