Skip to content
Run LLM Locally

Gemma 2 27B IT on the Arc A770 16 GB

Gemma 2 27B IT at Q4_K_M needs about 19.7 GB, more than the 16 GB of the Arc A770 16 GB: part of the model runs from system RAM, which slows generation. At Q3_K_M it does not fit in VRAM.

Gemma 2 27B IT

Q4_K_M · 8K context

CPU/GPU offload

The model needs about 19.7 GB but only 15.5 GB of VRAM is usable: 4.1 GB must be loaded into system RAM. It runs, 79% on the GPU, but more slowly.

Your hardware
Arc A770 16 GB · 64 GB RAM
Estimated memory
~19.7 GB estimate
VRAM (usable memory)
15.5 GB
GPU only
No
CPU/GPU offload
Required (4.1 GB)
Generation speed
3.5–7 tok/s estimate
Largest context in fast memory
None
Confidence: low

Based on: memory bandwidth (560 GB/s) and model size · system RAM bandwidth (~51 GB/s)

Memory breakdown

~19.7 GB (18.9 GB–20.5 GB) needed · estimate

4.1 GB must be offloaded to system RAM

  • Model weights15.5 GB
  • KV cache2.2 GB
  • Runtime overhead1.1 GB
  • Safety margin0.9 GB
Assumptions: context 8K tokens · KV cache f16 · llama.cpp (SYCL) · flash attention on · weights = real file size · sliding-window layers keep 4,096 tokens

EstimateData updated 24 days agoSource: Run LLM Locally engine, Hugging Face config.json + model card

What makes this estimate less reliable

  • no benchmark for this hardware and model yet
  • split CPU/GPU runs vary a lot with the CPU and PCIe link
  • system RAM speed unknown (dual-channel DDR4-3200 assumed)

Other quantizations on the Arc A770 16 GB

Gemma 2 27B IT on Arc A770 16 GB + 64 GB RAM · 8K context
Gemma 2 27B IT Q8_08K~31.7 GB · 1.2–2.5 tok/s estimatePossible but slow
Gemma 2 27B IT Q6_K8K~25.2 GB · 1.8–3.5 tok/s estimatePossible but slow
Gemma 2 27B IT Q5_K_M8K~22.4 GB · 2.5–5 tok/s estimateCPU/GPU offload
Gemma 2 27B IT Q4_K_M8K~19.7 GB · 3.5–7 tok/s estimateCPU/GPU offload
Gemma 2 27B IT IQ4_XS8K~17.9 GB · 5.5–11 tok/s estimateCPU/GPU offload
Gemma 2 27B IT Q3_K_M8K~16.5 GB · 8–16 tok/s estimateCPU/GPU offload

Alternatives

Hardware that runs Gemma 2 27B IT Q4_K_M entirely in memory

Prices and links follow the market you choose.

Radeon RX 7900 XTX

AMD

ExcellentRuns Gemma 2 27B IT Q4_K_M entirely in memory

Price not available

Availability unknown

View on amazon.com

Affiliate link

AMD Ryzen AI Max+ 395

Unified memory PC

CompatibleRuns Gemma 2 27B IT Q4_K_M entirely in memory

Price not available

Availability unknown

View on amazon.com

Affiliate link

GeForce RTX 2080 Ti

Lowest memory tier

CompatibleRuns Gemma 2 27B IT Q4_K_M entirely in memory

Price not available

Availability unknown

View on amazon.com

Affiliate link

No current offer imported for this market: the link opens a marketplace search.

Frequently asked questions

Can the Arc A770 16 GB run Gemma 2 27B IT?
The model needs about 19.7 GB but only 15.5 GB of VRAM is usable: 4.1 GB must be loaded into system RAM. It runs, 79% on the GPU, but more slowly.
Which quantization should I use on the Arc A770 16 GB?
None fits entirely in VRAM at 8K context; the smallest, Q3_K_M, still needs offload.
What is the largest context on the Arc A770 16 GB?
At Q4_K_M, the model does not fit entirely in VRAM even at a short context.
How fast is Gemma 2 27B IT (Arc A770 16 GB)?
We estimate 3.5–7 tok/s (estimate, confidence low). Based on: memory bandwidth (560 GB/s) and model size, system RAM bandwidth (~51 GB/s).