AMD Radeon RX 470 (4GB)
Compare AMD Radeon RX 470 (4GB) against another machine in the Arena
Reference: Radeon 400 series on Wikipedia
- Class
- GPU
- Memory
- 4 GB GDDR5
- Runs (Q4_K_M)
- —
- Bandwidth
- 211 GB/s
- TDP
- 120 W
- Released
- 2016-08-04
- Price (US)
- $39 used as of 2026-08
Sourcing and disambiguation notes
AMD's reference RX 470 launched as a 4GB card at $170 MSRP; some board partners also shipped an 8GB variant with the same 211 GB/s bandwidth, since only capacity differs. The machine behind the llama.cpp Vulkan figures in the source is the 4GB card, and no benchmark record is filed from it: at the source's own 512-token test length the weights alone (3.79 GB) plus a fp16 KV cache (0.27 GB) already exceed the card's 4 GB before any backend overhead, so the run cannot be represented honestly without misstating either the context length or the overhead. eBay used listings span roughly $15-110 depending on condition; $39 is the tracked average for a card discontinued in 2016. Any 'new' listing is new-old-stock rather than a real retail channel, so new is left empty rather than carrying the unobtainable 2016 launch MSRP.
Specification sources
- https://wccftech.com/amd-rx-470-460-specs-performance-launch-dates/ — press, 2016-08-04
- https://github.com/ggml-org/llama.cpp/discussions/10879 — forum, 2025-09-08
Measurements
Decode is token generation — the speed you feel while an answer streams. Prefill is prompt processing — the wait before it starts. Why bandwidth predicts decode speed.
No records yet for this hardware. Know of a published benchmark on it? Submit the link.
What verified, single-source and estimated mean, and the same rows with every filter and sort in the benchmarks explorer.
Speed over time
What it can run
Fit is arithmetic, not a measurement — how it is computed. Against 4 GB; models that fit are listed largest first.
No modeled quant fits in 4 GB.