GPU directory · Radeon RX · RDNA 4 · gfx1201
Cited, not measured on the local desktop
AMD Radeon RX 9070
16 GB gfx1201 RDNA 4 — the non-XT 9070, not the 16 GB 9070 XT and not the 12 GB 9070 GRE. AMD product page and driver listing cite 16 GB GDDR6, up to 640 GB/s, a 256-bit bus on the AMD 9000-series quick-reference, and 220 W typical board power. Official on the cited ROCm 10.1 matrix as gfx1201 — same LLVM target as the measured R9700, half the VRAM. House 35B Q4–Q6 files do not fit this VRAM on the calculator — expect RAM offload or a listed Q3 file, both estimates. Buy the R9700 when you want measured 32 GB recipes. Not a measured local desktop card.
If the app is not installed, packages are not published yet. There is no public download URL. Read the Stack Lab install notes . If Stack Lab is already running on this machine, open the local app .
How we test — Measured cells cite official amdai-bench/run@1 cards. Estimates stay labeled. Q3 is estimate-only. Q1/Q2 stay off.
Official identity
- Card
- AMD Radeon RX 9070
- gfx
- gfx1201
- Published VRAM
- 16 GB
- Hardware label, not a bench
- ROCm matrix
- Official
- gfx1201 on the 10.1 matrix.
Cited support row: ROCm 10.1.0 compatibility matrix and Linux system requirements. We do not invent a support claim past that listing. AMD product page: RX 9070.
HSA_OVERRIDE: Not needed on current 10.x.
What fits on this card (256K context)
Shopper verdict at 4K, then follow the pair page for 32K and full context. Q4 and up. No printed prices.
| Model | Verdict |
|---|---|
| Qwen3.6-35B on RX 9070 | Fits with RAM offload |
| Qwen 3.8 Flash on RX 9070 | Fits with RAM offload |
| gpt-oss 20B on RX 9070 | Needs 2 GPUs |
| Gemma 3 27B on RX 9070 | Fits with RAM offload |
| Llama 3.3 70B on RX 9070 | Fits with RAM offload |
| GLM 5.3 Flash on RX 9070 | Fits with RAM offload |
| Qwen3.6 27B on RX 9070 | Fits with RAM offload |
| Qwen3.8 27B on RX 9070 | Fits with RAM offload |
What fits in VRAM
Qwen3.6-35B-A3B on llama.cpp, f16 KV, one GPU. Q6 / Q5 / Q4 rows are calculator estimates. UD-Q3_K_XL is a labeled estimate (Hugging Face listing for Qwen3.6-35B-A3B-UD-Q3_K_XL.gguf (16.8 GB)) — not a local desktop run. Speed stays not measured. Change quant or context in the calculator.
| Quant | Kind | Weights | 4K total | 32K total | Speed |
|---|---|---|---|---|---|
| Q6_K | Estimate | 26.7 GB | 27.7 GB Does not fit this GPU | 28.2 GB Does not fit this GPU | not measured |
| Q5_K_M | Estimate | 24.8 GB | 25.7 GB Does not fit this GPU | 26.2 GB Does not fit this GPU | not measured |
| Q4_K_M | Estimate | 19.8 GB | 20.6 GB Does not fit this GPU | 21.2 GB Does not fit this GPU | not measured |
| UD-Q3_K_XL | Q3 estimate | 15.6 GB | 16.4 GB Does not fit this GPU | 17.0 GB Does not fit this GPU | not measured |
Weights notes: Q6_K — Estimate: total_params × 6.56 bpw (nominal GGUF Q6_K; UD-Q6_K_XL uses the same bucket) / 8. Active params do not shrink the file. Q5_K_M — Hugging Face listing for Qwen3.6-35B-A3B-UD-Q5_K_XL.gguf (26.6 GB). Converted 26.6 × 10⁹ bytes → GiB. Not measured VRAM. Q4_K_M — Estimate: total_params × 4.85 bpw (nominal GGUF Q4_K_M) / 8. Active params do not shrink the file. UD-Q3_K_XL — Hugging Face listing for Qwen3.6-35B-A3B-UD-Q3_K_XL.gguf (16.8 GB). Converted 16.8 × 10⁹ bytes → GiB. Not measured VRAM.
System RAM
Minimum system RAM needed and RAM needed at full context for Qwen3.6-35B-A3B on llama.cpp at Q4_K_M, 32K selected and the model’s full context. Estimate until a first-party amdai-bench/run@1 card exists for this SKU. Rounded up to a real DIMM kit.
Minimum system RAM needed
32 GB kit
Estimate · 29.2 GiB raw · 32K context
RAM needed at full context
48 GB kit
Estimate · 33.6 GiB raw · 256K full context
Estimate. Offloaded weights + KV in RAM + runtime + OS/desktop + headroom, rounded up to a 32 GB DIMM kit.
Breakdown
- Offloaded weights ·
- 3.76 GiB
- Separable embed in RAM ·
- 0 GiB
- KV in RAM at selected ·
- 0.69 GiB
- KV in RAM at full context ·
- 5.06 GiB
- Runtime / backend ·
- 0.71 GiB
- OS / desktop baseline ·
- 8.00 GiB
- Headroom ·
- 16.0 GiB
llama.cpp-first stack
-
Step 1
llama.cpp first
Vulkan is the consumer default on this site. HIP when rocminfo lists gfx1201. Official on the cited ROCm 10.1 matrix — not Vulkan-only. Ollama and vLLM are not the house path for this card.
-
Step 2
Same gfx1201 family as the R9700
gfx1201 is the same LLVM target as the measured 32 GB R9700 and the 16 GB 9070 XT. Install the gfx1201 ROCm family package. Do not mix a gfx110x 7900 or gfx1200 9060 XT pin.
-
Step 3
16 GB non-XT, not the XT or GRE
This sheet sizes the 16 GB non-XT 9070. AMD product page: 16 GB GDDR6, up to 640 GB/s, 220 W typical board power. The 16 GB 9070 XT and the 12 GB 9070 GRE are different listings. House 35B Q4–Q6 files do not fit this VRAM — expect RAM offload or a lighter listed Q3 file, both estimates.
-
Step 4
Setup, then Stack Lab doctor
First-run setup is Ubuntu. Open Stack Lab doctor for the gfx1201 install family. House daily and Flash profiles are R9700 / local desktop — they do not apply to this 16 GB card. Packages are not published yet.
Compare with…
Head-to-head pages use the same VRAM, bandwidth, and calculator fit already on the sheets. Q3 is a labeled estimate. No invented prices. RX 9070 against the other published sheets:
Also: PRO W6800 · PRO W7700 · PRO W7800 · PRO W7900 · RX 6700 XT · RX 6800 XT · RX 6950 XT · RX 7600 · RX 7600 XT · RX 7700 XT · RX 7800 XT · RX 7900 GRE · RX 7900 XT · RX 7900 XTX · RX 9060 XT · Strix Halo
Check price · no live or invented prices
AMD Radeon RX 9070
16 GB non-XT listing. Not the 16 GB 9070 XT and not the 12 GB 9070 GRE. No live or invented price.
- Amazon: check price
- Newegg: check price
- B&H Photo: check price
- Micro Center: check price
- AMD Store: check price
Return terms vary by seller and item. Newegg's Hassle-Free page covers only Sold-and-Shipped-by-Newegg listings; marketplace items use the seller's policy. See the listing's Warranty & Returns tab.
As an Amazon Associate I earn from qualifying purchases. Some Newegg links use Rakuten Advertising and some B&H links use B&H’s affiliate program when those IDs are configured. Micro Center and AMD Store links are plain “also available at” links — no tracking. Buttons say “Check price” and never invent or live-quote a price. Full disclosure.
Configurations we have
No first-party amdai-bench/run@1 card exists for this SKU. llama.cpp first. Empty cells stay empty. Fit numbers on this page are calculator estimates. Shopper RAM lines stay Minimum system RAM needed and RAM needed at full context — labeled estimate until measured.
No tested recipe yet — contribute. This card stays listed so the gap is visible.