2-GPU systems including RX 7900 XTX LLM benchmarks

194 results · 303 measurements · 10 sources

Explore all results Image benchmarks Video benchmarks Submit a result

Device specifications

Variant	Memory	Memory type	Bus	Bandwidth	FP16 matrix	INT8 matrix
RX 7900 XTX ↗	24 GB	GDDR6	384-bit	960 GB/s	123 TFLOPS	123 TOPS

Manufacturer specifications. Shared system RAM is not dedicated VRAM. Unreported values remain —; measured LLM results retain their reported configuration.

Selected reported tests

24 of 194 results

Source-reported results, selected across reports by documentation quality. Match configurations before comparing values.


View ReportCustom evaluation	Qwen3.8-Flash-NextUD-Q3_K_XLModel ref. ↗	AIME24LLST protocol 1.0 subset · 10 pinned LLST v1.0 questions; not the full AIME24 benchmark32K configured context	30 %Score · n = 100 · Strict numeric extraction	—	—	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine llama.cpp Backend ROCm / HIP Harness Local LLM Standard Test (LLST) 1.0 / EvalScope Prompt / output ? / ? tokens Context 32768 Concurrency Not reported Date 2026-09-17 Full record →
View ReportLocalMaxxing	Signal-3.8-27BQ8_0Model ref. ↗	Token speed	—	801.75 tok/sPrefill · unknown	26.45 tok/sGeneration · unknown	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Arch Linux Engine llama.cpp Backend Not reported Harness LocalMaxxing community speed test; exact client where reported Prompt / output 533 / 256 tokens Context 32768 Concurrency Not reported Date 2026-09-15 Full record →
View ReportGeneric LLM	Qwen3.8-27BW4A16 AutoRound GPTQWeights ↗	Token speed	—	1,016.56 tok/sPrefill · per request	27.55 tok/sGeneration · per request	Mixed: RX7900XTX24GB + RX7700XT12GB · 2 GPUsPipeline parallel Configuration Memory Not reported OS Ubuntu Engine vLLM Backend ROCm / HIP Harness Author vLLM request test Prompt / output 4096 / 256 tokens Context Not reported Concurrency 1 Date 2026-09-04 Full record →
View ReportGeneric LLM	Qwen3.8-27BW4A16-AutoRound-GPTQModel ref. ↗	Token speedLong output gen512	—	—	97.38 tok/sGeneration · per request	2 × RX 7900 XTXTensor parallel · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Ubuntu (native Linux) Engine SGLang Backend ROCm / HIP Harness Author long-context streaming, concurrent-wave and prefill-interference client Prompt / output ? / 512 tokens Context 196608 Concurrency 1 Date 2026-09-18 Full record →
View Reportllama-bench	Llama-2-7BQ4_0Weights ↗	Token speed	—	1,898.68 tok/sPrefill · mean · per request	—	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Arch Linux / Linux 6.12.61-1-lts Engine llama.cpp Backend Vulkan Harness llama-bench Prompt / output 512 / 0 tokens Context Not reported Concurrency 1 Date 2025-12-11 Full record →
View Reportllama-bench	gemma-4-31B-itQ8_0Model ref. ↗	Token speed	—	920.55 tok/sPrefill · mean · per request	—	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS CachyOS x86_64 Engine llama.cpp Backend ROCm / HIP Harness llama-bench Prompt / output 512 / 0 tokens Context Not reported Concurrency 1 Date 2026-08-23 Full record →
View Reportllama-bench	gemma-4-31B-itQ8_0Model ref. ↗	Token speed	—	851.56 tok/sPrefill · mean · per request	—	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS CachyOS x86_64 Engine llama.cpp Backend Vulkan Harness llama-bench Prompt / output 512 / 0 tokens Context Not reported Concurrency 1 Date 2026-08-23 Full record →
View Reportllama-bench	gemma-4-31B-itIQ4_NL_L_AMDWeights ↗	Token speed	—	1,381.37 tok/sPrefill · mean · per request	—	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Not reported Engine llama.cpp Backend ROCm 6.4.4 Harness llama-bench Prompt / output 1024 / 0 tokens Context Not reported Concurrency 1 Date Not reported Full record →
View Reportllama-bench	Llama-2-7BQ4_0Weights ↗	Token speed	—	1,668.19 tok/sPrefill · mean · per request	—	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Not reported Engine llama.cpp Backend Vulkan Harness llama-bench Prompt / output 512 / 0 tokens Context Not reported Concurrency 1 Date 2025-01-08 Full record →
View ReportGeneric LLM	Mistral-Small-3.2-24B-Instruct-2506Q4_0_SWeights ↗	Token speed	43.3 tok/sThroughput · aggregate	—	—	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Not reported Engine llama.cpp Backend ROCm 6.4.4 Harness Custom serving benchmark (not named) Prompt / output ? / ? tokens Context 131072 Concurrency 1 Date Not reported Full record →
View ReportCustom evaluation	Qwen3.8-Flash-NextUD-Q3_K_XLModel ref. ↗	IFEvalLLST protocol 1.0 subset · 20 pinned LLST v1.0 questions; not the full IFEval benchmark32K configured context	95 %Score · n = 100 · Source aggregate; exact strict/loose and prompt/instruction submetric unspecified	—	—	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine llama.cpp Backend ROCm / HIP Harness Local LLM Standard Test (LLST) 1.0 / EvalScope Prompt / output ? / ? tokens Context 32768 Concurrency Not reported Date 2026-09-17 Full record →
View ReportLocalMaxxing	Signal-3.8-27BQ8_0Model ref. ↗	Token speed	—	668.3 tok/sPrefill · unknown	26.19 tok/sGeneration · unknown	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Arch Linux Engine llama.cpp Backend Not reported Harness LocalMaxxing community speed test; exact client where reported Prompt / output 533 / 256 tokens Context 32768 Concurrency Not reported Date 2026-09-15 Full record →
View ReportGeneric LLM	Qwen3.8-27BW4A16 AutoRound GPTQWeights ↗	Token speed	—	995.61 tok/sPrefill · per request	24.94 tok/sGeneration · per request	Mixed: RX7900XTX24GB + RX7700XT12GB · 2 GPUsPipeline parallel Configuration Memory Not reported OS Ubuntu Engine vLLM Backend ROCm / HIP Harness Author vLLM request test Prompt / output 16384 / 256 tokens Context Not reported Concurrency 1 Date 2026-09-04 Full record →
View ReportGeneric LLM	Qwen3.8-27BW4A16-AutoRound-GPTQModel ref. ↗	Token speedLong output gen512	—	—	44.85 tok/sGeneration · per request	2 × RX 7900 XTXTensor parallel · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Ubuntu (native Linux) Engine SGLang Backend ROCm / HIP Harness Author long-context streaming, concurrent-wave and prefill-interference client Prompt / output ? / 512 tokens Context 196608 Concurrency 1 Date 2026-09-18 Full record →
View Reportllama-bench	Llama-2-7BQ4_0Weights ↗	Token speed	—	—	85.3 tok/sGeneration · mean · per request	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Arch Linux / Linux 6.12.61-1-lts Engine llama.cpp Backend Vulkan Harness llama-bench Prompt / output 0 / 128 tokens Context Not reported Concurrency 1 Date 2025-12-11 Full record →
View Reportllama-bench	gemma-4-31B-itQ8_0Model ref. ↗	Token speed	—	—	22.97 tok/sGeneration · mean · per request	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS CachyOS x86_64 Engine llama.cpp Backend ROCm / HIP Harness llama-bench Prompt / output 0 / 128 tokens Context Not reported Concurrency 1 Date 2026-08-23 Full record →
View Reportllama-bench	gemma-4-31B-itQ8_0Model ref. ↗	Token speed	—	—	24.42 tok/sGeneration · mean · per request	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS CachyOS x86_64 Engine llama.cpp Backend Vulkan Harness llama-bench Prompt / output 0 / 128 tokens Context Not reported Concurrency 1 Date 2026-08-23 Full record →
View Reportllama-bench	gemma-4-31B-itIQ4_NL_L_AMDWeights ↗	Token speed	—	1,494.88 tok/sPrefill · mean · per request	—	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Not reported Engine llama.cpp Backend ROCm 6.4.4 Harness llama-bench Prompt / output 2048 / 0 tokens Context Not reported Concurrency 1 Date Not reported Full record →
View Reportllama-bench	Llama-2-7BQ4_0Weights ↗	Token speed	—	—	100.62 tok/sGeneration · mean · per request	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Not reported Engine llama.cpp Backend Vulkan Harness llama-bench Prompt / output 0 / 128 tokens Context Not reported Concurrency 1 Date 2025-01-08 Full record →
View ReportGeneric LLM	Mistral-Small-3.2-24B-Instruct-2506Q4_0_SWeights ↗	Token speed	32.7 tok/sThroughput · mean · per request	—	—	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Not reported Engine llama.cpp Backend ROCm 6.4.4 Harness Custom serving benchmark (not named) Prompt / output ? / ? tokens Context 131072 Concurrency 1 Date Not reported Full record →
View ReportCustom evaluation	Qwen3.8-Flash-NextUD-Q3_K_XLModel ref. ↗	C-EvalLLST protocol 1.0 subset · 20 pinned LLST v1.0 questions; not the full C-Eval benchmark32K configured context	90 %Score · n = 100 · 20 subjects; five-shot development examples	—	—	2 × RX 7900 XTXMulti-GPU · split unknown · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine llama.cpp Backend ROCm / HIP Harness Local LLM Standard Test (LLST) 1.0 / EvalScope Prompt / output ? / ? tokens Context 32768 Concurrency Not reported Date 2026-09-17 Full record →
View ReportLocalMaxxing	gemma-4-26B-A4B-itQ8_0Model ref. ↗	Token speed	—	1,995.1 tok/sPrefill · unknown	91.26 tok/sGeneration · unknown	2 × RX 7900 XTXLayer split · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Arch Linux Engine llama.cpp Backend Not reported Harness LocalMaxxing community speed test; exact client where reported Prompt / output 538 / 256 tokens Context 8192 Concurrency Not reported Date 2026-09-15 Full record →
View ReportGeneric LLM	Qwen3.8-27BW4A16 AutoRound GPTQWeights ↗	Token speed	—	933.16 tok/sPrefill · per request	24.78 tok/sGeneration · per request	Mixed: RX7900XTX24GB + RX7700XT12GB · 2 GPUsPipeline parallel Configuration Memory Not reported OS Ubuntu Engine vLLM Backend ROCm / HIP Harness Author vLLM request test Prompt / output 32768 / 256 tokens Context Not reported Concurrency 1 Date 2026-09-04 Full record →
View ReportGeneric LLM	Qwen3.8-27BW4A16-AutoRound-GPTQModel ref. ↗	Token speedSingle-stream decode	—	—	103.82 tok/sGeneration · per request	2 × RX 7900 XTXTensor parallel · 24 GB VRAM/GPU Configuration Memory 24 GB VRAM/GPU OS Ubuntu (native Linux) Engine SGLang Backend ROCm / HIP Harness Author long-context streaming, concurrent-wave and prefill-interference client Prompt / output ? / ? tokens Context 196608 Concurrency 1 Date 2026-09-18 Full record →

Explore all 194 results →

Models and software in this collection

Model families: Gemma 4, Laguna, Llama 2, Mistral Small 3.2, Qwen3, Qwen3.6, Qwen3.8

Engines: SGLang, llama.cpp, vLLM

Methodology and evidence standards · Download the research dataset

2-GPU systems including RX 7900 XTX LLM benchmarks

Selected reported tests

LLMs benchmarked on the RX 7900 XTX