AMD Radeon RX 9070 XT 16GB Specs, Benchmarks & Pricing
The AMD Radeon RX 9070 XT is a high-performance graphics card released in March 2025, featuring the RDNA 4 architecture based on the Navi 48 GPU. It offers 64 compute units with 4,096 stream processors, 128 2nd-generation AI Accelerators, and 16GB of GDDR6 memory running at 20 Gbps. The card is designed for high-refresh 1440p and 4K gaming, providing exceptional performance with a boost clock reaching 3.1 GHz in factory-overclocked models. With a TDP of 304W, it delivers 48.7 TFLOPS of FP32 compute performance. The RX 9070 XT features 64MB of 3rd-generation Infinity Cache, 64 3rd-generation Ray Tracing accelerators, and supports FSR 4 (FidelityFX Super Resolution 4), AMD's machine-learning upscaling solution exclusive to RDNA 4. Built on TSMC's 4nm process using the Navi 48 XTX die, the card is positioned to compete with NVIDIA's RTX 4070 Ti and RTX 5070 at a starting price of $599. It uses a PCIe 5.0 x16 interface and offers 640 GB/s memory bandwidth on a 256-bit bus.
- Release Date: March 6, 2025
- MSRP: $599 USD
- GPU Architecture: RDNA 4
- Hardware-Accelerated GEMM Operations:FP64 FP32 TF32 BF16 FP16 FP8 FP4 INT8 INT4
- CUDA Compute Capability : n/a
Strengths
- Strong Counter-Strike 2 (1440p) performance (top 9% of GPUs)
- Strong Counter-Strike 2 (1080p) performance (top 0% of GPUs)
Specifications for AMD Radeon RX 9070 XT
| Specification | Performance Ranking |
|---|---|
| FP32 TFLOPs | |
| FP16 TFLOPs | |
| Tensor Core Count | |
| Memory Capacity (GB) | |
| Memory Bandwidth (GB/s) | |
| Int8 TOPs | |
| FP8 TFLOPs | |
| FP4 TFLOPs | Data not available |
Gaming Benchmarks for AMD Radeon RX 9070 XT
Rankings based on 816,293 real-world benchmark results from OpenBenchmarking.org
| Benchmark | Performance Ranking |
|---|---|
| 3DMark Wild Life Extreme (4K) | |
| Counter-Strike 2 (4K) | |
| Counter-Strike 2 (1440p) | |
| Counter-Strike 2 (1080p) | |
| FurMark (GL) (4K) | |
| FurMark (GL) (1440p) | |
| FurMark (GL) (1080p) | |
| FurMark (VK) (4K) | |
| FurMark (VK) (1440p) | |
| FurMark (VK) (1080p) |
Real-time AMD Radeon RX 9070 XT GPU Prices
Compare Price/Performance to other GPUs
Compare AMD Radeon RX 9070 XT to Another GPU
Price History
AMD Radeon RX 9070 XT Price History
This chart tracks AMD Radeon RX 9070 XT prices over the past 6 months based on eBay listings. The yellow line shows the average of the three lowest-priced listings each day—a useful indicator for finding deals.
Trend: Over this period, the lowest average price has been up 12.8%. The current lowest average is $818, compared to a typical $701 over the period. The lowest average price is calculated as the average of the three lowest-priced listings each day.
References
- https://www.amd.com/en/products/graphics/desktops/radeon/9000-series/amd-radeon-rx-9070xt.html
- https://www.amd.com/en/products/specifications/graphics
- https://en.wikipedia.org/wiki/RDNA_4
- https://en.wikipedia.org/wiki/Radeon_RX_9000_series
- https://www.tomshardware.com/pc-components/gpus/amd-rdna4-rx-9000-series-gpus-specifications-pricing-release-date
- https://www.tomshardware.com/pc-components/gpus/amd-radeon-rx-9070-xt-review
- https://www.gsmarena.com/amd_announces_pricing_and_availability_for_rx_9070_and_9070_xt-news-66772.php
Notes
- tensorCoreCount tracks the 2nd-generation AI Accelerators as "tensor cores". The RX 9070 XT has 128 AI Accelerators (2 per CU × 64 CUs).
- fp32TFLOPS of 48.7 calculated based on 4,096 stream processors × 2.97 GHz boost clock × 2 (dual-issue) = 48.66 TFLOPS, rounded to 48.7 TFLOPS as reported by AMD.
- fp16TFLOPS of 194.8 represents FP16 Matrix throughput (dense) via the 2nd-generation AI Accelerators (WMMA), per AMD product page convention (Peak Half Precision (FP16 Matrix) Performance). RDNA 4 matrix throughput is 4x FP32 vector (verified against AMD-published values for RX 9070=145, RX 9060=86, RX 9070 GRE=137). AMD also publishes FP16 Vector = FP32 Vector = 48.7 TFLOPS (SIMD path); that vector value is NOT used here in favor of the matrix path which is the cross-vendor analog of NVIDIA tensor-core FP16. See policy in .claude/agents/gpu-researcher.md.
- int8TOPS of 389 and fp8TFLOPS of 389 are the peak INT8 Matrix (dense) and FP8 Matrix (dense, E5M2/E4M3) performance per the AMD official product page (https://www.amd.com/en/products/graphics/desktops/radeon/9000-series/amd-radeon-rx-9070xt.html), which also lists structured-sparsity values of 779 TOPS INT8 Matrix and 779 TFLOPS FP8 Matrix (2x dense). Both fields were previously missing from this file entirely; added during the Phase 5 precision-filters audit (2026-09-02) directly from AMD's published compute specifications table. The dense fp8TFLOPS value equals both int8TOPS (389) and 2x fp16TFLOPS (194.8), consistent with AMD's halving-bitwidth-doubles-throughput Matrix Core pattern confirmed across all six RDNA 4 SKUs audited this batch.
- supportedHardwareOperations updated to add FP8 (Phase 5 audit, 2026-09-02). AMD's official product page publishes an FP8 Matrix performance row (389 TFLOPS dense) for the RX 9070 XT, confirming 2nd-generation AI Accelerator FP8 support is present on this SKU. FP8 was previously omitted from this file as a data-entry gap — it predates FP8 being checked against AMD's spec table — not because the hardware lacks FP8. Direct verification this batch confirmed FP8 is published on AMD's official product pages for all six RDNA 4 desktop/workstation SKUs, spanning both the Navi 48 die (this card, plus RX 9070 GRE/AI PRO R9700/RX 9070) and the Navi 44 die (RX 9060/RX 9060 XT), so FP8 WMMA support is architecture-wide (GFX12/RDNA 4), not die-specific. See the canonical RDNA 4 row in .claude/agents/gpu-researcher.md.
- fp4TFLOPS is intentionally omitted (not set to null-with-unverifiedSpecs) because no FP4 Matrix row is published on the AMD product page and no FP4 WMMA input format is documented in the RDNA4 Instruction Set Architecture Reference Guide. This is a confirmed hardware negative (RDNA 4 has no FP4 support), not a research gap, so unverifiedSpecs does not apply per project convention.
- Memory bandwidth calculated as 256-bit interface × 20 Gbps = 640 GB/s, enhanced by 64MB 3rd-generation Infinity Cache.
- TDP is 304W, with some custom models featuring up to 330W power limits.
- Process node is TSMC 4nm with Navi 48 XTX GPU die.
- Clock speeds vary by model - reference specs show 2,400 MHz game clock and 2,970 MHz boost, while factory-overclocked models like XFX Mercury reach 3,100 MHz boost.
- RDNA 4 introduces optimized Compute Units, 3rd-generation Ray Tracing accelerators, and 2nd-generation AI Accelerators.
- FSR 4 is exclusive to RDNA 4 GPUs and leverages the AI accelerators for ML-based upscaling.
- The RX 9070 XT competes directly with NVIDIA's RTX 4070 Ti and RTX 5070, offering competitive performance at a more attractive price point.
- MSRP of $599 USD sourced from AMD official announcement per GSMArena and Tom's Hardware coverage.