AMD Radeon Pro W7600 8GB Specs, Benchmarks & Pricing
The AMD Radeon Pro W7600 is a professional workstation graphics card launched on August 3, 2023, featuring AMD's RDNA 3 architecture on the Navi 33 GPU. Built on TSMC's 6nm process as a monolithic die containing 13.3 billion transistors across 204 mm², the card features 32 compute units with 2,048 stream processors and 64 AI Accelerators for enhanced machine learning performance. It includes 32 second-generation ray tracing accelerators and 8GB of GDDR6 memory on a 128-bit bus, delivering 288 GB/s of bandwidth with 32MB of AMD Infinity Cache. The W7600 offers 21.4 TFLOPS of peak FP32 performance and 42.8 TFLOPS of peak FP16 performance. With a TDP of 130W and positioned at $599, this single-slot card targets mainstream professional content creation, CAD, and engineering workflows. It features four DisplayPort 2.1 outputs with UHBR 20 support providing up to 80 Gb/s of display link bandwidth, making it suitable for multi-display professional workflows up to 8K resolution. The card requires a single 6-pin PCIe power connector and uses a PCIe 4.0 x8 interface.
- Release Date: August 3, 2023
- MSRP: $599 USD
- GPU Architecture: RDNA 3
- Hardware-Accelerated GEMM Operations:FP64 FP32 TF32 BF16 FP16 FP8 FP4 INT8 INT4
- CUDA Compute Capability : n/a
Considerations
- Lower FP16 compute performance compared to other GPUs
- Lower tensor core count compared to other GPUs
- Below average 3DMark Wild Life Extreme (1080p) performance
Specifications for AMD Radeon Pro W7600
| Specification | Performance Ranking |
|---|---|
| FP32 TFLOPs | |
| FP16 TFLOPs | |
| Tensor Core Count | |
| Memory Capacity (GB) | |
| Memory Bandwidth (GB/s) | |
| Int8 TOPs | Data not available |
| FP8 TFLOPs | Data not available |
| FP4 TFLOPs | Data not available |
Gaming Benchmarks for AMD Radeon Pro W7600
Rankings based on 636 real-world benchmark results from OpenBenchmarking.org
| Benchmark | Performance Ranking |
|---|---|
| 3DMark Wild Life Extreme (1080p) |
Real-time AMD Radeon Pro W7600 GPU Prices
Compare Price/Performance to other GPUs
Compare AMD Radeon Pro W7600 to Another GPU
Price History
AMD Radeon Pro W7600 Price History
This chart tracks AMD Radeon Pro W7600 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 down 27.8%. The current lowest average is $444. The lowest average price is calculated as the average of the three lowest-priced listings each day.
Product Identifiers
Manufacturer Part Numbers (1)
- AMD OPN
- 100-300000077
Available from 2 Partners (2 products)
- AMD Radeon Pro W7600, 8 GB GDDR6, Full Height, PCIe 4.0x8, 4 DP Graphics Card
- 490-BJZF(part number)
- AMD Radeon Pro W7600 Graphics Card
- 8D6B9AA(part number)
References
- https://www.amd.com/en/products/graphics/workstations/radeon-pro/w7600.html
- https://www.amd.com/en/newsroom/press-releases/2023-8-3-new-amd-radeon-pro-w7000-series-workstation-graphi.html
- https://ir.amd.com/news-events/press-releases/detail/1147/new-amd-radeon-pro-w7000-series-workstation-graphics-cards
- https://www.cgchannel.com/2023/08/amd-launches-radeon-pro-w7600-and-radeon-pro-w7500/
- https://www.storagereview.com/review/amd-radeon-pro-w7500-and-w7600-review
- https://www.tomshardware.com/news/amd-radeon-pro-w7600-and-w7500-revealed
- https://www.techpowerup.com/312011/amd-announces-radeon-pro-w7600-and-w7500-graphics-cards
- https://www.bhphotovideo.com/c/product/1781098-REG/amd_100_300000077_radeon_pro_w7600_graphic.html
- https://www.dell.com/en-us/shop/dell-amd-radeon-pro-w7600-8-gb-gddr6-full-height-pcie-40x8-4-dp-graphics-card/apd/490-bjzf/graphic-video-cards
- https://wccftech.com/amd-radeon-pro-w7600-workstation-gpu-benchmarked-navi-33-2048-cores-8-gb-vram/
- https://gpuopen.com/learn/wmma_on_rdna3/
Notes
- FP32 TFLOPS of 21.4 sourced from AMD's official product page "Peak Single Precision" specification. Note: Earlier press materials stated 20.0 TFLOPS; currently selected value reflects the final product page peak spec despite discrepancies with launch documentation.
- fp16TFLOPS of 42.8 represents FP16 Matrix throughput (dense) via RDNA 3 AI Accelerators (WMMA), per AMD product page convention. This equals 2x FP32 vector on RDNA 3 silicon. AMD also publishes FP16 Vector = FP32 Vector on the same product page; that SIMD path 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.
- tensorCoreCount of 64 represents the AI Accelerators (2 per compute unit × 32 compute units), consistent with RDNA 3 architecture
- Memory bandwidth calculated as 128-bit bus × 18 Gbps effective = 288 GB/s, confirmed by multiple sources
- RDNA 3 architecture includes dedicated AI Accelerators supporting Wave Matrix Multiply Accumulate (WMMA) operations for machine learning workloads
- RDNA 3 AI Accelerators support FP16, BF16, INT8, and INT4 data types for matrix operations
- Built on TSMC 6nm (N6) process as a monolithic die (not chiplet design like Navi 31/32)
- Navi 33 die specifications - 204 mm² die size, 13.3 billion transistors
- Released August 3, 2023 at $599 USD MSRP per AMD official press release
- MSRP of $599 USD sourced from AMD official press release dated August 3, 2023
- AMD Ordering Part Number (OPN) 100-300000077 sourced from multiple retailers including B&H Photo, Amazon, Newegg, and Micro Center
- UPC 727419314855 and EAN 0727419314855 are retail product codes documented for reference only
- Dell part number 490-BJZF sourced from Dell.com official product page
- HP part number 8D6B9AA sourced from HP channel partners
- int8TOPS is null because AMD does not publish separate INT8 TOPS specifications for RDNA 3 workstation GPUs; INT8 is supported via AI Accelerators but runs at same speed as FP16
- Card features 32MB AMD Infinity Cache for improved memory performance
- Single-slot design with blower-style cooler, requires single 6-pin PCIe power connector
- PCIe 4.0 x8 interface (not x16)
- Phase 5 audit (2026-09-02): supportedHardwareOperations corrected from [FP16, FP32, BF16, INT8, INT4] to the canonical RDNA 3 set [FP16, BF16, INT8, INT4] per AMD GPUOpen, "How to accelerate AI applications on RDNA 3 using WMMA" (https://gpuopen.com/learn/wmma_on_rdna3/), which documents WMMA input formats on RDNA 3 Compute Units as FP16, BF16, and 8-bit/4-bit integer (IU8/IU4) only. FP32 was removed: it is a plain shader-ALU precision, not accelerated through the AI Accelerator/WMMA matrix path, matching the canonical RDNA 3 row in .claude/agents/gpu-researcher.md.
- fp8TFLOPS and fp4TFLOPS are left null as a confirmed hardware negative, not a missing-data gap: RDNA 3's WMMA AI Accelerators support only FP16, BF16, and 8-bit/4-bit integer input formats per AMD GPUOpen (https://gpuopen.com/learn/wmma_on_rdna3/). FP8 support was added one generation later in RDNA 4, and no AMD consumer/prosumer GPU family has shipped FP4 matrix hardware to date. Per the not-found-marking rules in .claude/agents/gpu-researcher.md, this confirmed negative is NOT added to unverifiedSpecs.