The AMD Radeon PRO W6800X Duo graphics card which features dual Big Navi 21 GPUs beaten NVIDIA's GeForce RTX 3090 and RTX A6000 in the Octane Render benchmark. The benchmarks posted by Bare Feats show more than just Octane Render and also compare the monster card in other content creation applications against older Pro & RX cards.
AMD Radeon PRO W6800X Duo $5000 Graphics Card Is 14% Faster Than NVIDIA RTX A6000 In Octane Render Benchmark
The AMD Radeon PRO W6800X Duo graphics card is the only dual-chip graphics card available in the RDNA 2 family and it's exclusive to the Apple Mac Pro. The card was announced last month & costs $5000 US to purchase from the Apple Store. The card comes in the MPX module and is only compatible with the Mac Pro 2019 system. Following are some of its highlights:
Two W6800X GPUs, each with 60 compute units and 3840 stream processors64GB of GDDR6 memory (32GB per GPU), each with 512GB/s memory bandwidthUp to 30.2 teraflops single precision or 60.4 teraflops half precisionFour Thunderbolt 3 ports and one HDMI 2.0 port on cardFour DisplayPort connections routed to the system to support internal Thunderbolt 3 portsSupport for up to eight 4K displays, four 5K displays, or six Pro Display XDRsFull-height MPX Module fills an MPX bay and uses extra power and PCIe bandwidthOnboard Infinity Fabric Link connection connects two W6800X GPUs at up to 84GB/s in each direction; external Infinity Fabric Link connection enables two W6800X Duo modules to connect four W6800X GPUs
AMD Radeon PRO W6800X Duo GPU Specifications
The flagship variant of all is the dual Big Navi 21 GPU-based Radeon Pro W6800X which offers up to 64 GB of GDDR6 memory and two 60 CU GPU dies that offer a total of 120 Compute Units or 7680 cores. The clock speed for each core is maintained at 1979 MHz, offering up to 30.2 TFLOPs of FP32 and 60.4 TFLOPs of FP16 Compute power. The card offers up to 1 TB/s of memory bandwidth with its dual 256-bit bus interface. The two cut-down Navi 21 GPU dies should bring the total wattage of the DUO to around 400W.
AMD Radeon PRO W6000 Series Graphics Cards
| Graphics Card | GPU | Process Node | Compute Units / Cores | Compute Rate (FP32) | Compute Rate (FP16) | VRAM | Memory Bus | Memory Bandwidth | TDP | Price |
|---|---|---|---|---|---|---|---|---|---|---|
| Radeon Pro W6800X Duo | Navi 21 | TSMC 7nm | 60 / 7680 | 30.2 0TFLOPs | 60.40 TFLOPs | 64 GB GDDR6 | 256-bit x2 | 1 TB/s | 400W | TBA |
| Radeon Pro W6900X | Navi 21 | TSMC 7nm | 80 / 5120 | 22.20 TFLOPs | 44.40 TFLOPs | 32 GB GDDR6 | 256-bit | 512 GB/s | 300W | TBA |
| Radeon Pro W6800X | Navi 21 | TSMC 7nm | 60 / 3840 | 16.00 TFLOPs | 32.00 TFLOPs | 32 GB GDDR6 | 256-bit | 512 GB/s | 300W | TBA |
| Radeon Pro W6800 | Navi 21 | TSMC 7nm | 60 / 3840 | 17.83 TFLOPs | 35.66 TFLOPs | 32 GB GDDR6 | 256-bit | 512 GB/s | 250W | $2249 US |
| Radeon Pro W6600 | Navi 23 | TSMC 7nm | 28 / 1792 | 10.40 TFLOPs | 20.8 TFLOPs | 8 GB GDDR6 | 128-bit | 224 GB/s | 100W | $649 US |
| Radeon Pro W6600M | Navi 23 | TSMC 7nm | 28 / 1792 | 10.40 TFLOPs | 20.8 TFLOPs | 8 GB GDDR6 | 128-bit | 224 GB/s | 90W | TBA |
| Radeon Pro W6500M | Navi 24 | TSMC 6nm | 16/ 1024 | 5.30 TFLOPs | 10.6 TFLOPs | 4 GB GDDR6 | 64-bit | 128 GB/s | 50W | TBA |
| Radeon Pro W6400 | Navi 24 | TSMC 6nm | 12 / 768 | 3.50 TFLOPs | 7.00 TFLOPs | 4 GB GDDR6 | 64-bit | 128 GB/s | 50W | $229 US |
| Radeon Pro W6300M | Navi 24 | TSMC 6nm | 12 / 768 | 3.37 TFLOPs | 6.74 TFLOPs | 2 GB GDDR6 | 32-bit | 64 GB/s | 25W | TBA |









