AMD Radeon RX Vega 64 Liquid Cooling

AMD Radeon RX Vega 64 Liquid Cooling is a factory-integrated liquid-cooled variant of the Vega 10 GPU, built on the GCN 5.0 architecture using a 14 nm process at GlobalFoundries. It packs 12,500 million transistors on a 495 mm² die, yielding a transistor density of 25.3M per mm², and launched on 2017-08-06 with a launch MSRP of 699 USD. This card targets high-refresh and high-resolution gaming, though its production status is now end-of-life, with Polaris as its predecessor and Navi as its successor.

Benchmark Performance

The FACT PACK lists no explicit benchmark scores for this card, and its percentileVsAllGpus sits at 50, meaning it lands exactly in the middle of all GPUs tracked by the database. The average benchmark score is 0, which indicates that no standardized performance samples are available for this specific liquid-cooled SKU in the current dataset. However, the raw compute specifications provide a clear picture of its theoretical capabilities. The GPU delivers 13.74 TFLOPS of FP32 performance, which is substantial for a 2017-era flagship. Its FP16 throughput doubles to 27.48 TFLOPS via a 2:1 ratio, a feature that can accelerate certain compute workloads but is less relevant for gaming.

The pixel rate is 107.3 GPixel/s, and the texture rate is 429.3 GTexel/s, driven by 4096 shading units, 256 TMUs, and 64 ROPs. These numbers suggest that the card can sustain high fill rates at 1440p and even 4K, though the lack of benchmark data means we cannot quantify real-world deltas against competitors. The clock speeds are notable: a base clock of 1406 MHz and a boost clock of 1677 MHz. The liquid cooling solution allows these clocks to remain stable under sustained load, which is critical because the air-cooled Vega 64 often throttled. The data implies that this variant should outperform the standard Vega 64 in sustained workloads, but without nearestRivals entries, we cannot state exact percentage advantages. The 50th percentile ranking suggests it is neither a top-tier performer nor a weak one—it sits in the middle of the field, which is consistent with a high-end card from 2017 that has since been surpassed by newer architectures.

Who Should Consider It

Based on the compute and memory specifications, this card is best suited for gamers targeting 1440p with high refresh rates or 4K at moderate settings. The 13.74 TFLOPS FP32 throughput and 483.8 GB/s bandwidth provide enough headroom for demanding titles at these resolutions. At 1080p, the card is overkill for most esports titles, as the 64 ROPs and 107.3 GPixel/s pixel rate can easily handle 144Hz or 240Hz displays. For 1440p, the 8 GB HBM2 frame buffer (detailed in the Memory Subsystem section) is adequate for current games, but future titles with heavy texture packs may strain the capacity. At 4K, the card can run many games at medium to high settings, but ultra settings with ray tracing enabled will likely push it beyond its limits—the RT performance is absent from the FACT PACK, so we cannot quantify that. The liquid cooling ensures consistent boost clocks, which is a boon for long gaming sessions where air-cooled cards might drop performance due to thermal throttling. Users who play in well-ventilated cases and prefer a quieter experience under load will find the integrated liquid loop beneficial, though the dual-slot design and 272 mm length require adequate case clearance.

Memory Subsystem

The RX Vega 64 Liquid Cooling features 8 GB of HBM2 memory on a 2048-bit bus, delivering a bandwidth of 483.8 GB/s. The memory clock is listed as 945 MHz, with an effective rate of 1890 Mbps. This is a significant advantage over GDDR5-based competitors of its era, as the wide bus and high bandwidth reduce bottlenecks in memory-intensive scenarios like 4K textures and high-resolution rendering. The 8 GB capacity is sufficient for 1440p gaming with ultra settings in most titles, but it becomes a limiting factor at 4K in games that exceed 8 GB of VRAM usage. The 2048-bit bus width is exceptionally wide, which allows the memory controller to fetch data more efficiently, reducing latency and improving minimum frame rates. For high-resolution gaming, the 483.8 GB/s bandwidth ensures that the GPU’s 4096 shading units are not starved for data, which is often the case with narrower-bus cards. However, the HBM2 memory operates at a relatively modest 945 MHz, and the effective 1890 Mbps is lower than later HBM2 implementations, so the bandwidth advantage is not as large as the bus width alone might suggest. The data indicates that for 1440p and below, memory performance is a non-issue; at 4K, the capacity, not bandwidth, is the more likely bottleneck.

How It Compares

The FACT PACK does not include any nearestRivals entries, so this section cannot reference specific competitor names, scores, or deltaPct values. The percentileVsAllGpus of 50 places the card in the median position across all GPUs in the database. This implies that it performs roughly on par with the average GPU tracked, but that average includes many lower-end cards, so the Vega 64 Liquid’s raw specs are clearly above that median. Without rival data, we can only infer that its 13.74 TFLOPS and 483.8 GB/s bandwidth place it in the upper tier of 2017 GPUs, likely competing with other flagship cards of that generation. The absence of benchmark scores means we cannot state a percentage lead or deficit against any specific product. The data is silent on this front, so any comparison would require speculation, which is outside the scope of this analysis.

FAQ

Q: What is the launch MSRP of this card?

A: The launch MSRP is 699 USD, as listed in the FACT PACK.

Q: Does this card support DirectX 12 Ultimate?

A: The FACT PACK lists DirectX support as 12 (12_1), which is DirectX 12 with Feature Level 12_1, but it does not mention DirectX 12 Ultimate, so that cannot be confirmed.

Q: How much VRAM does it have and what type?

A: It has 8 GB of HBM2 memory on a 2048-bit bus, with a bandwidth of 483.8 GB/s.

Q: What is the TDP and recommended PSU?

A: The TDP is 345 W, and the suggested PSU is 700 W.

Q: What power connectors does it require?

A: It requires two 8-pin power connectors.

Q: Is this card still in production?

A: No, the production status is listed as end-of-life.

Power and Cooling

The TDP is rated at 345 W, which is a high figure for a 2017 GPU, reflecting the power-hungry Vega 10 architecture. The suggested PSU is 700 W, which provides adequate headroom for the card plus a typical CPU and peripherals. Power is delivered via two 8-pin connectors, which is standard for high-end cards of this era. The liquid cooling solution is integrated, meaning it is a closed-loop system that does not require additional radiators or tubing. The card is dual-slot, with dimensions of 272 mm in length, 112 mm in height, and 40 mm in width (10.7 x 4.4 x 1.6 inches). This form factor is longer than many air-cooled cards, so case compatibility must be checked. The liquid cooling allows the GPU to maintain its 1677 MHz boost clock under sustained load, which is a key advantage over air-cooled variants that often drop clocks. However, the 345 W TDP means the card generates significant heat, and the liquid loop must dissipate it through a radiator that is not detailed in the FACT PACK. The 700 W PSU recommendation is a safe baseline, but users with high-end CPUs or multiple drives may need more. The 2x 8-pin connectors indicate that the card draws power from both slots, which is typical for flagship GPUs.

Ray Tracing and Feature Set

The FACT PACK does not list any ray tracing cores (rtCores) or tensor cores (tensorCores) for this card, and both fields are null. This indicates that the RX Vega 64 Liquid Cooling does not have dedicated hardware for ray tracing or AI-accelerated tensor operations. Instead, it relies on the GCN 5.0 architecture’s traditional shader-based approach. The API support includes DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. DirectX 12 (12_1) provides feature level 12_1, which includes some advanced rendering features but not the hardware-accelerated ray tracing found in DirectX 12 Ultimate. Vulkan 1.3 support is notable, as it allows for low-level access to the GPU, which can improve performance in Vulkan-based games. OpenGL 4.6 is also supported, though this is less relevant for modern gaming. The display outputs include 1x HDMI 2.0b and 3x DisplayPort 1.4a, which support high refresh rates at 1440p and 4K, but not the higher bandwidth of DisplayPort 2.0. The 13.74 TFLOPS FP32 performance can handle software-based ray tracing effects in some games, but without dedicated RT cores, the performance will be significantly lower than cards with hardware acceleration. The FP16 throughput of 27.48 TFLOPS (2:1) could accelerate certain compute workloads, but it does not translate to gaming features like DLSS, which requires tensor cores. Overall, the feature set is solid for its time, but it lacks the modern ray tracing and AI capabilities found in later GPU generations.