AMD Radeon Pro V340 8 GB

The AMD Radeon Pro V340 8 GB is a professional workstation graphics card built on the GCN 5.0 architecture, featuring the Vega 10 chip manufactured on a 14 nm process at GlobalFoundries. It was released on August 25, 2018, and is positioned within the Radeon Pro Vega generation, succeeding the Radeon Pro Polaris series and preceding the Radeon Pro Navi. The card is now end-of-life, and the available data shows a percentile ranking of 50 among all GPUs, indicating it sits at the median performance tier in the historical database.

Benchmark Performance

The Radeon Pro V340 8 GB delivers a peak single-precision (FP32) compute throughput of 10.75 TFLOPS, while its half-precision (FP16) output reaches 21.50 TFLOPS, achieved through a 2:1 ratio architecture. This places the card in a competitive position for its era, though the benchmark data for this specific unit shows an average score of 0, meaning no aggregated performance samples are available for direct comparison. The percentile rank of 50 indicates that, based on the broader database, this GPU performs at the midpoint of all recorded graphics cards — neither a standout leader nor a laggard, but a balanced professional offering.

In terms of raw throughput metrics, the card sustains a pixel rate of 96.00 GPixel/s and a texture rate of 336.0 GTexel/s. These figures are derived from its 3584 shading units, 224 texture mapping units (TMUs), and 64 raster operations pipelines (ROPs). The shading unit count is substantial for a workstation card of this generation, allowing the V340 to handle compute-heavy workloads with reasonable efficiency. The FP32 throughput of 10.75 TFLOPS is a meaningful indicator for double-precision-adjacent tasks, though note that the card’s architecture is optimized for single-precision and half-precision operations.

Without nearestRivals data, a direct percentage comparison against specific competitors is not possible from the FACT PACK alone. However, the 50th percentile ranking implies that in mixed professional and consumer workloads, the V340 achieves performance roughly on par with the median GPU in the database. For demanding tasks like 3D rendering or scientific simulation, the 10.75 TFLOPS figure suggests it can handle moderate compute loads, but it would trail newer or higher-tier cards. The 2:1 FP16 ratio is notable — it indicates that for machine learning inference or training tasks using half precision, the card effectively doubles its throughput, though dedicated tensor cores are absent (see Ray Tracing section).

Ray Tracing and Feature Set

The Radeon Pro V340 does not include dedicated ray tracing (RT) cores or tensor cores — these fields are null in the specification data. Instead, the card relies on the GCN 5.0 architecture’s compute units to handle any ray tracing or AI workloads, which means performance in those areas is driven entirely by the 3584 shading units and the FP16 acceleration noted above. For professional applications that utilize DirectX Raytracing (DXR), the card supports DirectX 12 with feature level 12_1, which includes basic DXR support, but without hardware acceleration, ray tracing performance will be significantly lower than on cards with dedicated RT hardware.

The API support is comprehensive for its generation: DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3 are all present. This ensures compatibility with modern CAD, DCC, and scientific visualization software that rely on these APIs. Vulkan 1.3 support is particularly relevant for cross-platform compute workloads, while OpenGL 4.6 covers legacy professional applications. The card lacks any tensor core functionality, meaning AI-based features like denoising or DLSS-style upscaling must be handled through compute shaders, which will consume shading unit resources. For the feature set, the V340 provides a solid baseline for rasterization-based rendering and compute, but it is not designed for path-traced workflows or advanced neural network acceleration.

The display output is limited to a single mini-DisplayPort 1.4a connector. This is a notable constraint for multi-monitor professional setups, as users will need to rely on daisy-chaining (if supported by the monitor) or additional adapters. The bus interface is PCIe 3.0 x16, which is standard for the era and provides adequate bandwidth for most workloads, though it may bottleneck data transfer for very large datasets compared to newer PCIe 4.0 or 5.0 interfaces.

How It Compares

Since the nearestRivals array is empty in the provided FACT PACK, a direct comparison against specific competitor models is not possible from the given data. The card’s position must therefore be inferred from its percentile rank and architectural traits.

Relative to the broader GPU database: With a percentile rank of 50, the V340 sits exactly at the median. This means half of all GPUs in the database outperform it and half underperform it. For a workstation card released in 2018, this is a reasonable standing — it was not a flagship, but it was solidly mid-pack, capable of handling professional tasks without being at the cutting edge. The 10.75 TFLOPS FP32 throughput aligns with this median positioning, as many consumer and prosumer cards from that era fell within a similar range.

Relative to its predecessor (Radeon Pro Polaris): The V340 represents a generational step forward in architecture, moving from Polaris to Vega 10. The die size of 495 mm² and transistor count of 12,500 million are indicative of a larger, more complex chip than typical Polaris offerings. This suggests a meaningful improvement in compute density, though specific performance deltas are not quantified in the FACT PACK.

Relative to its successor (Radeon Pro Navi): The successor would likely offer improved efficiency and newer features, but without numerical data, the comparison is qualitative. The V340’s 14 nm process and GCN 5.0 architecture place it a full generation behind Navi’s RDNA-based design, implying that the successor would deliver higher performance per watt and possibly better ray tracing support. However, the V340’s 8 GB HBM2 memory and 2048-bit bus remain competitive for memory bandwidth, which is a key strength for compute-heavy tasks.

FAQ

Q: What is the compute performance of the AMD Radeon Pro V340 8 GB?

A: The card delivers 10.75 TFLOPS of FP32 performance and 21.50 TFLOPS of FP16 performance, with the FP16 figure achieved via a 2:1 ratio.

Q: Does the Radeon Pro V340 support hardware ray tracing?

A: No, the card has no dedicated RT cores. It supports DirectX 12 (12_1) which includes DXR, but ray tracing must be processed through the standard shading units.

Q: What is the memory configuration of this card?

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

Q: What APIs are supported by the Radeon Pro V340?

A: The card supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3.

Q: How many display outputs does the card have?

A: It has one mini-DisplayPort 1.4a output, which limits multi-monitor setups without additional hardware.

Q: What is the production status of the Radeon Pro V340?

A: The card is end-of-life, having been released on August 25, 2018.

Power and Cooling

The Radeon Pro V340 has a thermal design power (TDP) of 230 W, which dictates its cooling and power delivery requirements. The card is dual-slot in width, indicating that it uses a substantial heatsink and fan assembly to dissipate heat. For power, it requires two 8-pin PCIe power connectors, and AMD recommends a system power supply of at least 550 W. This is a moderate power draw for a workstation GPU, allowing it to be installed in most professional systems with a reasonable PSU, but the dual 8-pin requirement means older or lower-wattage power supplies may need upgrading.

The 230 W TDP, combined with the 14 nm process node, suggests that the card runs relatively hot under full load, especially given the large die size of 495 mm². The dual-slot cooler is necessary to manage thermals during sustained compute workloads. The dimensions are 267 mm in length (10.5 inches) and 111 mm in height (4.4 inches), which is a standard size for a dual-slot card, but users should verify case clearance — the length is on the longer side, though not extreme. The power connectors are located on the top edge of the card, requiring careful cable management in smaller chassis.

Cooling performance is not quantified in the FACT PACK, but the dual-slot design and 230 W TDP imply that the card is designed for consistent operation under professional workloads, which often run at high utilization for extended periods. The suggested 550 W PSU provides a reasonable headroom above the TDP, assuming the rest of the system (CPU, drives, etc.) does not consume excessive power. For multi-GPU configurations, a higher-wattage PSU would be necessary, but the FACT PACK does not provide multi-GPU scaling data.

Memory Subsystem

The Radeon Pro V340 is equipped with 8 GB of HBM2 memory, a significant advantage for professional workloads that require high bandwidth. The memory operates at a bus width of 2048 bits, which is exceptionally wide compared to GDDR6-based cards, and the memory clock is rated at 945 MHz (1890 Mbps effective). This configuration yields a total memory bandwidth of 483.8 GB/s — a figure that is critical for handling large datasets, high-resolution textures, and compute tasks that frequently access memory.

For high-resolution workloads, such as 4K or 8K video editing, 3D rendering, or scientific visualization, the 483.8 GB/s bandwidth ensures that the GPU is not starved for data. The 8 GB capacity, while modest by modern standards, is sufficient for many professional applications that were current at the time of release, though it may limit the card in very large scene files or multi-GPU memory pooling scenarios. The HBM2 technology provides a more energy-efficient memory interface compared to GDDR5, contributing to the card’s overall power profile.

The 2048-bit bus width is a key differentiator — it allows the card to maintain high throughput even at relatively low memory clock speeds (945 MHz). This architecture is well-suited for compute-heavy tasks where memory bandwidth is the bottleneck, such as fluid dynamics simulations or large matrix operations. However, the 8 GB capacity may be a limiting factor for modern ultra-high-resolution textures or machine learning models that exceed this size. For the era and the card’s median percentile ranking, the memory subsystem is a strong point, offering a balance of capacity and bandwidth that outperforms many peers with wider capacity but narrower buses.