AMD Radeon R9 380 OEM

The AMD Radeon R9 380 OEM is a GCN 3.0 architecture graphics card built on the 28 nm process at TSMC. It houses the Antigua chip, packing 5,000 million transistors on a 366 mm² die. As an end-of-life product from the Pirate Islands (R9 300) generation, it sits at the 50th percentile of all GPUs in the database, indicating a squarely mid-range performance tier. With a dual-slot cooler and dual 6-pin power connectors, it is a substantial but not extreme card for its era.

Benchmark Performance

The R9 380 OEM’s raw compute output is defined by its 1,792 shading units, 112 texture mapping units, and 32 raster operations pipelines. These specifications translate to a peak FP32 throughput of 3.290 TFLOPS, which is identical to its FP16 rate, confirming a 1:1 ratio with no dedicated half-rate path. The card achieves a pixel fill rate of 29.38 GPixel/s and a texture fill rate of 102.8 GTexel/s. These numbers place the card in a familiar performance envelope for its class: sufficient for 1080p gaming in its generation, but with clear limits at higher resolutions or with heavy post-processing.

The benchmark percentile data shows the card at the 50th percentile against all GPUs in the database, meaning exactly half of all tracked graphics cards perform better, and half perform worse. This is a classic mid-pack position. The average benchmark score is zero, which suggests no direct aggregate score was computed for this specific OEM variant, but the percentile ranking provides the relative context. In practical terms, the 3.290 TFLOPS compute figure indicates the card can handle contemporary esports titles and moderately demanding AAA games at medium to high settings, but it will struggle with more recent, compute-heavy titles that require higher shader throughput.

Without direct rival scores in the provided data, the analysis relies on the architectural characteristics. The R9 380 OEM’s 102.8 GTexel/s texture rate is the single most telling number for standard rasterization: it allows for high-detail texture filtering at 1080p without bottlenecking the shading units. The 29.38 GPixel/s pixel rate, meanwhile, is a more modest figure, suggesting that fill-rate-bound scenarios, such as 1440p rendering or heavy alpha effects, will cause noticeable frame rate drops. The card’s performance is fundamentally balanced around a 1080p target, with headroom for older titles at 1440p if settings are reduced.

Memory Subsystem

The R9 380 OEM is equipped with 4 GB of GDDR5 memory on a 256-bit bus. The memory clock runs at 1375 MHz, which translates to 5.5 Gbps effective data rate. This configuration yields a total memory bandwidth of 176.0 GB/s. This is a critical specification for high-resolution workloads: 4 GB of VRAM is sufficient for 1080p textures in most titles from its release era, but the 176.0 GB/s bandwidth is the limiting factor for 1440p or 4K.

At 1080p, the 176.0 GB/s bandwidth is generally adequate for the 3.290 TFLOPS compute throughput, allowing the shading units to stay fed without excessive stutter. However, at higher resolutions, the demand for texture data and frame buffer access grows disproportionately. The 256-bit bus width helps mitigate this, but the effective 5.5 Gbps speed is not exceptional for the generation. The data indicates a balanced design: the 4 GB capacity prevents texture thrashing in modern games, but the bandwidth will be saturated in scenarios with high resolution and high anti-aliasing. For users pushing beyond 1080p, the memory subsystem will be the primary bottleneck, capping performance well before the GPU’s compute limits are reached.

How It Compares

The nearestRivals array in the fact pack is empty, so a direct deltaPct comparison against specific competitor cards cannot be made from this data. The analysis must instead rely on the percentile position and the absolute architectural metrics. The card’s 50th percentile rank is the only comparative anchor, placing it in a broad mid-range bracket.

Given the absence of rival data, the comparison is purely contextual. Against newer generation cards, the R9 380 OEM’s 28 nm process and GCN 3.0 design will inevitably lag in both performance and power efficiency. Against older cards, such as those from the Volcanic Islands generation (its direct predecessor), it offers a generational uplift in transistor count, from a smaller die to a 5,000 million transistor configuration, which supports higher clock speeds and more shading units. The leap from predecessor to successor — from Volcanic Islands to Arctic Islands — shows a clear progression, but within its own generation, the R9 380 OEM occupies a specific niche: it is not the top-tier card, but it is far from an entry-level part. Its 190 W TDP and 450 W suggested PSU requirement indicate a card that demands a respectable power supply, but not a flagship-level one.

FAQ

Q: What is the memory bandwidth of the Radeon R9 380 OEM?

A: The card has a memory bandwidth of 176.0 GB/s, derived from a 256-bit bus and 5.5 Gbps effective GDDR5 memory speed.

Q: Does the R9 380 OEM support DirectX 12?

A: Yes, it supports DirectX 12 with a feature level of 12_0, along with OpenGL 4.6 and Vulkan 1.2.170.

Q: What is the peak FP32 compute performance?

A: The peak FP32 performance is 3.290 TFLOPS, with FP16 performance also at 3.290 TFLOPS, indicating a 1:1 ratio.

Q: How much VRAM does the card have and what type?

A: It has 4 GB of GDDR5 VRAM, which is a standard capacity for its generation.

Q: What is the recommended power supply wattage?

A: The suggested power supply is 450 W, and the card itself has a 190 W TDP and requires two 6-pin power connectors.

Q: What display outputs are available?

A: The card provides 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2 outputs.

Ray Tracing and Feature Set

The R9 380 OEM does not include dedicated ray tracing or tensor cores; these fields are null in the specification data. This is consistent with its GCN 3.0 architecture, which predates the hardware-accelerated ray tracing found in later AMD RDNA and CDNA architectures. Instead, the card relies on traditional rasterization for rendering. Its feature set for modern APIs includes DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.2.170, which allows it to run modern games and applications, but without the performance benefits of dedicated RT or AI acceleration hardware.

The absence of RT cores means any ray tracing effects in games must be handled by the standard shader units, which will severely impact performance given the 3.290 TFLOPS compute budget. Similarly, the lack of tensor cores means no dedicated hardware for AI-based upscaling techniques; any such features would rely on compute shaders, which would again tax the limited FP32 throughput. The card’s feature set is therefore firmly rooted in its 2015-era design: it supports the necessary API hooks for modern software compatibility, but it lacks the specialized silicon required for contemporary graphics features. Users should expect the R9 380 OEM to function as a capable DirectX 12 and Vulkan rasterizer, but it is not equipped for hardware ray tracing or AI-accelerated rendering.