ATI Mobility Radeon HD 3870

Memory Subsystem

The ATI Mobility Radeon HD 3870 is equipped with 512 MB of GDDR3 memory, paired with a 256-bit memory bus. This configuration yields a memory bandwidth of 54.40 GB/s, a figure that reflects the mobile GPU's positioning within the M8x generation of Mobility HD 3800 parts. For high-resolution workloads, the 256-bit interface is a critical asset, as it allows the memory controller to feed the shader array with data more efficiently than narrower-bus designs of the same era.

The effective memory speed is rated at 1700 Mbps, with the memory clocked at 850 MHz. In practice, this translates to a bandwidth ceiling that can sustain 1080p-class textures and moderate anti-aliasing without severe bottlenecks, though the 512 MB capacity may prove restrictive for extremely large texture sets in later titles. The combination of a 256-bit bus and GDDR3 technology was a standard for performance mobile GPUs in this period, delivering a balance between power consumption and throughput.

Given the 55 nm process node from TSMC, the memory controller and the rest of the die are manufactured with a transistor count of 666 million on a 192 mm² die. The transistor density of 3.5M per mm² is modest by modern standards, but the memory subsystem's efficiency is more dependent on the bus width than on process scaling. For users targeting high resolutions, the data indicates that the 54.40 GB/s bandwidth is a limiting factor compared to desktop parts of the same generation, yet it remains adequate for the laptop displays of its time, which typically peaked at 1920x1200.

How It Compares

The FACT PACK provides no nearestRivals entries for this GPU. Consequently, direct comparative analysis against specific competitor models cannot be derived from the available data. The percentile ranking of 50 against all GPUs indicates that this part sits at the median of the entire GPU landscape, meaning roughly half of all GPUs ever benchmarked are faster and half are slower. This is a useful positional marker, but it lacks the granularity of a head-to-head comparison.

Without rival names or deltaPct values, the analysis must rely on the broader context of the M8x generation. The Mobility Radeon HD 3870 is a successor to the M7x family and precedes the M9x family, placing it as a mid-generation refresh. Its architecture, TeraScale, was AMD's unified shader design, which replaced older separate vertex and pixel pipelines. The 320 shading units, 16 texture mapping units, and 16 ROPs define its raw compute capacity, but without rival data, these numbers cannot be contextualized as percentages ahead or behind.

The absence of benchmark scores (avgBenchmarkScore is 0) further limits quantitative comparison. The percentile of 50 is the only statistical anchor, suggesting that in a historical database of GPU performance, this mobile part lands exactly at the midpoint. For a notebook GPU from mid-2008, that is a reasonable standing, as mobile parts typically lagged their desktop counterparts by a significant margin, but the data does not allow for a more precise statement.

Benchmark Performance

The benchmark data for the ATI Mobility Radeon HD 3870 is sparse: the FACT PACK lists an average benchmark score of 0 and a percentile of 50. A score of zero is anomalous, likely indicating that no standardized benchmarks were recorded in the database for this specific SKU, or that the part was not subjected to the full test suite. The percentile of 50, however, is a meaningful data point, as it represents the GPU's standing relative to all other GPUs in the database.

Interpreting the percentile: a score of 50 means the Mobility HD 3870 outperforms half of all GPUs cataloged. In practice, this places it above entry-level integrated graphics and older discrete parts, but below the high-end desktop GPUs of its day. The raw compute metrics from the FACT PACK support this: the FP32 performance is 422.4 GFLOPS, the pixel rate is 10.56 GPixel/s, and the texture rate is 10.56 GTexel/s. These figures indicate a part that can handle contemporary games at medium settings, but not at maximum details with high resolutions.

The lack of nearestRivals data means that percentage deltas (deltaPct) cannot be computed. In a typical hardware analysis, one would report that the HD 3870 is, for example, 30% faster than a specific rival in multi-core synthetic tests. Here, no such statements are possible. The only valid comparative statement is the percentile-based one: it sits at the 50th percentile, meaning it is exactly average relative to the entire GPU population. This is a neutral position, neither a performance leader nor a laggard.

For mobile users of the era, this translated to playable frame rates in DirectX 10 titles at 1366x768 or 1440x900, with reduced detail settings. The 10.56 GPixel/s pixel fill rate and 10.56 GTexel/s texture fill rate are balanced, indicating no obvious bottleneck between the ROPs and TMUs. The 422.4 GFLOPS FP32 throughput is the limiting factor for shader-heavy workloads, but it is consistent with a 55 nm, 55 W mobile part.

FAQ

Q: What is the memory configuration of the ATI Mobility Radeon HD 3870?

A: It ships with 512 MB of GDDR3 memory on a 256-bit bus, delivering a bandwidth of 54.40 GB/s.

Q: What is the transistor count and die size of this GPU?

A: The M88 chip contains 666 million transistors on a 192 mm² die, manufactured on a 55 nm process at TSMC.

Q: What is the pixel fill rate?

A: The pixel fill rate is 10.56 GPixel/s, while the texture fill rate is also 10.56 GTexel/s.

Q: Does this GPU support DirectX 10.1?

A: Yes, it supports DirectX 10.1 (10_1) and OpenGL 3.3. It does not support Vulkan.

Q: What is the thermal design power (TDP)?

A: The TDP is 55 W, which is typical for a high-performance mobile GPU of its generation.

Q: What is the release date and production status?

A: It was released on June 3, 2008, and is currently marked as end-of-life in production.

Ray Tracing and Feature Set

The ATI Mobility Radeon HD 3870 does not include dedicated ray tracing cores or tensor cores, as those technologies were not introduced until much later in GPU architecture evolution. The FACT PACK lists rtCores and tensorCores as null values, confirming their absence. Instead, this GPU relies on the TeraScale architecture's unified shader design, which allocates all 320 shading units to general-purpose compute, vertex processing, and pixel shading as needed.

The API support is limited to DirectX 10.1 (with a shader model of 10_1) and OpenGL 3.3. Vulkan is not supported, which is expected given the 2008 release date, as Vulkan was not finalized until 2016. The DirectX 10.1 support is a notable feature for its time, as it enabled more advanced per-pixel lighting and shadow techniques compared to DirectX 10.0. However, it lacks the hardware-accelerated ray tracing features found in modern GPUs.

The feature set also includes a PCIe 2.0 x16 bus interface, which was the standard for high-bandwidth communication with the host system at the time. Display outputs are listed as "Portable Device Dependent," meaning the exact ports (e.g., HDMI, VGA, DisplayPort) depend on the laptop manufacturer's implementation. The GPU's 16 ROPs and 16 TMUs handle rasterization and texture filtering, respectively, and the 422.4 GFLOPS FP32 throughput is the ceiling for shader computations.

For users interested in modern features like variable rate shading, mesh shaders, or hardware ray tracing, this GPU offers none of those. It is a pure rasterization engine, designed for the DirectX 10 era. The absence of Vulkan support also precludes modern cross-platform graphics APIs. In summary, the feature set is firmly rooted in its 2008 timeframe, with no forward-looking capabilities beyond DirectX 10.1 and OpenGL 3.3.