NVIDIA Playstation 3 GPU 28nm

The NVIDIA Playstation 3 GPU 28nm, also known as the RSX-28nm, is a console-class graphics processor manufactured by Sony on a 28 nm process. It represents a die-shrink iteration of the original RSX architecture, built on NVIDIA’s Curie design with 302 million transistors on a 68 mm² die, yielding a transistor density of 4.4M per mm². This entry focuses on the chip’s measured specifications and its context within the console GPU segment, as the benchmark database shows no direct performance scores or rival comparisons for this specific variant.

Benchmark Performance

The RSX-28nm has no recorded benchmark scores in the database, and its average benchmark score is listed as 0, placing it at the 50th percentile of all GPUs tracked by the system. This percentile figure is not a measure of performance but rather a positional indicator, suggesting that the chip sits in the middle of the database’s distribution, largely because it is a console part with limited synthetic testing data. The absence of nearestRivals data means there are no deltaPct values to compare against other GPUs, so any performance interpretation must rely on the chip’s internal specifications rather than head-to-head results.

The pixel rate is 4.400 GPixel/s, and the texture rate is 13.20 GTexel/s, which are the primary throughput indicators available. These figures, when viewed together, suggest a GPU designed for a fixed-function console workload rather than a general-purpose desktop part. The 8 ROPs paired with 24 TMUs indicate a balanced but modest rasterization pipeline, typical of a late-generation console refresh meant to maintain compatibility with existing titles rather than push new graphical heights. Without FP32 or FP16 throughput values, the compute capability cannot be quantified, but the Curie architecture’s legacy design points to a shader model focused on older DirectX 9-era effects, not modern compute-heavy workloads.

Ray Tracing and Feature Set

The RSX-28nm does not include dedicated ray tracing cores or tensor cores, as these fields are null in the specification data. This absence is expected for a GPU based on the Curie architecture, which predates NVIDIA’s RT and Tensor core hardware by over a decade. The chip’s feature set is instead defined by its API support: it offers OpenGL ES 1.1, with no DirectX, Vulkan, or newer OpenGL versions listed. This limited API stack is significant for developers, as it restricts the GPU to older rendering paths and precludes any hardware-accelerated ray tracing effects.

The display output is a single HDMI 1.3 port, which supports 1080p video output but lacks the bandwidth for higher resolutions or refresh rates beyond early console standards. The absence of a bus interface field suggests this is an integrated or semi-integrated console part, not a discrete card with a standard PCIe interface. For feature set analysis, the data indicates a chip that is functionally frozen in time, designed to run existing console libraries at their original settings rather than to enable new graphical features. The lack of tensor cores also means no AI-accelerated upscaling or DLSS-style features are present, which is consistent with the chip’s 2013-era origins.

Who Should Consider It

Given the 256 MB GDDR3 memory and 20.80 GB/s bandwidth, the RSX-28nm is suited exclusively for legacy console gaming at 720p or 1080p with reduced settings. The memory capacity is severely constrained by modern standards, but it was sufficient for the PlayStation 3’s typical game loads, which were optimized for this exact hardware. The 4.400 GPixel/s fill rate and 13.20 GTexel/s texture rate are adequate for 720p rendering with moderate texture detail, but the chip will struggle with 1080p in demanding scenes due to the limited bandwidth and ROP count.

For resolution-based recommendations, the data supports 720p as the primary target, with 1080p achievable only in less complex titles or with significant visual compromises. The 8 ROPs are the bottleneck at higher resolutions, as pixel fill rate drops relative to the number of pixels being drawn. The 24 TMUs can feed textures at a reasonable rate for the era, but the 20.80 GB/s bandwidth will become a limiting factor when texture-heavy scenes require large data transfers. This chip is not suitable for modern gaming, productivity tasks, or any compute-oriented workload, as the API support and memory size preclude such use cases.

How It Compares

The database lists no nearest rivals for the RSX-28nm, meaning there are no direct comparison scores or deltaPct values to report. This absence is notable because the chip is a mid-cycle revision of the original RSX, typically compared to the base 65nm or 40nm versions, but those are not in the fact pack. Without rival data, a positional analysis is impossible; the chip’s 50th percentile ranking is a placeholder rather than a competitive measurement.

In the broader console GPU context, the RSX-28nm would be compared to other seventh-generation console GPUs, such as the Xbox 360’s Xenos, but again, no such data is provided. The lack of benchmarks means the chip’s performance relative to any other GPU cannot be stated with numbers. What the data does show is that the RSX-28nm is an end-of-life product with a 21 W TDP, which is remarkably low for a GPU with 302 million transistors, indicating a heavily power-optimized design for console thermal constraints. The 28 nm process node from Sony’s foundry is the key differentiator from earlier RSX variants, likely offering better power efficiency but no architectural changes.

Memory Subsystem

The RSX-28nm is equipped with 256 MB of GDDR3 memory on a 128-bit bus, operating at 650 MHz with an effective data rate of 1300 Mbps. This configuration yields a memory bandwidth of 20.80 GB/s, which is the single most critical limitation for high-resolution gaming. The 128-bit bus width is narrow by modern standards, but it was a deliberate choice for cost and power reasons in the console market. For 720p gaming, the 20.80 GB/s bandwidth is sufficient to feed the 4.400 GPixel/s pixel rate and 13.20 GTexel/s texture rate without major stuttering.

At 1080p, the bandwidth becomes a bottleneck, as the additional pixels require more texture fetches and color buffer writes. The 256 MB capacity is also problematic for modern game assets, but for the PlayStation 3’s library, it was a known constraint that developers worked around with aggressive streaming and compression techniques. The GDDR3 type is older and lower-bandwidth than the GDDR5 used in later consoles, but it is consistent with the 2013 release date. The memory clock of 650 MHz is modest, and the effective 1300 Mbps rate is half of what early GDDR5 parts offered, reinforcing the chip’s position as a low-power, cost-optimized console part rather than a high-performance GPU.

FAQ

Q: What is the transistor count and die size of the RSX-28nm?

A: The chip contains 302 million transistors on a 68 mm² die, produced on a 28 nm process by Sony’s foundry, giving a transistor density of 4.4M per mm².

Q: Does the RSX-28nm support ray tracing or tensor cores?

A: No, the specification data lists null values for both RT cores and tensor cores, confirming the absence of dedicated ray tracing or AI acceleration hardware.

Q: What is the maximum memory bandwidth available to the GPU?

A: The memory subsystem provides 20.80 GB/s bandwidth, derived from 256 MB of GDDR3 on a 128-bit bus running at 650 MHz with a 1300 Mbps effective rate.

Q: Which APIs are supported by this GPU?

A: The only API listed is OpenGL ES 1.1; there is no DirectX, Vulkan, or desktop OpenGL support indicated in the fact pack.

Q: What is the thermal design power (TDP) of the RSX-28nm?

A: The TDP is rated at 21 W, which is exceptionally low for a GPU with this transistor count, reflecting a power-optimized console design.

Q: What display outputs does the chip provide?

A: The RSX-28nm features a single HDMI 1.3 output, which supports standard high-definition video but lacks the bandwidth for 4K or high refresh rates.

Power and Cooling

The RSX-28nm has a TDP of 21 W, a figure that underscores its efficiency-oriented design for the PlayStation 3’s slim form factor. This low power draw is a direct result of the 28 nm process node, which reduced the die size from earlier RSX variants while maintaining the same 302 million transistors. The fact pack does not list a suggested PSU or power connector requirements, which is typical for a console GPU where the power delivery is integrated into the console’s internal power supply rather than a user-replaceable component.

The chip’s dimensions are listed as 290 mm in length, 60 mm in height, and 230 mm in width, which describe the physical envelope of the console board or the card’s mounting area, not a standard expansion card. The cooling solution is not specified, but the 21 W TDP allows for a passive or low-speed fan design, as the heat output is minimal compared to desktop GPUs of the same era. The lack of a slot width or bus interface field confirms this is not a standalone GPU but a fixed console component. For the end user, there is no PSU selection or connector compatibility to consider, as the console’s internal power supply handles all delivery. The low TDP also means the chip can sustain its maximum pixel and texture rates without thermal throttling, assuming the console’s cooling is functional.