ATI Radeon 9800

The ATI Radeon 9800, built on TSMC's 150 nm process with 117 million transistors on a 215 mm² die, is a single-slot AGP 8x card from the Radeon R300 generation, released on 2003-02-28. It carries 128 MB of DDR memory on a 256-bit bus, delivering 18.56 GB/s of bandwidth, and sits at the 50th percentile of all GPUs in the database — a median position that reflects its era rather than any modern standing. With no recorded benchmark scores in the database, its evaluation rests entirely on its architectural specifications and feature set.

Benchmark Performance

The database lists no benchmark scores for the ATI Radeon 9800, and its average benchmark score is recorded as zero. This absence of data means the card cannot be ranked against specific rivals through measured deltas; instead, its performance profile must be inferred from its fixed-function pipeline. The card pairs 8 texture mapping units with 8 render output units, producing a pixel rate of 2.600 GPixel/s and a texture rate of 2.600 GTexel/s. These identical rates indicate a balanced fill-rate design, where pixel and texel throughput are matched — a common trait for cards of this generation that targeted DirectX 9.0 workloads at resolutions where fill-rate was the primary bottleneck.

The 256-bit memory bus is the standout feature here. At a memory clock of 290 MHz (580 Mbps effective), the card achieves 18.56 GB/s of bandwidth — a figure that would have been substantial for its release period, and one that directly supports the 2.600 GPixel/s pixel rate. A narrower bus would have starved the ROPs; the 256-bit interface ensures that texture-heavy scenes, which demand simultaneous pixel writes and texture fetches, do not stall on memory latency. The 128 MB frame buffer is modest by later standards, but for a 2003-era card it aligns with the DirectX 9.0 feature level (9_0) that the card supports.

Because the nearestRivals array is empty, there are no deltaPct values to report. The card's 50th percentile placement across all GPUs in the database is a median ranking, suggesting it outperforms roughly half of the entries and trails the other half — a reasonable outcome for an end-of-life product whose architecture predates unified shaders, hardware tessellation, and all modern compute APIs. In practical terms, the Radeon 9800 would be competitive within its own generation, but the data does not support any specific percentage comparison against named competitors.

Ray Tracing and Feature Set

The Radeon 9800 has no dedicated ray tracing cores and no tensor cores — the JSON fields for both are null. This is expected for a 2003 GPU, as hardware-accelerated ray tracing did not appear in consumer graphics until much later. The card's feature set is defined entirely by its API support: it implements DirectX 9.0 at the 9_0 shader model level and OpenGL 2.0. There is no Vulkan support, which is consistent with an architecture that predates the Vulkan specification by over a decade.

The DirectX 9.0 (9_0) support is the defining feature of this generation. It enables programmable pixel and vertex shaders, which the R350 chip (the "R350" die) exposes through its 8 TMUs and 8 ROPs. The absence of RT and tensor cores means the card cannot accelerate ray-traced lighting, denoising, or AI-based upscaling — none of which existed as consumer features at the time. The display outputs are equally period-appropriate: 1x DVI, 1x VGA, and 1x S-Video, covering analog and early digital connections. The card's OpenGL 2.0 support would have allowed contemporary games and professional OpenGL applications to run, but with no Vulkan path, modern titles are entirely out of reach.

The 150 nm process node and 117 million transistor count place the Radeon 9800 in an era where fixed-function and early programmable pipelines coexisted. Its 544.2K transistors per mm² density reflects the manufacturing limits of TSMC's 150 nm process. The feature set is complete for its time — DirectX 9.0 and OpenGL 2.0 cover the software ecosystem of the early 2000s — but it offers nothing for modern ray-traced or tensor-accelerated workloads.

Power and Cooling

The Radeon 9800 carries a thermal design power of 37 W, which is exceptionally low by contemporary standards but typical for a mid-range GPU of its generation. This modest TDP is reflected in the cooling solution: the card is single-slot, requiring no oversized heatsink or dual-fan arrangement. A single Molex power connector supplies the card, and the suggested power supply rating is 200 W — a figure that any modern desktop PSU would exceed by a wide margin, but which was a reasonable requirement in 2003 when system power draw was far lower.

The 37 W TDP means the card generates little waste heat, so a single-slot cooler is sufficient for sustained operation. The 1x Molex connector is the only external power input, indicating that the AGP 8x bus slot itself provides the remainder of the card's power. The 200 W suggested PSU is a system-level recommendation, accounting for the rest of the platform (CPU, drives, motherboard) rather than the GPU alone. Users upgrading from older cards would need to verify their PSU had a spare Molex connector, but the low power draw makes the card compatible with a wide range of systems from its era.

There is no length, height, or width data in the database, so physical clearance cannot be quantified. However, the single-slot design and low TDP suggest the card would fit in most AGP-era cases without issue. The end-of-life production status means the card is no longer manufactured, but its power requirements remain historically relevant for collectors or retro-builders assessing whether their existing PSU can handle it.

FAQ

Q: What DirectX version does the ATI Radeon 9800 support?

A: The card supports DirectX 9.0 at the 9_0 feature level, along with OpenGL 2.0. It does not support Vulkan.

Q: How much memory does the Radeon 9800 have, and what is its bus width?

A: It has 128 MB of DDR memory on a 256-bit bus, delivering 18.56 GB/s of bandwidth at a memory clock of 290 MHz (580 Mbps effective).

Q: Does the Radeon 9800 support ray tracing?

A: No. The card has no ray tracing cores and no tensor cores. It relies entirely on its 8 TMUs and 8 ROPs for rendering.

Q: What is the TDP of the Radeon 9800, and what power supply is recommended?

A: The TDP is 37 W, and the suggested PSU rating is 200 W. The card uses a single Molex power connector.

Q: What display outputs does the card provide?

A: The Radeon 9800 offers 1x DVI, 1x VGA, and 1x S-Video outputs.

Q: What is the production status of the Radeon 9800?

A: It is marked as end-of-life. It was released on 2003-02-28, with the Radeon R200 as its predecessor and the Radeon R400 AGP as its successor.

Q: How many texture mapping units and render output units does it have?

A: It has 8 TMUs and 8 ROPs, producing a texture rate of 2.600 GTexel/s and a pixel rate of 2.600 GPixel/s.

How It Compares

The database lists no nearest rivals for the ATI Radeon 9800, so there are no named competitors with score deltas to cite. Its 50th percentile ranking across all GPUs places it at the median of the entire database — a position that is more a reflection of the database's mix of modern and legacy hardware than of the card's own capabilities. Within its own generation, the Radeon 9800's 256-bit memory bus and 18.56 GB/s bandwidth would have positioned it above cards with narrower buses, but without rival entries in the nearestRivals field, no quantitative comparison is possible.

Against its predecessor, the Radeon R200, the 9800 represents a generational step forward in API support (DirectX 9.0 versus the older DirectX 8-era feature set) and memory bandwidth. Its successor, the Radeon R400 AGP, would eventually supersede it, but the 9800's 150 nm process and 117 million transistors were competitive for its release window. The absence of ray tracing and tensor cores is not a disadvantage in this context — no 2003 GPU had them — but it does mean the card cannot be compared on modern feature sets.

The 37 W TDP and 200 W PSU recommendation set the Radeon 9800 apart from both older and newer cards: it draws far less than the multi-hundred-watt GPUs of later generations, yet requires more than the slot-powered cards that preceded it. The single-slot cooler and single Molex connector keep installation simple. In the absence of rival data, the card's legacy is best summarized by its balanced fill-rate (2.600 GPixel/s and 2.600 GTexel/s), its generous-for-the-era 256-bit memory bus, and its median position in the database's performance distribution. It is a competent mid-range part from a transitional period in GPU architecture, neither a flagship nor a budget offering, and its end-of-life status confirms its historical role.