ATI Mobility FireGL 9000

The ATI Mobility FireGL 9000 is a mobile GPU in AMD’s Mobility FireGL generation, built on the Rage 7 architecture with the M9 chip. TSMC manufactured it on a 150 nm process with 36 million transistors on an 81 mm² die, producing a transistor density of 444.4K / mm². It connects through an AGP 4x interface and carries 64 MB of DDR memory on a 128-bit bus. Its listed API support covers DirectX 8.1 and OpenGL 1.4, and the production status is end-of-life, with FirePro Mobility listed as its successor.

Benchmark Performance

Benchmark data for this part is effectively absent. The benchmarks array is empty, the average benchmark score is 0, and the nearestRivals array contains no entries. This means no exact percentage deltas against named competitors can be computed from the FACT PACK. The only positional value available is percentileVsAllGpus, which is 50, placing the GPU at the midpoint of the database’s GPU population. That midpoint position, however, is not backed by any measured score in the record, so it should be treated as a database position marker rather than a validated performance result.

With no benchmark scores, throughput figures become the meaningful performance bounds. The GPU has 4 TMUs and 4 ROPs. The listed pixel rate is 1.000 GPixel/s, and the texture rate is also 1.000 GTexel/s. These matching values indicate a balanced fill-rate profile: the pixel backend and texture path are specified at the same throughput ceiling. Memory bandwidth is 6.400 GB/s, delivered over a 128-bit bus from DDR memory clocked at 200 MHz with a 400 Mbps effective data rate. That bandwidth is a tight limit for texture-heavy workloads, and the equal fill rates mean neither pixel writes nor texture reads have spare headroom over the other.

FP32 and FP16 throughput values are null in the record, and no shading unit count is listed. Without those data points, shader compute performance cannot be directly quantified. The 36 million transistor count and 81 mm² die size are structural facts about the chip, but they do not substitute for measured performance. Any comparison that relies on a tested score cannot be made for this entry.

Ray Tracing and Feature Set

The feature set is clearly limited by the Rage 7 architecture and the API entries in the data. RT cores are null, and tensor cores are null. There is no ray tracing acceleration and no tensor-based AI acceleration present in the record. The API list consists of DirectX 8.1 and OpenGL 1.4, with no Vulkan support. This restricts the GPU to software written for those feature levels.

The absence of Vulkan is particularly significant because it rules out modern cross-platform graphics workloads that depend on Vulkan. The absence of any DirectX version beyond 8.1 similarly caps the feature level for DirectX-based applications. Since FP32 and FP16 values are absent, compute throughput cannot be stated, but the API ceiling alone is enough to disqualify the card from modern rendering pipelines. This is not a compute-oriented accelerator; the data shows no dedicated blocks for ray tracing, tensor operations, or unspecified compute features.

Because RT and tensor cores are null, any workload that requires hardware-accelerated ray tracing or tensor processing will not run on this GPU. The only supported graphics paths are those compatible with DirectX 8.1 and OpenGL 1.4. That is a narrow compatibility window by current standards, and the feature set should be considered legacy-oriented.

How It Compares

The nearestRivals array is empty. There are no rival names, no rival scores, and no deltaPct values to report. Consequently, this entry cannot provide paragraph-by-paragraph comparisons against named competitors. The lack of nearest rivals means there is no measured competitive anchor in the data.

The only positional reference is percentileVsAllGpus, which is 50. That places the ATI Mobility FireGL 9000 at the exact middle of the database’s GPU list. In a normally populated entry, a 50th percentile would be supported by benchmark results; here, the average benchmark score is 0 and the benchmarks array is empty. Therefore, the percentile is an isolated marker rather than an outcome of tested workloads. It cannot be used to say that the GPU beats a specific percentage of rivals, because no such rivals are listed.

The successor field provides a generational reference: FirePro Mobility. The FACT PACK lists no scores for that successor, so a direct performance relationship between the two cannot be expressed as a percentage. The strongest comparison data in the pack is the compatibility profile: AGP 4x, DirectX 8.1, OpenGL 1.4, 64 MB DDR, 1.000 GPixel/s pixel rate, and 1.000 GTexel/s texture rate. Any other GPU would need its own entry with comparable data to form a meaningful head-to-head comparison.

Who Should Consider It

The ATI Mobility FireGL 9000 is for someone maintaining a legacy AGP 4x system that specifically needs DirectX 8.1 or OpenGL 1.4 support. It is not a candidate for modern, high-resolution rendering. The 64 MB frame buffer and 6.400 GB/s bandwidth are small enough that large frame buffers and texture sets will not fit comfortably, and the 1.000 GPixel/s pixel rate will limit fill-heavy settings.

Because no benchmark scores are present, resolution-specific recommendations cannot be grounded in measured data. The specifications indicate that modest detail settings and lower resolutions are the practical range. At higher demands, the frame buffer capacity, memory bandwidth, and fill rate will all act as bottlenecks. There is no path to ray-traced effects because RT cores are null, and no tensor-based features because tensor cores are null. The API support also prevents newer DirectX or Vulkan applications from running.

This GPU makes sense only for a narrow compatibility role. For software written to run under DirectX 8.1 or OpenGL 1.4, and for hardware built around AGP 4x, the Mobility FireGL 9000 can still be a relevant part. For anything outside that profile, the data in this FACT PACK is clear: the feature set and memory subsystem are too constrained.

FAQ

Q: What architecture does the ATI Mobility FireGL 9000 use?

A: It uses the Rage 7 architecture with the M9 chip. TSMC manufactures the chip on a 150 nm process with 36 million transistors on an 81 mm² die.

Q: How much memory does it have, and what kind?

A: It has 64 MB of DDR memory on a 128-bit bus. The memory clock is 200 MHz with a 400 Mbps effective data rate, resulting in 6.400 GB/s of bandwidth.

Q: Does it support ray tracing or tensor cores?

A: No. RT cores and tensor cores are null in the data. The API support is DirectX 8.1 and OpenGL 1.4, with no Vulkan listed.

Q: What bus interface does it use?

A: It uses AGP 4x.

Q: Is the ATI Mobility FireGL 9000 still in production?

A: No. Its production status is end-of-life, and the listed successor is FirePro Mobility.

Q: What are its fill rates?

A: The pixel rate is 1.000 GPixel/s, and the texture rate is 1.000 GTexel/s, based on 4 ROPs and 4 TMUs.

Memory Subsystem

The memory subsystem consists of 64 MB of DDR memory on a 128-bit bus. The physical memory clock is 200 MHz, and the effective data rate is 400 Mbps. This configuration yields a total memory bandwidth of 6.400 GB/s. That bandwidth is the available budget for all frame buffer reads, writes, and texture fetches.

High resolutions put pressure on both capacity and bandwidth. A 64 MB frame buffer cannot hold large color buffers, depth buffers, and textures at the same time. The 128-bit bus provides a moderate data path, but the effective memory clock keeps the final bandwidth at 6.400 GB/s. The GPU’s 1.000 GPixel/s pixel rate also limits how quickly rendering results can be written into memory. Together, these numbers describe a subsystem that is balanced for low-resolution, low-texture workloads but quickly strained by higher demands.

Because the memory type is DDR, the effective data rate is higher than the physical clock rate: 400 Mbps effective versus a 200 MHz clock. That distinction matters when calculating bandwidth. The listed width of 128 bit and the effective memory rate combine to produce 6.400 GB/s, which is the number that ultimately bounds texture streaming and frame buffer operations. For any workload that requires large amounts of data to move quickly, this is the hard ceiling.