Intel Iris Pro Graphics 580 Mobile

Intel Iris Pro Graphics 580 Mobile is an integrated graphics processor built on Intel's 14 nm+ process. Its chip is Skylake GT4e, architecture Generation 9.0, and it belongs to the HD Graphics-M (Skylake) generation. Released on 2015-08-31, it is now marked end-of-life. The data set lists no benchmark entries: the average benchmark score is 0 and the nearestRivals array is empty. The only relative ranking provided is percentileVsAllGpus, which is 50. That means the specification-level throughput figures below are the only numerical performance boundaries available.

Benchmark Performance

The benchmark section of the data contains no entries. Because the benchmarks array is empty and the average benchmark score is 0, no measured application-level score can be cited. The nearestRivals field is also empty, so there are no rival names, no rival scores, and no deltaPct values available for direct percentage comparisons. Any statement like "30% faster than X" would have no support in this FACT PACK.

What remains are theoretical throughput rates. The GPU has 576 shading units, 72 TMUs, and 9 ROPs. Its base clock is 350 MHz and its boost clock is 950 MHz; no game clock is listed. From those numbers, the data reports FP32 performance of 1,094.4 GFLOPS and FP16 performance of 2.189 TFLOPS at a 2:1 ratio. The 72 TMUs yield a texture rate of 68.40 GTexel/s, and the 9 ROPs yield a pixel rate of 8.550 GPixel/s. The FP16 throughput is presented as double the FP32 figure, which is a useful architectural signal even though no compute benchmark confirms it in practice.

These figures should be read as ceilings, not guarantees. A 950 MHz boost with 576 shading units gives a specific compute envelope, but the database does not translate that envelope into frames-per-second numbers. Likewise, 8.550 GPixel/s and 68.40 GTexel/s are fill-rate limits; actual rendering performance depends on workloads the data does not include.

The 50th-percentile ranking is curious because it is not accompanied by an average score. A median position among all GPUs suggests a middle-of-pack placement, but the absence of benchmark results leaves the basis of that percentile unclear. Without nearestRivals data, the benchmark analysis cannot go beyond raw throughput and a single percentile value.

Ray Tracing and Feature Set

The data fields for rtCores and tensorCores are null. No ray tracing cores are listed, and no tensor cores are listed. That means this part has no documented hardware path for ray traversal acceleration or tensor-style workloads. The API list is separate: DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. These are the supported graphics APIs in the data, and they give a sense of software compatibility. DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3 are all present as capability markers.

Notably, the architecture is Generation 9.0, built on the Skylake GT4e chip at 14 nm+. The bus interface is Ring Bus. Display outputs are listed as Portable Device Dependent, meaning the actual output configuration depends on the host portable device rather than being fixed on the GPU itself. For ray tracing, the absence of RT cores is the central fact: no hardware accelerator is declared, and the data provides no ray tracing benchmark results. API support alone does not establish real-time ray tracing performance.

The feature set is otherwise defined by its fixed-function units: 72 TMUs and 9 ROPs. Those are modest numbers in absolute terms, but the data does not include a comparison that says how modest. Vulkan 1.3 is the newest API in the list, yet no workload results show what that support enables in practice.

Memory Subsystem

The memory subsystem is entirely system-shared. Memory size is System Shared, memory type is System Shared, bus width is System Shared, and bandwidth is System Dependent. There is no dedicated VRAM capacity listed, no dedicated memory type, and no fixed bus width. Even the memory clock field is System Shared.

This has direct implications for high-resolution work. A dedicated frame buffer with fixed bandwidth would allow more predictable performance; this part has none in the data. Instead, the same system memory that the CPU uses also serves the GPU. The pixel rate is fixed at 8.550 GPixel/s, but the bandwidth needed to feed those pixels is not fixed. Consequently, high-resolution performance could vary meaningfully from one host system to another, depending on the system memory implementation. The data does not provide enough information to quantify that variation.

The ROP count of 9 is low alongside 72 TMUs and 576 shading units. At 950 MHz boost, the pixel rate is 8.550 GPixel/s. That suggests fill operations could be a limiting factor for high-resolution scenes, but without memory bandwidth numbers, the precise bottleneck cannot be isolated. Shared memory also means there is no separate VRAM pool to hold textures and geometry; large assets would rely on the host's memory capacity and bandwidth. The data does not state how much system memory can be used, nor does it describe allocation behavior.

FAQ

Q: What is the production status?

A: The production status is end-of-life.

Q: How much dedicated video memory does it have?

A: None is listed. Memory size, type, bus width, and memory clock are all listed as System Shared, with bandwidth as System Dependent.

Q: Does it support ray tracing?

A: The rtCores and tensorCores fields are null, so no ray tracing cores or tensor cores are documented. The API list includes DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3.

Q: What are the clock speeds?

A: The base clock is 350 MHz and the boost clock is 950 MHz. No game clock is listed.

Q: What is the TDP?

A: The TDP is 15 W. No power connectors and no suggested PSU are listed.

Q: When was it released?

A: The release date is 2015-08-31.

Who Should Consider It

The data does not contain a single benchmark score, so no exact resolution-and-settings recommendation can be grounded in measured performance. What the data does contain are throughput limits: 1,094.4 GFLOPS FP32, 68.40 GTexel/s texture rate, and 8.550 GPixel/s pixel rate. Those figures point to a modest integrated solution for portable devices. Display outputs are Portable Device Dependent, meaning the target hardware is a laptop or mobile system rather than a fixed desktop display configuration.

For high-resolution gaming, the absence of dedicated VRAM and the system-dependent bandwidth are meaningful concerns. A high-resolution frame buffer requires more memory and more bandwidth, but the data does not specify how much bandwidth is available. The fixed pixel rate of 8.550 GPixel/s also places an upper bound on filled pixels, and with just 9 ROPs, pixel-fill work could become a limiting factor. The data cannot support a recommendation for high-detail, high-resolution use.

For lower resolutions and lighter settings, the 576 shading units and 72 TMUs may be sufficient, but again there is no game score to confirm it. Users who require verified performance should note that the benchmark list is empty. The 15 W TDP and IGP slot width make it plausible for thin-and-light designs, but performance conclusions remain specification-based rather than score-based.

Power and Cooling

The only power figure in the data is TDP, which is 15 W. That is a low thermal envelope, but the data does not include cooler size, fan requirements, or chassis cooling specifications. Slot width is IGP, meaning the part is integrated rather than a discrete expansion card. The power connectors field is null, so no auxiliary power connector requirement is documented. The suggested PSU field is also null, so the database provides no power-supply recommendation.

Base clock 350 MHz and boost clock 950 MHz are the only clock figures tied to power behavior. No additional power states are listed. The absence of power connectors and a suggested PSU is consistent with an integrated design, but the data itself does not state that they are unnecessary. Cooling requirements are best described as modest; 15 W is the only number that defines the thermal scope.

How It Compares

The nearestRivals field in the data is empty. There are no rival names, no rival scores, and no deltaPct values to present. The only comparison point is percentileVsAllGpus, which is 50. That places this part at the midpoint of the database's all-GPU ranking, although the average benchmark score is 0 and no benchmark entries back the ranking. No predecessor or successor is listed, so there is no direct generational comparison available.

Because the nearestRivals array is empty, no per-rival paragraphs can be constructed from this data. This is a data limitation rather than a sign of parity with any specific GPU. The positional claim is limited to one percentile value and the raw throughput figures already covered. In the absence of rivals, the 50th-percentile rank and the empty benchmark results are the only comparative evidence the database supplies.