Intel Arc Graphics 32EU

Power and Cooling — TDP, PSU recommendation, connector requirements

The Intel Arc Graphics 32EU is an integrated graphics processor (IGP) built for the Arrow Lake-S platform. Its thermal design power is rated at 65 W, which is a modest figure for a chip that includes the CPU, GPU, and memory controller on a single package. Because this is an IGP, it does not carry dedicated power connectors; the slot width is listed as "IGP," meaning it occupies no expansion slot and draws power entirely through the motherboard's socket delivery. Consequently, there is no suggested PSU rating from the fact pack, and the system power supply is dictated by the rest of the platform components rather than this GPU.

The chip is fabricated on a 3 nm process at TSMC, with a die size of 243 mm² and a transistor count of 17,800 million. This yields a transistor density of 73.3 million transistors per square millimeter. The 65 W TDP is the only power figure provided, and it represents the entire package's thermal envelope, not just the graphics portion. Since the memory subsystem uses "System Shared" memory, the GPU's power draw varies with system memory bandwidth and capacity, but the fact pack provides no separate figures for memory-related power consumption.

For cooling, the fact pack offers no specific cooler recommendations. The 65 W TDP suggests that a capable air cooler designed for mainstream desktop processors should suffice, but no numeric thermal solution specifications are given. The bus interface is a Ring Bus, which is typical for integrated graphics within a CPU die, and the display outputs are listed as "Motherboard Dependent," meaning the available ports (HDMI, DisplayPort, etc.) are determined by the specific motherboard model rather than the GPU itself.

Ray Tracing and Feature Set — RT/tensor cores, API support from facts

The Intel Arc Graphics 32EU does not list dedicated ray tracing cores or tensor cores in the fact pack. The architecture is Xe-LPG, which is the low-power variant of Intel's Xe graphics architecture, and it includes 256 shading units, 16 texture mapping units (TMUs), and 8 raster output units (ROPs). The absence of RT and tensor core counts indicates that hardware-accelerated ray tracing and AI tensor operations are not separate functional blocks in this implementation, though the API support suggests some software-level capabilities.

The API support is robust for an integrated solution. It supports DirectX 12 Ultimate (12_2), which is the highest tier of DirectX 12, including features like variable rate shading and mesh shaders when implemented in software. OpenGL 4.6 is supported, and Vulkan 1.4 is listed, making this a modern API-compliant part. The fact pack does not specify whether ray tracing is handled via compute shaders or is absent entirely, but the lack of dedicated RT cores points toward compute-based approaches if any ray tracing is attempted.

The pixel rate is 15.60 GPixel/s, and the texture rate is 31.20 GTexel/s. These figures are derived from the base clock of 300 MHz and boost clock of 1950 MHz, with the ROP and TMU counts respectively. The FP32 performance is 998.4 GFLOPS, and FP16 performance is 1.997 TFLOPS (2:1 ratio), meaning the FP16 throughput is roughly double the FP32 rate. These are not high figures by discrete GPU standards, but they are consistent with an integrated part aimed at basic rendering and compute tasks.

Benchmark Performance — analyze scores vs rivals with exact % deltas

The fact pack lists no benchmark scores, no average benchmark score, and no nearest rivals. The avgBenchmarkScore field is 0, and the percentileVsAllGpus field is 50, which places this GPU at the median of all GPUs in the database. However, because there are no actual benchmark entries, this percentile is based on the hardware specifications alone rather than measured performance. The absence of nearestRivals data means no direct percentage comparisons can be made to other specific GPUs.

Given the lack of benchmark scores, the analysis must rely on the raw compute figures. The FP32 throughput of 998.4 GFLOPS is the primary measure of general-purpose shader performance. To contextualize this, the pixel rate of 15.60 GPixel/s and texture rate of 31.20 GTexel/s indicate that the 8 ROPs and 16 TMUs are the limiting factors for fill-rate-bound workloads. For a 256-shader-unit design at a 1950 MHz boost clock, these figures are internally consistent: 256 shaders × 2 operations per clock × 1.95 GHz ≈ 998.4 GFLOPS, and 16 TMUs × 1.95 GHz = 31.2 GTexel/s.

The percentile rank of 50 out of 100 suggests this is a mid-pack performer among all GPUs tracked, but the near-zero benchmark score field indicates that no empirical data has been collected. This is typical for newly released integrated graphics. The production status is "Active," and the release date is 2024-10-23, so the part is current. Without rival scores, the performance assessment is provisional: the 65 W TDP and 3 nm process suggest efficiency, but the absolute performance is limited by the 256 shader count and shared memory bandwidth.

Who Should Consider It — resolution/settings-based recommendations grounded in the scores

The Intel Arc Graphics 32EU is an integrated GPU with no dedicated VRAM; it uses system shared memory, and the bandwidth is "System Dependent." This means performance at high resolutions and detail settings will be constrained by the system's memory configuration. At 1080p, the FP32 throughput of 998.4 GFLOPS and pixel rate of 15.60 GPixel/s are sufficient for older or less demanding titles at low to medium settings, but the lack of benchmark data prevents precise settings recommendations.

For 1440p and above, the 8 ROPs will become a bottleneck for fill-rate-intensive effects like high-resolution shadows and post-processing. The texture rate of 31.20 GTexel/s is modest, so texture-heavy scenes will strain the TMUs. The DirectX 12 Ultimate and Vulkan 1.4 support means modern APIs are available, but the hardware's raw throughput is roughly one teraflop, which is well below what contemporary discrete GPUs offer.

Given the 50th percentile ranking, this GPU is positioned for basic desktop use, light gaming, and media playback rather than high-end gaming. Users with a 65 W TDP platform who need a display output without a discrete graphics card are the target audience. Systems with slower system memory will see lower performance, as the memory bandwidth is entirely dependent on the platform's RAM configuration. For 1080p esports titles at low settings, the data suggests playable frame rates, but no specific frame rate numbers are available to confirm this.

Memory Subsystem — VRAM size/type, bus width, bandwidth and what it means for high resolutions

The memory subsystem is entirely "System Shared," with no dedicated VRAM. The size, type, and bus width are all listed as "System Shared," and the bandwidth is "System Dependent." This is a fundamental difference from discrete GPUs, which have their own memory with fixed bandwidth characteristics. For the Intel Arc Graphics 32EU, the GPU accesses the same system memory as the CPU, meaning the available bandwidth is shared with the processor and varies with the memory configuration (e.g., dual-channel vs. single-channel, DDR4 vs. DDR5).

The practical consequence is that high-resolution gaming and large texture loads will be constrained by system memory bandwidth. At 1080p with moderate texture quality, the shared memory may suffice, but at 1440p or 4K, the bandwidth demand from the GPU will compete with the CPU for memory access, potentially causing stuttering or reduced frame rates. The fact pack provides no specific bandwidth figure, so the "System Dependent" label is the only guidance—it implies that a system with faster memory will yield better GPU performance.

For the FP32 throughput of 998.4 GFLOPS, the memory bandwidth requirement is roughly proportional: a GPU of this compute level typically needs 50-100 GB/s of bandwidth for balanced performance, but no such figure appears in the fact pack. The system shared memory also means the maximum VRAM capacity is whatever the system has installed, minus the OS and application usage. This makes the part unsuitable for workloads with large working sets, such as high-resolution texture packs or machine learning inference, unless the system has ample RAM.

FAQ

Q: What is the TDP of the Intel Arc Graphics 32EU?

A: The TDP is 65 W, which is the thermal envelope for the entire Arrow Lake-S package, not just the GPU portion.

Q: Does this GPU have dedicated ray tracing cores?

A: No, the fact pack lists no RT core count. The architecture is Xe-LPG with 256 shading units, 16 TMUs, and 8 ROPs, but no separate RT or tensor core blocks are specified.

Q: What API versions are supported?

A: The GPU supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: How much VRAM does it have?

A: It has no dedicated VRAM. The memory size, type, and bus width are all "System Shared," meaning it uses the system's main RAM.

Q: What is the boost clock speed?

A: The boost clock is 1950 MHz, with a base clock of 300 MHz.

Q: What is the FP32 performance in GFLOPS?

A: The FP32 throughput is 998.4 GFLOPS, and the FP16 throughput is 1.997 TFLOPS (2:1 ratio).

Q: When was it released?

A: The release date is 2024-10-23, and the production status is "Active."