AMD Radeon HD 7660G IGP

AMD Radeon HD 7660G IGP is an integrated graphics processor from AMD. Its architecture is TeraScale 3, and its generation is listed as “TeraScale 3 IGP (Trinity Mobile)”. The chip inside this product is named Devastator. The process node is 32 nm and the foundry is GlobalFoundries. The die contains 1,303 million transistors on a 246 mm² area, giving a transistor density of 5.3M / mm². This product was released on 2012-05-14 and its production status is end-of-life. The predecessor field is TeraScale 2 IGP, and the successor field is GCN 2.0 IGP.

Memory Subsystem

The memory fields in the fact pack are defined by shared-system values rather than by fixed specifications. The memory size is “System Shared”, the memory type is “System Shared”, and the bus width is “System Shared”. The bandwidth is “System Dependent”. The memory clock entry is also “System Shared”. There is no numeric VRAM size in the record, no fixed bus width, and no rated bandwidth. The memory subsystem therefore cannot be described as a discrete frame buffer. For high resolutions, the data set provides no fixed capacity figure and no guaranteed bandwidth. The only memory-related labels in the record are “System Shared” for the size, type, and bus width, plus “System Dependent” for bandwidth. What this means is that the GPU’s memory behavior will track the system’s memory behavior. The system memory parameters themselves are not in the fact pack, so no high-resolution capacity or bandwidth claim can be derived.

In addition to the memory fields, the core has fixed output rates: pixel rate is 5.488 GPixel/s and texture rate is 16.46 GTexel/s. These are the listed throughput ceilings for the pixel and texture paths. There are no other memory or clock values in the data set; the base, boost, and game clocks are all null, leaving “System Shared” as the only clock-related entry. The absence of a numeric bus width means the data path width cannot be compared with other products. The absence of a numeric bandwidth value means any memory throughput number would have to be supplied by the host system it is integrated into. In this record, the memory subsystem is described entirely through shared and dependent qualifiers.

Ray Tracing and Feature Set

The fact pack has no RT core entry and no tensor core entry; both fields are null. The architecture is TeraScale 3. API support is stated as DirectX 11.2 (11_0) and OpenGL 4.4. The Vulkan field is null, so no Vulkan version is listed. The GPU has 384 shading units, 24 texture mapping units, and 8 ROPs. The FP32 throughput is 526.8 GFLOPS, while the FP16 value is null. The pixel rate is 5.488 GPixel/s and the texture rate is 16.46 GTexel/s.

With no RT cores, the record contains no dedicated ray tracing acceleration hardware. With no tensor cores, there is no tensor acceleration entry in the data set. Because Vulkan is null, the supported API surface in the record is limited to the DirectX and OpenGL entries. The DirectX field itself is “11.2 (11_0)”, which distinguishes the API version and feature level in the fact pack. The OpenGL field is 4.4. The throughput values — 526.8 GFLOPS, 5.488 GPixel/s, and 16.46 GTexel/s — are the quantitative feature-set markers available for this IGP. No FP16 value is present, so half-precision throughput cannot be reported from the data set.

Power and Cooling

The TDP is 35 W. The slot width is IGP, and the bus interface is also IGP. No power connectors are listed. No suggested PSU is listed. Display outputs are described as “Portable Device Dependent”. The dimensions fields, including length, height, and width, are all null. This is an integrated part rather than a discrete expansion card. The power connector field being null means the fact pack does not specify any external power connector requirement. The suggested PSU field being null means no power supply recommendation is provided.

The cooling side of the record is similarly limited: there is no cooler specification, no board dimensions, and no slot-based installation details. The 35 W TDP is the only power-related number in the fact pack. The IGP slot width and IGP bus interface indicate that the product is not installed in a full-length slot. The portable-device-dependent display output field indicates that the display connections belong to the portable device rather than to the GPU itself. Together, the 35 W TDP and IGP designation describe a part that is integrated into a larger system and cooled within that system’s design envelope. No other power or cooling figures are present.

How It Compares

The nearestRivals field in the fact pack is an empty list. There are therefore no nearest rival names, no rival scores, and no deltaPct values to use for comparison. The only broad position metric is percentileVsAllGpus, which has a value of 50. That percentile places the product at the midpoint of the all-GPU percentile field. However, the avgBenchmarkScore field is 0, and the benchmarks array is empty.

In the product sequence, the predecessor is TeraScale 2 IGP and the successor is GCN 2.0 IGP. This sequence gives a positional context: it sits after a TeraScale 2 IGP generation and before a GCN 2.0 IGP generation. That is a lineage relationship, not a performance comparison. No rival-by-rival paragraphs can be written from the fact pack because no rivals are listed. The percentile value is the only comparative number in the record, and it is not supported by any benchmark score entries. Without nearestRivals, the data set cannot justify statements about being faster or slower than any specific product.

Benchmark Performance

The benchmarks array in the fact pack is empty, so no benchmark entries are available. The avgBenchmarkScore field is 0. There are no nearestRivals entries, so there are no deltaPct values. Without a benchmarks list, there are no scores to analyze against any other product. Without nearestRivals, there are no exact percentage deltas to quote. The only numeric performance-adjacent fields are avgBenchmarkScore 0 and percentileVsAllGpus 50. The percentile is the only comparative position, but it is unaccompanied by benchmark data. The data cannot support statements such as “30% ahead of a rival” or “20% behind a rival”. The fixed rates of 5.488 GPixel/s, 16.46 GTexel/s, and 526.8 GFLOPS are throughput values from the specification, not benchmark results. There are no recorded frame rates and no recorded synthetic scores. The benchmark performance section of this product record contains no measurements to interpret.

Who Should Consider It

Because there are no benchmark scores, this section cannot ground any recommendation in measured performance. The available grounds are the memory fields, the API fields, the feature fields, and the power and form-factor fields. The memory size is System Shared, the memory type is System Shared, and the bus width is System Shared, with System Dependent bandwidth. A user considering high resolutions has no fixed VRAM figure to plan around; the system memory will determine the frame buffer behavior. The API list includes DirectX 11.2 (11_0) and OpenGL 4.4, with no Vulkan value. Software that requires Vulkan is not represented in the feature set. With no RT cores and no tensor cores, hardware-accelerated ray tracing and tensor features are not present in the record. The compute and throughput resources are 384 shading units, 24 TMUs, 8 ROPs, and 526.8 GFLOPS FP32. These values define the execution resources available to supported workloads.

The TDP is 35 W, the slot width is IGP, and the display outputs are Portable Device Dependent. That points to a portable-device integration rather than a discrete add-in card. The production status is end-of-life, and the release date is 2012-05-14. The predecessor is TeraScale 2 IGP and the successor is GCN 2.0 IGP. For a potential user, the summary from the data is that no score-based resolution or settings recommendation exists in this fact pack. Any choice would have to be made on the basis of the shared memory behavior, the DirectX 11.2/OpenGL 4.4 API list, the 35 W TDP, and the absence of RT, tensor, and Vulkan entries.