NVIDIA GeForce GT 630 OEM

The NVIDIA GeForce GT 630 OEM is an end-of-life Kepler device from the GeForce 600 generation, fabricated by TSMC on a 28 nm process. The record identifies the GK107 chip, 1,270 million transistors, and a 118 mm² die size, with a transistor density of 10.8M/mm². The database entry contains no benchmark entries, no nearest rivals, an avgBenchmarkScore of 0, and a percentileVsAllGpus value of 50. Because measured scores are absent, every observation below is grounded in the specification block rather than in a performance-test result.

Benchmark Performance

The benchmarks array is empty, and the nearestRivals array is also empty. There are no rival names, no rival scores, and no deltaPct values to interpret. The only relative value in the record is percentileVsAllGpus, which is 50, and the only aggregate figure is avgBenchmarkScore, which is 0. Without a filled benchmark list, the percentile cannot be tied to any measured frame-rate or score distribution; it is a database position rather than a verified head-to-head result. Consequently, no percentage deltas against other GPUs can be reported from this data.

The raw throughput specifications provide the quantitative profile. The card has 192 shading units, 16 texture mapping units, and 16 raster operation units. Its FP32 compute rate is 336.0 GFLOPS. The texture rate is 14.00 GTexel/s, and the pixel rate is 3.500 GPixel/s. These numbers describe the maximum processing rates Recorded in the hardware specification. No base clock or boost clock is stored, so shader clock behavior is not part of the analysis. The only clock value in the entry is the memory clock of 891 MHz, with an effective data rate of 1782 Mbps. That makes the memory timing the sole clock-related anchor for the entire specification.

The absence of benchmark entries limits how strongly the raw rates can be interpreted. A GPU with these fill rates and compute figures can process small workloads within its stated limits, but the database does not provide evidence of sustained performance in any application. The 50 percentile value and 0 average score should be read as placeholders indicating that no sampled benchmark result exists, not as evidence of a competitive level. In short, the specification block defines an upper boundary of 336.0 GFLOPS, 14.00 GTexel/s, and 3.500 GPixel/s, while the actual performance position remains unmeasured in this record.

Ray Tracing and Feature Set

The rtCores and tensorCores fields are null. No dedicated ray tracing core count is recorded, and no tensor core count is recorded. Therefore, hardware-accelerated ray tracing and tensor-core-based features are not represented in the specification. This does not mean the GPU cannot render graphics; rather, the data contains no RT or tensor hardware to analyze.

API support is listed as DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The DirectX entry is important because it couples API 12 with feature level 11_0. The record therefore distinguishes between the API version and the hardware feature level. OpenGL 4.6 and Vulkan 1.2.175 are the other API compatibility points. These values define the software interface surface of the card, but they do not change the absence of RT and tensor core counts in the database.

The feature set also includes the physical and electrical characteristics. The card uses the Kepler architecture, with a 28 nm TSMC process, 1,270 million transistors, and a 118 mm² die. It is a single-slot card with a 50 W TDP, no power connectors, and a suggested PSU of 250 W. The host interface is PCIe 3.0 x16. Display outputs are 1x DVI, 1x HDMI 1.4a, and 1x VGA. The production status is end-of-life, and the release date is 2012-04-23. The predecessor is GeForce 500, and the successor is GeForce 700. These fields complete the platform-level picture: an older, low-power, single-slot card with a limited output set and no dedicated RT or tensor hardware listed.

Who Should Consider It

With no benchmark scores, a resolution-specific recommendation cannot be based on measured results. The specification does, however, make the hardware constraints explicit. The card has 1024 MB of DDR3 memory on a 128-bit bus, with 28.51 GB/s of bandwidth. It also has 192 shading units, 16 TMUs, and 16 ROPs. The FP32 rate is 336.0 GFLOPS, the texture rate is 14.00 GTexel/s, and the pixel rate is 3.500 GPixel/s.

The memory configuration is the most restrictive element for high-resolution work. A 1024 MB frame buffer is a fixed limit for any scene, while 28.51 GB/s of bandwidth governs how quickly texture and pixel data can move. Workloads with light memory pressure could run within the 336.0 GFLOPS and 14.00 GTexel/s limits, but large render targets and heavy texture use will hit either the capacity ceiling or the bandwidth ceiling first. The recorded output support includes a VGA connector, which suits legacy display devices, and HDMI 1.4a, which is an older HDMI revision.

The physical design is suited to constrained systems. The card is 145 mm long, or 5.7 inches, and uses one slot. It draws 50 W, has no auxiliary power connector, and pairs with a suggested 250 W PSU. That makes installation straightforward in systems with minimal power delivery. It is not, however, a current-generation recommendation: the production status is end-of-life, and the release date is 2012-04-23. The data does not support a claim of high-settings capability; it supports a narrow profile of legacy, low-power, modest-throughput use.

How It Compares

The nearestRivals array is empty. There are no rival GPUs in the record, and therefore no rival scores or deltaPct values. The usual per-rival comparison cannot be constructed from this fact pack because no comparison objects exist. The only nearby product relationships are the predecessor and successor fields, which list GeForce 500 and GeForce 700. These are family-level transitions, not scored benchmarks. The percentileVsAllGpus value is 50, but with no benchmark distribution behind it, that number cannot be used as evidence that the card sits at the midpoint of measured performance. The avgBenchmarkScore is 0, reinforcing that no average exists. Without nearestRivals entries, the database cannot produce a single exact percentage delta for this GPU. Any assertion that one GPU is faster or slower than another would require data not present in this entry.

Memory Subsystem

The memory subsystem is fully specified. Capacity is 1024 MB of DDR3. The bus width is 128 bit. The memory clock is 891 MHz, with 1782 Mbps effective data rate. Peak bandwidth is 28.51 GB/s. These four numbers define the memory interface’s throughput limits.

For high resolutions, the 1024 MB capacity is a hard ceiling. Any scene requiring more than 1024 MB for geometry, textures, or render targets cannot be contained within the frame buffer. The 128-bit bus is the narrow transfer path, while the effective 1782 Mbps data rate sets how many transfers occur per second. Together, they produce the 28.51 GB/s ceiling. The pixel rate of 3.500 GPixel/s and texture rate of 14.00 GTexel/s indicate that the GPU can generate pixel and texture work at certain rates, but those rates depend on memory access. With 28.51 GB/s of bandwidth, sustained high-resolution rendering is likely to be constrained by memory throughput before the shading units reach their 336.0 GFLOPS limit.

The memory interface is also distinct from the PCIe 3.0 x16 host interface. The PCIe link handles data between the card and the CPU, while the 128-bit bus handles data between the GPU and its local DDR3 memory. The card’s 50 W TDP and lack of power connectors suggest a memory configuration designed for low power draw. The 145 mm length and single-slot shape complete a compact, conservative memory and board design. The record’s numbers converge on one conclusion: 1024 MB, a 128-bit bus, and 28.51 GB/s form a modest memory subsystem that will constrain demanding workloads before the compute and fill-rate limits are reached.