NVIDIA GeForce GTX 560 Ti OEM

The NVIDIA GeForce GTX 560 Ti OEM is a GeForce 500-series graphics card built by NVIDIA around the GF110 chip and the Fermi 2.0 architecture. Manufactured at TSMC on a 40 nm process, the die integrates 3,000 million transistors across 520 mm², giving a transistor density of 5.8M / mm². The card is dated 2011-03-29 and carries an end-of-life production status. Its position in the database is the 50th percentile among all GPUs, yet the benchmarks array is empty and the average benchmark score is 0. The name includes an OEM suffix, although the fact pack contains no additional OEM-specific details such as altered clocks or custom board dimensions. Accordingly, all conclusions below are drawn from the specification fields and the percentile value, not from workload measurements.

Power and Cooling

The card's thermal and power design is anchored by a 210 W TDP, a suggested PSU figure of 550 W, and two 6-pin power connectors. The 550 W value is a system-level PSU recommendation, distinct from the 210 W card-level thermal design point. The board is dual-slot with a length of 229 mm, or 9 inches. The large GF110 die — 520 mm² at 40 nm, containing 3,000 million transistors — is the physical source of the 210 W envelope, and the 40 nm process with a density of 5.8M / mm² frames the cooling challenge. A 550 W power supply is the listed recommendation, and the two 6-pin connectors supply auxiliary current; the bus interface is PCIe 2.0 x16. The dual-slot cooler occupies an adjacent expansion slot, and the 229 mm length must be accommodated by the chassis. No core clock is listed, so frequency-dependent thermal analysis cannot be performed. The only clock present is the memory clock of 950 MHz, stated as 3.8 Gbps effective. The fact pack does not describe the cooler's height or width, so clearance beyond length is unspecified. In practice, a 210 W board with two 6-pin inputs requires a PSU with those leads and a case with a dual-slot rear opening.

Memory Subsystem

The memory subsystem is defined by 1280 MB of GDDR5 on a 256-bit bus, with a bandwidth of 121.6 GB/s. The memory clock is 950 MHz, which the fact pack converts to 3.8 Gbps effective; the 256-bit bus and the effective data rate together account for the stated bandwidth. For high resolutions, capacity is the first limit: 1280 MB is a fixed ceiling for framebuffer and texture data. The chip's pixel rate of 16.10 GPixel/s and texture rate of 32.21 GTexel/s can consume memory bandwidth quickly, but the 121.6 GB/s figure is sufficient for those fill rates. The more immediate constraint is the capacity ceiling, because framebuffers, depth buffers, and texture mip chains all compete for the same 1280 MB. A 256-bit bus is generally matched with GDDR5 at this clock speed to reach the listed bandwidth; the fact pack lists no memory overclock or secondary profile. The memory type is GDDR5, and the effective data rate of 3.8 Gbps is the only rate supplied. The absence of a core clock prevents a memory-to-core ratio analysis. Overall, the memory subsystem is balanced for 1280 MB-oriented workloads and will become binding at resolution or texture settings that demand more than that capacity.

Who Should Consider It

The fact pack does not include measured frame rates, so recommendations rest on the listed specifications and the 50th percentile rank. Because the benchmarks array is empty, there is no verified performance tier beyond the median position. The 1280 MB memory capacity points to scenarios where texture and framebuffer demand stays below that threshold. The API field lists DirectX 12 (11_0) and OpenGL 4.6, but Vulkan is null, which excludes engines that rely entirely on Vulkan. Display output consists of 2x DVI, 1x HDMI 1.3a, and 1x DisplayPort, enabling multi-monitor connection within a digital output set. The release date of 2011-03-29 and the end-of-life status make it a legacy part; it is most plausible in used or OEM systems. The GeForce 500 generation and Fermi 2.0 architecture define the feature scope, and the predecessor/successor fields — GeForce 400 and GeForce 600 — frame it as a middle-generation product. Users with a 550 W PSU and two 6-pin leads can satisfy the power requirement, and the dual-slot 229 mm board fits standard ATX-class openings. For high-resolution gaming, the 1280 MB buffer is the strongest constraint; for older DirectX 11-era workloads, the 352 shading units, 44 TMUs, and 40 ROPs are the resources listed in the fact pack. The empty benchmarks array means any claim of specific resolution or settings performance would exceed the data.

How It Compares

The nearestRivals array in the fact pack is empty, meaning there are no rival names, scores, or deltaPct values to include in this section. Without those entries, no percentage delta can be stated, and a per-rival paragraph is not possible. The only relative metric is percentileVsAllGpus, which is 50. That value places the GTX 560 Ti OEM at the median of the database's GPU list. The predecessor field lists GeForce 400 and the successor field lists GeForce 600, but those are generation labels, not rival benchmark entries. The fact pack does not contain scores for either adjacent generation. Therefore, any comparison to specific GeForce 400 or GeForce 600 cards would be ungrounded. The empty rivals array also means the database has not classified performance deltas against competing products, so all comparative sentences would be fabrication. The card's own generation, GeForce 500, has no other entries in the fact pack. Given this absence, the section's content is limited to the observation that no rival data exists and the median percentile is the sole quantitative position marker.

Benchmark Performance

The benchmarks array is empty and avgBenchmarkScore is 0, so there are no measured scores to analyze. The theoretical throughput is specified instead: FP32 performance is 1,030.7 GFLOPS, pixel fill is 16.10 GPixel/s, and texture fill is 32.21 GTexel/s. These rates come from 352 shading units, 44 texture mapping units, and 40 render output units. The memory bandwidth of 121.6 GB/s is the data path that feeds these units. Because nearestRivals is empty, the theoretical rates cannot be placed against competing cards. The 50th percentile rank is the only aggregate comparison, and avgBenchmarkScore being 0 means no positive average exists to corroborate it. The memory clock of 950 MHz / 3.8 Gbps effective is the only clock in the fact pack, so performance-per-clock analysis is impossible. The DirectX 12 (11_0) and OpenGL 4.6 fields define the API surface, while Vulkan is null. The hardware is Fermi 2.0 on GF110 with a 40 nm TSMC process, and it belongs to the GeForce 500 generation. The combination of 1,030.7 GFLOPS FP32 and 121.6 GB/s memory bandwidth gives a compute-to-bandwidth ratio that would support bandwidth-sensitive workloads, but without workloads or rival scores no benchmark conclusion is possible. The 50th percentile rank is the sole benchmark result in the fact pack, and it is not accompanied by a margin of error or a workload description. The 40 ROPs realize the 16.10 GPixel/s pixel throughput, while the 44 TMUs realize the 32.21 GTexel/s texture rate; these are hardware-configured relationships, not measured outcomes. In the absence of benchmarks, these theoretical rates and the median rank are the only quantitative performance statements that can be made.