AMD Radeon Vega 11 vs NVIDIA GeForce GTX 680 Comparison

AMD Radeon Vega 11 and NVIDIA GeForce GTX 680 represent two very different approaches to GPU design: one is a 15 W integrated processor from 2019, the other a 195 W dedicated card from 2012. Benchmark aggregation places them nearly level, with the AMD part averaging 14352 points against NVIDIA’s 14150, a gap of only 1.4% in AMD’s favor. The two GPUs trade blows decisively depending on the API, with the GTX 680 dominating Vulkan and OpenCL while the Vega 11 crushes Metal, making the "better" choice entirely dependent on the workload.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon Vega 11 edges out the NVIDIA GeForce GTX 680, with an average benchmark score of 14352 versus 14150. That puts the GTX 680 at a 1.4% deficit relative to the Vega 11 in the aggregated data.

Q: How does the Vega 11 perform in Metal benchmarks compared to the GTX 680?

A: The Vega 11 is dramatically faster in Geekbench Metal, scoring 17392 against the GTX 680’s 7897. This represents a 120.2% advantage for the AMD integrated GPU, by far the largest margin in any test.

Q: Where does the GTX 680 beat the Vega 11?

A: The GTX 680 wins in both OpenCL and Vulkan. In Geekbench OpenCL, it scores 16906 versus 13392 (a 20.8% lead for NVIDIA), and in Geekbench Vulkan it scores 17646 versus 12273 (a 30.4% lead for NVIDIA).

Q: What are the performance percentiles of these two GPUs relative to all GPUs?

A: The Vega 11 sits at the 56th percentile of all GPUs, while the GTX 680 sits at the 55th percentile. This confirms that despite their architectural differences, both are positioned in a similar mid-range performance tier.

Q: Which GPU has more shading units and texture mapping units?

A: The GTX 680 has 1536 shading units and 128 TMUs, while the Vega 11 has 704 shading units and 44 TMUs. The NVIDIA card also has 32 ROPs versus the AMD part’s 8 ROPs.

Q: What are the launch dates for these products?

A: The NVIDIA GeForce GTX 680 was released on 2012-03-21, while the AMD Radeon Vega 11 was released on 2019-09-29. Both are now end-of-life products.

The Verdict

The data paints a clear picture of two GPUs that excel in different environments. For users running Metal-based applications — primarily macOS ecosystems — the AMD Radeon Vega 11 is the unequivocal choice, offering a 120.2% performance advantage over the GTX 680 in that specific API. Its 1.971 TFLOPS of FP32 compute and 3.942 TFLOPS of FP16 (2:1 ratio) suggest a modern architecture tuned for compute-heavy tasks, despite its meager 15 W power envelope.

The NVIDIA GeForce GTX 680, however, is the pick for Vulkan and OpenCL workloads. Its 3.250 TFLOPS of FP32 performance, backed by 1536 shading units and 128 TMUs, translates into a 30.4% lead in Vulkan and a 20.8% lead in OpenCL. The GTX 680 also brings dedicated 2 GB GDDR5 memory on a 256-bit bus with 192.3 GB/s bandwidth, versus the Vega 11’s system-shared memory, which could matter in texture-heavy scenes.

For a system builder constrained to an integrated GPU, the Vega 11 is the only option of the two that fits an IGP form factor. For anyone with a PCIe slot and a power supply capable of handling a 195 W card with dual 6-pin connectors, the GTX 680 offers superior raw throughput in modern cross-platform APIs. The GTX 680’s 450 W suggested PSU requirement is a notable system-level consideration, but its benchmark wins in two of three tests make it the stronger all-rounder for general-purpose GPU compute.

Head-to-Head Benchmarks

The single most lopsided result in this comparison comes from Geekbench Metal, where the AMD Radeon Vega 11 posts 17392 points against the GTX 680’s 7897. That 120.2% delta is staggering, more than doubling the NVIDIA card’s score. The Vega 11’s GCN 5.0 architecture appears to have a far more efficient Metal implementation, likely benefiting from its 2019 release date and modern driver maturity.

The Geekbench OpenCL test flips the script. Here, the GTX 680 scores 16906, which is 20.8% higher than the Vega 11’s 13392. The NVIDIA card’s 3.250 TFLOPS FP32 throughput and 135.4 GTexel/s texture rate give it a decisive edge in this legacy compute API. The AMD part’s 61.60 GTexel/s texture rate and 1.971 TFLOPS FP32 simply cannot match that raw throughput.

Geekbench Vulkan shows the widest margin for NVIDIA, with the GTX 680 scoring 17646 versus 12273 for the Vega 11. That 30.4% gap suggests the Kepler architecture’s 1536 shading units scale better with Vulkan’s low-level draw call overhead. The GTX 680’s 33.86 GPixel/s pixel rate also exceeds the Vega 11’s 11.20 GPixel/s, which may contribute to its Vulkan superiority.

The aggregate benchmark score — 14352 for Vega 11 versus 14150 for GTX 680 — is remarkably close, differing by just 1.4%. This near-tie masks the extreme variance across APIs. The GTX 680 wins two tests, but the Vega 11’s single Metal victory is so overwhelming that it nearly cancels out both NVIDIA wins in the average.

Specification Differences

The most fundamental difference is the form factor and power envelope. The Vega 11 is an IGP with a 15 W TDP and no power connectors, while the GTX 680 is a dual-slot card with 2x 6-pin power connectors and a 195 W TDP. The GTX 680’s suggested PSU is 450 W; the Vega 11 has no such requirement.

Memory configuration diverges sharply. The GTX 680 has 2 GB of GDDR5 on a 256-bit bus with 192.3 GB/s bandwidth, while the Vega 11 uses system-shared memory with system-dependent bandwidth. The NVIDIA card’s dedicated VRAM is a clear advantage for local data access.

Clock speeds differ in magnitude and behavior. The Vega 11 has a 300 MHz base clock and 1400 MHz boost clock, while the GTX 680 runs at 1006 MHz base and 1058 MHz boost. The GTX 680’s memory clock is 1502 MHz (6 Gbps effective), whereas the Vega 11’s memory clock is listed as system shared.

The bus interface is another key differentiator: the Vega 11 uses IGP, while the GTX 680 uses PCIe 3.0 x16. Display outputs on the Vega 11 are motherboard dependent, while the GTX 680 offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.

Dimensions favor the IGP. The GTX 680 measures 256 mm in length, 111 mm in height, and 38 mm in width, while the Vega 11 has no listed physical dimensions due to its integrated nature.

Architecture Differences

The Vega 11 is built on GlobalFoundries’ 12 nm process with 4,940 million transistors on a 210 mm² die, yielding a transistor density of 23.5M per mm². The GTX 680 uses TSMC’s 28 nm node with 3,540 million transistors on a 294 mm² die, giving it a density of 12.0M per mm². The Vega 11’s newer process allows for more than double the transistor density.

Architecturally, the Vega 11 uses GCN 5.0 (chip: Picasso), while the GTX 680 uses the Kepler architecture (chip: GK104). The Vega 11 belongs to the "Vega IGP (Picasso)" generation, and the GTX 680 to "GeForce 600". The Vega 11’s predecessor is GCN 3.0 IGP and its successor is Vega II IGP; the GTX 680’s predecessor is GeForce 500 and its successor is GeForce 700.

Compute resources heavily favor NVIDIA. The GTX 680 has 1536 shading units, 128 TMUs, and 32 ROPs, versus the Vega 11’s 704 shading units, 44 TMUs, and 8 ROPs. Pixel rate and texture rate follow suit: the GTX 680 achieves 33.86 GPixel/s and 135.4 GTexel/s, while the Vega 11 manages 11.20 GPixel/s and 61.60 GTexel/s.

API support shows a mixed picture. Both support DirectX 12 and OpenGL 4.6, but the Vega 11 supports DirectX 12 (12_1) while the GTX 680 only supports DirectX 12 (11_0). The Vega 11 supports Vulkan 1.3, whereas the GTX 680 supports Vulkan 1.2.175. The Vega 11 also lists FP16 at 3.942 TFLOPS (2:1), while the GTX 680 has no FP16 data. The GTX 680’s launch MSRP was 499 USD, though that figure is historical context given both products are end-of-life.