AMD Radeon HD 8790M vs NVIDIA Quadro K3100M Comparison

Where Each One Wins

The benchmark data splits cleanly between these two mobile GPUs, with each taking one discipline. In Geekbench OpenCL, the NVIDIA Quadro K3100M posts 6154 points against the AMD Radeon HD 8790M's 5017, a margin of 18.5 percent. That is a substantial lead in general-purpose compute workloads, which tend to favor the larger Kepler chip's raw throughput. The Quadro K3100M also appears in a Metal benchmark with a score of 3823, though the AMD part has no comparable Metal result in the database, so cross-platform comparison is limited to the two shared tests.

The AMD Radeon HD 8790M counters in Geekbench Vulkan, scoring 6365 against the Quadro's 5484, a 16.1 percent advantage. Vulkan is a lower-level API that rewards efficient command processing and driver overhead management, and the GCN 1.0 architecture appears to handle it better here. This split suggests the AMD part is the stronger choice for Vulkan-based gaming or modern API workloads, while the NVIDIA part holds the edge in OpenCL compute tasks. The average benchmark scores reflect this divide differently: the AMD part averages 5691 across its recorded tests, while the NVIDIA part averages 5154, largely because the Quadro's Metal result drags its average down. The percentile rankings tell a similar story, with the AMD part at the 33rd percentile of all GPUs and the NVIDIA part at the 30th, a narrow gap that masks the very different performance profiles in individual tests.

Architecture Differences

The two chips come from different design philosophies. The AMD Radeon HD 8790M uses the Mars chip built on GCN 1.0 architecture, fabricated by TSMC on a 28 nm process. It packs 950 million transistors into a 77 mm² die, yielding a transistor density of 12.3 million per square millimeter. The NVIDIA Quadro K3100M uses the GK104 chip with Kepler architecture, also on TSMC's 28 nm node, but with 3,540 million transistors across a 294 mm² die, for a density of 12.0 million per square millimeter. The density figures are nearly identical, which makes sense given the shared process node, but the NVIDIA die is roughly 3.8 times larger and carries over 3.7 times more transistors.

That size difference translates directly into resource counts. The Quadro K3100M has 768 shading units, 64 texture mapping units, and 32 ROPs. The Radeon HD 8790M has 384 shading units, 24 TMUs, and just 8 ROPs. The NVIDIA part doubles the shader count, nearly triples the TMU count, and quadruples the ROP count. Clock speeds go the other way: the Radeon runs at 850 MHz base and 900 MHz boost, while the Quadro is locked at 706 MHz for both base and boost. The Radeon's higher clocks partially compensate for its smaller execution resources, but not enough to close the gap in most throughput-bound scenarios.

Memory subsystems also diverge sharply. The Radeon HD 8790M has 2 GB of GDDR5 on a 128-bit bus, with 1000 MHz memory clock translating to 4 Gbps effective and 64.00 GB/s of bandwidth. The Quadro K3100M has 4 GB of GDDR5 on a 256-bit bus, with 800 MHz memory clock (3.2 Gbps effective) and 102.4 GB/s of bandwidth. The Quadro's bus width is double, and its bandwidth is 60 percent higher, which matters for large working sets and texture-heavy workloads. The Radeon counters with faster memory clock speed, but the narrower bus limits total throughput.

API support differs in one notable direction. The Radeon HD 8790M supports DirectX 12 (11_1), while the Quadro K3100M supports DirectX 12 (11_0). Both support OpenGL 4.6, and the NVIDIA part has a slightly newer Vulkan version at 1.2.175 versus 1.2.170. The Radeon's higher DirectX feature level could matter for certain modern titles, though the practical impact depends on the specific workload.

Head-to-Head Benchmarks

The Geekbench OpenCL result is the Quadro K3100M's strongest statement. Its 6154 score beats the Radeon HD 8790M's 5017 by 18.5 percent. Given that the Quadro has double the shading units, 2.67 times the TMUs, and 4 times the ROPs, the OpenCL margin actually looks modest relative to the hardware disparity. The Radeon's higher clocks and GCN architecture likely narrow what could have been a much larger gap. For OpenCL compute tasks like image processing, physics simulation, or data-parallel workloads, the Quadro is clearly the better performer.

The Vulkan result flips the script. The Radeon HD 8790M scores 6365 against the Quadro's 5484, a 16.1 percent advantage. This is remarkable because the Quadro has far more raw execution resources. The Radeon's GCN 1.0 architecture, despite being older in design lineage, apparently handles Vulkan's explicit command model more efficiently. The result implies that driver quality and architecture-level scheduling matter more than raw shader counts in modern API scenarios. The Radeon also achieves this win while consuming less power on paper, though the database does not list a TDP for the AMD part.

Context from the nearest rivals helps position these scores. The Radeon HD 8790M's average of 5691 sits within 0.7 percent of the NVIDIA GeForce GTX 670MX (5721) and GTX 550 Ti (5731), and it is 1.6 percent ahead of the NVIDIA Quadro M500M (5604). The Quadro K3100M's average of 5154 is 1.6 percent behind the AMD Radeon R7 240 (5063 is the rival, so the Quadro is ahead by 1.8 percent, and the database lists that rival with a delta of 1.8), and it trails the NVIDIA Quadro 4000M (5211) by 1.1 percent. These deltas show both parts are closely clustered with their peers, despite the large swing between OpenCL and Vulkan results.

Specification Differences

The two GPUs differ across nearly every major specification category. The Radeon HD 8790M uses the Mars chip with GCN 1.0 architecture, while the Quadro K3100M uses GK104 with Kepler. The Radeon has 950 million transistors on a 77 mm² die; the Quadro has 3,540 million on 294 mm². Both use TSMC's 28 nm process, with densities of 12.3M/mm² and 12.0M/mm² respectively.

Clock speeds favor the Radeon: 850 MHz base and 900 MHz boost versus 706 MHz for both base and boost on the Quadro. Memory clocks also favor the Radeon at 1000 MHz (4 Gbps effective) versus 800 MHz (3.2 Gbps effective). However, the Quadro has 4 GB of memory versus 2 GB, a 256-bit bus versus 128-bit, and 102.4 GB/s bandwidth versus 64.00 GB/s.

Execution resources favor the Quadro overwhelmingly: 768 shading units versus 384, 64 TMUs versus 24, and 32 ROPs versus 8. Pixel rate is 11.30 GPixel/s versus 7.200 GPixel/s, texture rate is 45.18 GTexel/s versus 21.60 GTexel/s, and FP32 throughput is 1,084.4 GFLOPS versus 691.2 GFLOPS. The Quadro has a listed TDP of 75 W; the Radeon has no TDP listed. Both use MXM modules, but the Radeon uses MXM-A (3.0) while the Quadro uses MXM-B (3.0). Both have no power connectors and portable-device-dependent display outputs. The Radeon supports DirectX 12 (11_1); the Quadro supports DirectX 12 (11_0). Both list OpenGL 4.6, with Vulkan 1.2.170 on the Radeon and 1.2.175 on the Quadro. The Radeon released on 2013-03-31, the Quadro on 2013-07-22, both are end-of-life, and neither has a launch MSRP listed.

FAQ

Q: Which GPU wins in Geekbench OpenCL?

A: The NVIDIA Quadro K3100M wins with a score of 6154 versus the AMD Radeon HD 8790M's 5017, a lead of 18.5 percent.

Q: Which GPU wins in Geekbench Vulkan?

A: The AMD Radeon HD 8790M wins with 6365 points against the Quadro K3100M's 5484, an advantage of 16.1 percent.

Q: How do the memory configurations compare?

A: The Quadro K3100M has 4 GB of GDDR5 on a 256-bit bus with 102.4 GB/s bandwidth; the Radeon HD 8790M has 2 GB on a 128-bit bus with 64.00 GB/s bandwidth.

Q: What is the transistor count difference?

A: The Quadro K3100M has 3,540 million transistors, while the Radeon HD 8790M has 950 million, a gap of roughly 3.7 times.

Q: Which GPU has higher clock speeds?

A: The Radeon HD 8790M runs at 850 MHz base and 900 MHz boost, while the Quadro K3100M is fixed at 706 MHz for both.

Q: Do both GPUs support the same DirectX version?

A: No. The Radeon HD 8790M supports DirectX 12 (11_1), while the Quadro K3100M supports DirectX 12 (11_0).

The Verdict

The data points to a workload-dependent choice rather than a clear overall winner. The NVIDIA Quadro K3100M is the pick for OpenCL compute workloads, where its 18.5 percent lead in Geekbench OpenCL reflects the advantage of 768 shading units, 64 TMUs, 32 ROPs, and 102.4 GB/s of memory bandwidth. The Quadro also offers double the VRAM at 4 GB, which matters for large datasets or multi-application workflows. Its 75 W TDP is documented, making it a known quantity for thermal planning in mobile workstations.

The AMD Radeon HD 8790M is the pick for Vulkan-based workloads, where its 16.1 percent lead in Geekbench Vulkan demonstrates that architecture efficiency can overcome hardware resource disadvantages. The Radeon also has a higher DirectX feature level (11_1 versus 11_0), which could provide better compatibility with certain modern titles. Its smaller 2 GB framebuffer and 128-bit bus are clear limitations for memory-intensive tasks, and its 64.00 GB/s bandwidth is barely over half of the Quadro's.

For buyers who prioritize OpenCL compute, the Quadro K3100M is the data-supported choice. For those who prioritize Vulkan performance or DirectX feature level, the Radeon HD 8790M holds the edge. The average benchmark scores show the Radeon ahead at 5691 versus 5154, but that average includes the Quadro's Metal score, which has no AMD counterpart, so it should not be treated as a direct comparison. The percentile rankings, 33rd for AMD and 30th for NVIDIA, suggest both parts occupy similar tiers in the overall GPU landscape. In short, the Quadro wins where raw throughput matters, and the Radeon wins where modern API efficiency matters.