AMD Radeon HD 8690M vs NVIDIA GeForce GTX 750 Comparison

Head-to-Head Benchmarks

The database contains a single direct comparison between the NVIDIA GeForce GTX 750 and the AMD Radeon HD 8690M: the Geekbench OpenCL test. In this measurement, the GTX 750 records a score of 9,315, while the HD 8690M posts 6,137. The resulting delta is 51.8% in favor of the NVIDIA part, making it the clear winner in this head-to-head matchup.

That 51.8% margin is substantial. It means the GTX 750 delivers over half again as much compute throughput in OpenCL workloads as the AMD mobile chip. For context, the GTX 750’s average benchmark score across all recorded tests is 7,222, while the HD 8690M’s average is 6,137. The GTX 750 sits at the 40th percentile among all GPUs in the database, whereas the HD 8690M sits at the 36th percentile. The performance gap in the head-to-head test is larger than the gap in their percentile rankings would suggest, indicating that the OpenCL workload specifically favors the GTX 750’s architecture more than other general workloads might.

The GTX 750 also has additional benchmark results beyond OpenCL. It records 4,274 in Geekbench Metal and 8,078 in Geekbench Vulkan. The HD 8690M has no Metal or Vulkan scores in the database, so those tests cannot be compared directly. However, the Vulkan score of 8,078 is noteworthy because it is close to the OpenCL score, suggesting the GTX 750 maintains consistent performance across different compute APIs. The Metal score is lower, but that is typical for a GPU not primarily designed for Apple’s ecosystem.

When placed against its nearest rivals in the database, the GTX 750’s average score of 7,222 places it just 0.3% ahead of the AMD Radeon Vega 8 Mobile (7,203), 0.7% ahead of the NVIDIA GeForce GTX 560 SE (7,171), and 0.7% ahead of the Intel Iris Pro Graphics P580 (7,170). Interestingly, the GTX 970 appears in the nearest rivals list with an average score of 7,157, which is 0.9% lower than the GTX 750. This is a surprising result, as the GTX 970 is generally considered a higher-tier product, but the database’s average scores reflect the specific benchmark mix recorded.

The HD 8690M’s nearest rivals include the Intel Iris Pro Graphics 6200 (6,117, 0.3% lower), the NVIDIA RTX A400 (6,078, 1% lower), and the NVIDIA GeForce MX230 (6,077, 1% lower). On the higher side, the AMD Radeon HD 6950 posts 6,210, which is 1.2% above the HD 8690M. These deltas are all small, indicating that the HD 8690M sits in a tightly packed performance cluster. The GTX 750, by contrast, has a slightly wider spread among its nearest rivals, but its absolute scores are consistently higher.

Architecture Differences

The two GPUs come from fundamentally different design philosophies. The GTX 750 is built on NVIDIA’s Maxwell architecture, using the GM107 chip. The HD 8690M uses AMD’s Graphics Core Next 1.0 (GCN 1.0) architecture, based on the Sun chip. Both are fabricated on a 28 nm process at TSMC, and their transistor densities are nearly identical: 12.6 million transistors per square millimeter for the GTX 750 and 12.3 million for the HD 8690M. However, the total transistor counts diverge sharply. The GTX 750 packs 1,870 million transistors into a die size of 148 mm², while the HD 8690M has 690 million transistors on a 56 mm² die. The GTX 750’s die is roughly 2.6 times larger and carries more than 2.7 times the transistor count.

The compute resources also differ. The GTX 750 has 512 shading units, 32 texture mapping units (TMUs), and 16 raster output units (ROPs). The HD 8690M has 320 shading units, 20 TMUs, and 8 ROPs. In every category, the GTX 750 has more hardware. The pixel rate of the GTX 750 is 17.36 GPixel/s, compared to 7.80 GPixel/s for the HD 8690M. The texture rate is 34.72 GTexel/s versus 19.50 GTexel/s. The FP32 compute throughput is 1,111.0 GFLOPS for the GTX 750 and 624.0 GFLOPS for the HD 8690M. These figures align with the OpenCL benchmark result: the GTX 750 has roughly 78% more FP32 throughput, and the measured performance advantage is 51.8%, indicating some efficiency differences or memory bottlenecks in the AMD part.

Clock speeds tell part of the story. The GTX 750 runs at a base clock of 1020 MHz with a boost of 1085 MHz. The HD 8690M runs at 925 MHz base and 975 MHz boost. The NVIDIA chip has a higher base clock by 95 MHz and a higher boost clock by 110 MHz. But clock speed alone does not explain the performance gap; the larger shading unit count and wider memory bus are more significant factors.

Memory configuration is another major divergence. The GTX 750 has 1024 MB of GDDR5 memory on a 128-bit bus, yielding a bandwidth of 80.19 GB/s. The HD 8690M has 2048 MB of GDDR5 memory on a 64-bit bus, yielding 32.00 GB/s. The GTX 750’s bandwidth is 2.5 times higher. The effective memory clock is 5 Gbps for the GTX 750 and 4 Gbps for the HD 8690M. Interestingly, the HD 8690M has twice the memory capacity, but that does not compensate for the much lower bandwidth in compute workloads.

Feature support also differs. The GTX 750 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4. The HD 8690M supports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. The DirectX feature level is slightly different (11_0 versus 11_1), and the Vulkan version is newer on the NVIDIA side. Both support OpenGL 4.6.

The bus interface differs as well: the GTX 750 uses PCIe 3.0 x16, while the HD 8690M uses PCIe 3.0 x8. This halves the available bandwidth to the host system for the AMD part, which can matter in some workloads that stream data from system memory.

Where Each One Wins

The GTX 750 wins the only direct head-to-head benchmark recorded in the database, and it does so by a wide margin. The OpenCL score of 9,315 versus 6,137 represents a 51.8% advantage. This indicates that for general-purpose compute tasks using OpenCL, the GTX 750 is the stronger choice by a significant amount.

Beyond that single test, the GTX 750 also has Vulkan and Metal scores in the database, while the HD 8690M has none. The Vulkan score of 8,078 suggests the GTX 750 can handle modern graphics APIs well. The Metal score of 4,274 is lower but still present, meaning the GTX 750 has recorded performance in Apple’s compute framework, whereas the HD 8690M has no such data.

The HD 8690M’s strengths are not visible in the benchmark data, as it has no wins in any recorded test. However, the specification table shows it has 2 GB of memory versus the GTX 750’s 1 GB. In scenarios where memory capacity matters more than bandwidth, such as holding larger textures or datasets that do not fit in 1 GB, the HD 8690M could have an advantage. But the database does not contain a benchmark that isolates this factor.

For gaming and graphics workloads, the GTX 750’s higher pixel rate (17.36 GPixel/s versus 7.80 GPixel/s) and texture rate (34.72 GTexel/s versus 19.50 GTexel/s) point toward better rasterization performance. The higher ROP count (16 versus 8) also supports more efficient fill-rate-bound tasks.

For compute-heavy applications like machine learning inference, physics simulations, or video encoding, the GTX 750’s higher FP32 throughput (1,111.0 GFLOPS versus 624.0 GFLOPS) and wider memory bus (128-bit versus 64-bit) are clear advantages. The memory bandwidth alone is 2.5 times higher, which often becomes the limiting factor in such workloads.

Power consumption is another consideration. The GTX 750 has a TDP of 55 W, while the HD 8690M’s TDP is not listed in the database. The GTX 750 also has a suggested PSU rating of 250 W, requires no power connectors, and is a single-slot card. The HD 8690M has no listed slot width, power connector, or PSU requirement, which is typical for a mobile GPU. In a desktop context, the GTX 750 is a low-power card that does not need auxiliary power. In a laptop context, the HD 8690M is designed for mobility, but the database does not provide its power draw.

Specification Differences

The following specifications differ between the two GPUs:

• Chip: GM107 (NVIDIA) versus Sun (AMD)
• Architecture: Maxwell versus GCN 1.0
• Generation: GeForce 700 versus Solar System (HD 8600M)
• Transistors: 1,870 million versus 690 million
• Die Size: 148 mm² versus 56 mm²
• Transistor Density: 12.6M / mm² versus 12.3M / mm²
• Base Clock: 1020 MHz versus 925 MHz
• Boost Clock: 1085 MHz versus 975 MHz
• Memory Clock: 1253 MHz (5 Gbps effective) versus 1000 MHz (4 Gbps effective)
• Memory Size: 1024 MB versus 2048 MB
• Bus Width: 128 bit versus 64 bit
• Bandwidth: 80.19 GB/s versus 32.00 GB/s
• Shading Units: 512 versus 320
• TMUs: 32 versus 20
• ROPs: 16 versus 8
• Pixel Rate: 17.36 GPixel/s versus 7.80 GPixel/s
• Texture Rate: 34.72 GTexel/s versus 19.50 GTexel/s
• FP32: 1,111.0 GFLOPS versus 624.0 GFLOPS
• TDP: 55 W versus not listed
• Slot Width: Single-slot versus not listed
• Power Connectors: None versus not listed
• Suggested PSU: 250 W versus not listed
• Bus Interface: PCIe 3.0 x16 versus PCIe 3.0 x8
• Display Outputs: 2x DVI, 1x mini-HDMI 1.4a versus not listed
• DirectX: 12 (11_0) versus 12 (11_1)
• Vulkan: 1.4 versus 1.2.170
• Dimensions: Length 145 mm (5.7 inches) versus not listed
• Release Date: 2014-02-17 versus 2013-02-28
• Predecessor: GeForce 600 versus London
• Successor: GeForce 900 versus Gem System
• Launch MSRP: 119 USD versus not listed

The GTX 750 was released in February 2014, about a year after the HD 8690M, which came in February 2013. Both are end-of-life products.

FAQ

Q: Which GPU is faster in OpenCL compute?

A: The NVIDIA GeForce GTX 750 scores 9,315 in Geekbench OpenCL, while the AMD Radeon HD 8690M scores 6,137. The GTX 750 is 51.8% faster in this test.

Q: How much memory does each GPU have?

A: The GTX 750 has 1024 MB of GDDR5 memory on a 128-bit bus. The HD 8690M has 2048 MB of GDDR5 memory on a 64-bit bus.

Q: What is the memory bandwidth difference?

A: The GTX 750 provides 80.19 GB/s of bandwidth, while the HD 8690M provides 32.00 GB/s. The NVIDIA part has 2.5 times the memory bandwidth.

Q: Which GPU has more shading units?

A: The GTX 750 has 512 shading units, compared to 320 for the HD 8690M. It also has 32 TMUs versus 20, and 16 ROPs versus 8.

Q: What are the architecture differences?

A: The GTX 750 uses NVIDIA’s Maxwell architecture with the GM107 chip. The HD 8690M uses AMD’s GCN 1.0 architecture with the Sun chip. Both are built on a 28 nm process at TSMC, but the GTX 750 has 1,870 million transistors versus 690 million for the AMD part.

Q: What is the FP32 compute throughput?

A: The GTX 750 delivers 1,111.0 GFLOPS of FP32 performance, while the HD 8690M delivers 624.0 GFLOPS. The GTX 750 has roughly 78% more raw compute throughput.

Q: Does the GTX 750 have better API support?

A: The GTX 750 supports Vulkan 1.4 and DirectX 12 (11_0). The HD 8690M supports Vulkan 1.2.170 and DirectX 12 (11_1). Both support OpenGL 4.6. The NVIDIA part has a newer Vulkan version.