AMD Radeon R9 M265X vs NVIDIA GeForce GTX 660 Comparison

Head-to-Head Benchmarks

The only shared benchmark between these two cards is Geekbench OpenCL, and the result is decisive. The NVIDIA GeForce GTX 660 scores 11,347, while the AMD Radeon R9 M265X scores 8,851. That is a 28.2% advantage for the GTX 660, which is a substantial gap for two GPUs sitting at nearly the same overall percentile ranking. In practical terms, the GTX 660 is completing the same OpenCL workload roughly a quarter faster than the R9 M265X.

Looking at the broader benchmark picture, the GTX 660 has three recorded scores: 4,305 in Geekbench Metal, 11,347 in OpenCL, and 11,415 in Vulkan. Its average benchmark score across all tests is 9,022. The R9 M265X has only one recorded score, the 8,851 OpenCL result, which also serves as its average. This means the R9 M265X has no data for Metal or Vulkan, so any comparison there is impossible. When you average the GTX 660's three scores, it lands at 9,022, which is only 1.9% higher than the R9 M265X's single result. That sounds close, but the average is dragged down by the Metal score. In the one test both cards actually run, the GTX 660 is clearly ahead.

The percentile data tells a similar story with a twist. The GTX 660 sits at the 45th percentile of all GPUs, while the R9 M265X sits at the 44th percentile. That is nearly identical positioning in the overall hierarchy. However, the nearest rivals for each card reveal how the field around them is packed. The GTX 660's closest competitor is the NVIDIA TITAN V CEO Edition, which scores 9,037 and is only 0.2% ahead. The GTX 660 also trails the GeForce GTX 560 (9,058, 0.4% ahead) but beats the AMD Radeon 550X (8,918, 1.2% behind) and the AMD Radeon Pro WX 5100 (8,863, 1.8% behind). The R9 M265X, meanwhile, is 0.1% behind the Radeon Pro WX 5100, 0.8% behind the Radeon 550X, and 1.2% ahead of both the RTX 3050 A Mobile and the GTX 460 v2. So while the GTX 660 wins the head-to-head, both cards are swimming in a very dense pool of similarly performing GPUs. A 28.2% win in OpenCL does not translate to a massive percentile jump because the R9 M265X is already close to the GTX 660's neighborhood in the aggregate.

The wins column is straightforward: the GTX 660 has one win, the R9 M265X has zero. That is the entire head-to-head record. There is no benchmark where the AMD card comes out on top. The delta in OpenCL is the only quantitative comparison available, and it favors NVIDIA by a wide margin.

Architecture Differences

The GTX 660 is built on NVIDIA's Kepler architecture using the GK106 chip. It is a 28 nm part fabricated at TSMC, with 2,540 million transistors on a 221 mm² die. That works out to a transistor density of 11.5 million transistors per square millimeter. The R9 M265X uses AMD's GCN 1.0 architecture with the Venus chip. It is also 28 nm from TSMC, but the die is much smaller at 123 mm² and holds 1,500 million transistors. The transistor density is slightly higher at 12.2 million per square millimeter, which is interesting because the AMD chip packs more transistors per area despite having fewer total transistors.

Core counts differ significantly. The GTX 660 has 960 shading units, 80 texture mapping units, and 24 ROPs. The R9 M265X has 640 shading units, 40 TMUs, and 16 ROPs. That means the GTX 660 has 50% more shading units, double the TMUs, and 50% more ROPs. This hardware advantage directly explains the OpenCL result. The GTX 660 also runs at much higher clocks: 980 MHz base and 1,032 MHz boost for the NVIDIA card, versus 575 MHz base and 625 MHz boost for the AMD card. Even the memory clocks are far apart. The GTX 660 runs its GDDR5 at 1,502 MHz with 6 Gbps effective, while the R9 M265X runs at 1,000 MHz with 4 Gbps effective.

Memory configuration is another major split. Both cards have 2 GB of GDDR5, but the GTX 660 uses a 192-bit bus delivering 144.2 GB/s of bandwidth. The R9 M265X is limited to a 128-bit bus and 64.00 GB/s. That is more than a 2x advantage in memory bandwidth for the GTX 660. The R9 M265X cannot feed its cores as quickly, which hurts in memory-intensive workloads.

The resulting compute rates tell the story numerically. The GTX 660 achieves 20.64 GPixel/s pixel fill rate, 82.56 GTexel/s texture rate, and 1.981 TFLOPS FP32 compute. The R9 M265X manages 10.00 GPixel/s, 25.00 GTexel/s, and 800.0 GFLOPS FP32. Every rate metric is more than double on the NVIDIA side. Pixel rate is 2.06x higher, texture rate is 3.3x higher, and FP32 is 2.48x higher. The R9 M265X's lower clocks and fewer cores create a wide gap in raw throughput.

The API support differs slightly. Both support DirectX 12, but the GTX 660 is at 12 (11_0) while the R9 M265X is at 12 (11_1). OpenGL is identical at 4.6 for both. Vulkan is close, with the GTX 660 at version 1.2.175 and the R9 M265X at 1.2.170. Neither card has ray tracing or tensor cores. The GTX 660 supports PCIe 3.0 x16, and so does the R9 M265X. The R9 M265X has no listed display outputs, while the GTX 660 offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.

The GTX 660 is a desktop card with a 140 W TDP, dual-slot width, a 1x 6-pin power connector, and a 300 W suggested PSU. It is 241 mm or 9.5 inches long. The R9 M265X has no TDP, slot width, power connector, or dimension data listed. It is a mobile part from the "Gem System (R9 M200)" generation, which explains the missing power specs. The GTX 660 is from the GeForce 600 generation and was released on 2012-09-05. The R9 M265X came later on 2014-03-20. Both are end-of-life products now.

The Verdict

The data is unambiguous: the NVIDIA GeForce GTX 660 is the faster card. In the only shared benchmark, it beats the R9 M265X by 28.2% in OpenCL. Its shading unit count is 50% higher, its TMU count is double, its ROP count is 50% higher, and its memory bandwidth is more than double. Every compute rate — pixel, texture, and FP32 — is at least twice as high on the GTX 660. If you are choosing between these two for a workload that uses OpenCL, the GTX 660 is the obvious pick.

However, there is a context issue. The R9 M265X is a mobile GPU from the R9 M200 series, while the GTX 660 is a desktop card. They are not direct competitors in the market. The R9 M265X has no TDP or power connector data because it is designed for laptops. The GTX 660 requires a 300 W PSU and a 6-pin connector. If you are building a desktop, the GTX 660 wins easily. If you are looking at a laptop, the R9 M265X is the only option of the two.

The percentile rankings are nearly identical at 45 vs 44, which suggests that in aggregate performance, these cards are close. But the aggregate for the GTX 660 includes a weak Metal score of 4,305 that drags its average down. The R9 M265X has no Metal score to compare. The single OpenCL data point is the cleanest comparison, and it strongly favors NVIDIA. The GTX 660's nearest rival, the TITAN V CEO Edition, is only 0.2% ahead, meaning the GTX 660 is competitive with much more expensive hardware in this metric. The R9 M265X's nearest rivals are all within 1.2% in either direction, so it sits in a tightly packed mid-range cluster.

Who should pick which? If you need the higher OpenCL score, the GTX 660. If you need a low-power mobile part with no external power requirements, the R9 M265X. There is no benchmark where the R9 M265X wins, so the choice hinges entirely on form factor and power constraints, not performance.

FAQ

Q: Which GPU is faster in the shared OpenCL benchmark?

A: The NVIDIA GeForce GTX 660 scores 11,347 in Geekbench OpenCL, while the AMD Radeon R9 M265X scores 8,851. The GTX 660 wins by 28.2%.

Q: How do the two cards compare in memory bandwidth?

A: The GTX 660 has a 192-bit bus with 144.2 GB/s bandwidth, while the R9 M265X has a 128-bit bus with 64.00 GB/s. The GTX 660 delivers more than double the bandwidth.

Q: Do both cards support the same DirectX version?

A: Both support DirectX 12, but the GTX 660 is at 12 (11_0) and the R9 M265X is at 12 (11_1). The AMD card has a slightly higher feature level.

Q: What is the FP32 compute performance difference?

A: The GTX 660 achieves 1.981 TFLOPS FP32, while the R9 M265X achieves 800.0 GFLOPS FP32. The GTX 660 is approximately 2.48x higher.

Q: Are these cards close in overall performance ranking?

A: Yes. The GTX 660 is at the 45th percentile of all GPUs with an average benchmark score of 9,022, while the R9 M265X is at the 44th percentile with an average of 8,851.

Q: Which card has more shading units?

A: The GTX 660 has 960 shading units, while the R9 M265X has 640. That is a 50% advantage for the NVIDIA card.

Where Each One Wins

The GTX 660 wins every measurable performance category. It has the higher OpenCL score, the higher pixel rate, the higher texture rate, the higher FP32 throughput, the higher memory bandwidth, and the higher core counts. It also has a much higher boost clock at 1,032 MHz versus 625 MHz. If the workload is compute-heavy, texture-heavy, or memory-heavy, the GTX 660 is the card to use.

The R9 M265X wins in power efficiency and physical footprint. It has no listed TDP, which means it is designed to operate within mobile thermal limits. It has no power connector requirement and no slot width listed, so it fits in laptop chassis. The GTX 660 needs a 140 W TDP budget, a dual-slot cooler, a 6-pin power connector, and a 300 W PSU. The R9 M265X also uses a smaller die at 123 mm² versus 221 mm², so it is cheaper to produce and generates less heat.

The R9 M265X also has a slight edge in transistor density at 12.2M per mm² versus 11.5M for the GTX 660, though this is irrelevant to performance. Both cards are 28 nm parts from TSMC. The R9 M265X's advantage is purely in packaging efficiency, not in raw output.

There is no workload where the R9 M265X beats the GTX 660 in the provided data. The only way the AMD card "wins" is by being the only option in a mobile form factor. If you are constrained to a laptop, the R9 M265X is what you get. If you have desktop power delivery, the GTX 660 is superior in every numerical comparison available.

Specification Differences

Chip and Architecture: The GTX 660 uses NVIDIA's GK106 chip with Kepler architecture. The R9 M265X uses AMD's Venus chip with GCN 1.0 architecture.

Transistor Count and Die Size: The GTX 660 has 2,540 million transistors on a 221 mm² die. The R9 M265X has 1,500 million transistors on a 123 mm² die. Transistor density is 11.5M per mm² for the GTX 660 and 12.2M per mm² for the R9 M265X.

Clock Speeds: The GTX 660 runs at 980 MHz base and 1,032 MHz boost. The R9 M265X runs at 575 MHz base and 625 MHz boost. Memory clocks are 1,502 MHz (6 Gbps effective) for the GTX 660 and 1,000 MHz (4 Gbps effective) for the R9 M265X.

Memory Bus and Bandwidth: The GTX 660 has a 192-bit bus with 144.2 GB/s bandwidth. The R9 M265X has a 128-bit bus with 64.00 GB/s bandwidth. Both have 2 GB of GDDR5.

Core Configuration: The GTX 660 has 960 shading units, 80 TMUs, and 24 ROPs. The R9 M265X has 640 shading units, 40 TMUs, and 16 ROPs.

Compute Rates: The GTX 660 delivers 20.64 GPixel/s, 82.56 GTexel/s, and 1.981 TFLOPS FP32. The R9 M265X delivers 10.00 GPixel/s, 25.00 GTexel/s, and 800.0 GFLOPS FP32.

API Support: DirectX is 12 (11_0) for the GTX 660 and 12 (11_1) for the R9 M265X. OpenGL is 4.6 for both. Vulkan is 1.2.175 for the GTX 660 and 1.2.170 for the R9 M265X.

Power and Physical: The GTX 660 has a 140 W TDP, dual-slot width, 1x 6-pin power connector, 300 W suggested PSU, and is 241 mm (9.5 inches) long. The R9 M265X has no listed TDP, slot width, power connector, suggested PSU, or dimensions.

Other Differences: The GTX 660 has display outputs (2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2); the R9 M265X has none listed. The GTX 660 was released on 2012-09-05; the R9 M265X on 2014-03-20. The GTX 660 has a launch MSRP of 229 USD; the R9 M265X has none listed.