AMD Radeon 660M vs NVIDIA GeForce GTX 590 Comparison

# NVIDIA GeForce GTX 590 vs AMD Radeon 660M

The NVIDIA GeForce GTX 590 and AMD Radeon 660M occupy opposite ends of the GPU spectrum: one is a 2011 dual-GPU flagship designed for maximum desktop performance, while the other is a 2022 integrated graphics processor built for mobile efficiency. Despite their different origins, the benchmark data reveals a surprisingly close contest in raw compute performance, with the AMD part edging ahead in the only shared test. The GTX 590 holds a 52nd percentile ranking among all GPUs, while the Radeon 660M sits slightly higher at the 55th percentile, indicating that both parts land in the mid-range of modern GPU performance despite their age or integrated nature.

Where Each One Wins

The AMD Radeon 660M wins the only direct head-to-head benchmark available: Geekbench OpenCL. It scores 12,876 points versus the GTX 590's 12,830, a margin of just 0.4%. This is a narrow victory, but it is consistent with the Radeon 660M's higher average benchmark score of 13,812 across two tests, compared to the GTX 590's single-test average of 12,830. The Radeon 660M also demonstrates strength in Vulkan workloads, scoring 14,748 in Geekbench Vulkan—a test the GTX 590 does not appear in at all. This suggests the AMD part has broader API support and better modern-API performance.

The GTX 590, however, wins in raw memory bandwidth and pixel throughput. Its dedicated 1536 MB GDDR5 memory on a 384-bit bus delivers 164.0 GB/s, dwarfing the Radeon 660M's system-shared memory with system-dependent bandwidth. The GTX 590 also has a higher pixel rate at 19.46 GPixel/s, although the Radeon 660M counters with a higher texture rate at 45.60 GTexel/s. The GTX 590's 48 ROPs versus the Radeon's 16 ROPs gives it an advantage in fill-rate-bound scenarios, while the Radeon's 24 TMUs are fewer than the GTX 590's 64 TMUs but run at higher clocks.

The Radeon 660M wins decisively in power efficiency and form factor. Its 40 W TDP is a fraction of the GTX 590's 365 W, and it requires no power connectors or dedicated PSU recommendation, whereas the GTX 590 needs two 8-pin connectors and a 750 W PSU. The Radeon is an IGP with no dimensions, while the GTX 590 is a dual-slot, 279 mm card.

Architecture Differences

The two GPUs come from entirely different architectural eras. The GTX 590 uses NVIDIA's GF110 chip based on Fermi 2.0 architecture, built on a 40 nm process at TSMC. It packs 3,000 million transistors on a 520 mm² die, yielding a transistor density of 5.8M per mm². The Radeon 660M uses AMD's Rembrandt chip based on RDNA 2.0, built on a 6 nm process, also at TSMC. The Rembrandt die contains 13,100 million transistors across 208 mm², achieving a dramatically higher density of 63.0M per mm²—over ten times the transistor density of the Fermi part.

The GTX 590 has 512 shading units, 64 TMUs, and 48 ROPs. The Radeon 660M has 384 shading units, 24 TMUs, and 16 ROPs, but it adds 6 ray-tracing cores, which the GTX 590 lacks entirely. The Radeon also supports FP16 compute at 2.918 TFLOPS (2:1), while the GTX 590 has no FP16 capability listed. The GTX 590's FP32 performance is 1,244.2 GFLOPS, while the Radeon 660M reaches 1,459.2 GFLOPS, making the AMD part about 17% faster in raw FP32 throughput.

Memory architecture differs fundamentally. The GTX 590 uses dedicated 1536 MB GDDR5 on a 384-bit bus with fixed 164.0 GB/s bandwidth. The Radeon 660M uses system-shared memory, meaning its capacity, bus width, and bandwidth are all system-dependent. The GTX 590's memory clock is 854 MHz (3.4 Gbps effective), while the Radeon's memory clock is listed as "System Shared."

The GTX 590 supports DirectX 12 (11_0), OpenGL 4.6, and has no Vulkan support listed. The Radeon 660M supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The GTX 590 uses PCIe 2.0 x16, while the Radeon uses PCIe 4.0 x8. The GTX 590 has fixed display outputs (3x DVI, 1x mini-DisplayPort), while the Radeon's outputs are portable-device-dependent.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon 660M has an average benchmark score of 13,812 across two tests (OpenCL and Vulkan), while the NVIDIA GeForce GTX 590 has an average of 12,830 from its single OpenCL test. The Radeon's percentile ranking of 55 also exceeds the GTX 590's 52.

Q: How close is the OpenCL performance between the two?

A: The AMD Radeon 660M scores 12,876 in Geekbench OpenCL versus 12,830 for the GTX 590, a delta of only 0.4%. This is within the margin of error when compared to their nearest rivals—the GTX 590's closest rival, AMD Radeon Pro 455, scores 12,831 (0% delta), while the Radeon 660M's closest rival, NVIDIA RTX A2000 Mobile, scores 13,821 (-0.1% delta).

Q: Does the GTX 590 support ray tracing?

A: No. The GTX 590 has no ray-tracing cores listed in its specifications, while the AMD Radeon 660M includes 6 ray-tracing cores. The Radeon also supports DirectX 12 Ultimate (12_2), which includes ray-tracing features, whereas the GTX 590 is limited to DirectX 12 (11_0).

Q: How do their power requirements compare?

A: The GTX 590 has a TDP of 365 W and requires two 8-pin power connectors plus a 750 W suggested PSU. The AMD Radeon 660M has a TDP of just 40 W, requires no power connectors, and has no suggested PSU rating since it is an integrated processor.

Q: Which GPU is more transistor-dense?

A: The AMD Radeon 660M achieves 63.0M transistors per mm² on a 6 nm process, while the NVIDIA GTX 590 has 5.8M per mm² on a 40 nm process. The Radeon's Rembrandt die contains 13,100 million transistors versus the GTX 590's 3,000 million.

Q: What memory does each GPU use?

A: The GTX 590 has 1536 MB of dedicated GDDR5 memory on a 384-bit bus with 164.0 GB/s bandwidth. The Radeon 660M uses system-shared memory with system-dependent bandwidth and bus width, meaning its memory performance depends entirely on the host system's configuration.

Specification Differences

• Process Node: GTX 590 is 40 nm; Radeon 660M is 6 nm.
• Transistors: GTX 590 has 3,000 million; Radeon 660M has 13,100 million.
• Die Size: GTX 590 is 520 mm²; Radeon 660M is 208 mm².
• Transistor Density: GTX 590 is 5.8M / mm²; Radeon 660M is 63.0M / mm².
• Base Clock: GTX 590 has no listed base clock; Radeon 660M has 1500 MHz.
• Boost Clock: GTX 590 has no listed boost clock; Radeon 660M has 1900 MHz.
• Memory Size: GTX 590 is 1536 MB; Radeon 660M is System Shared.
• Memory Type: GTX 590 is GDDR5; Radeon 660M is System Shared.
• Memory Bus: GTX 590 is 384 bit; Radeon 660M is System Shared.
• Memory Bandwidth: GTX 590 is 164.0 GB/s; Radeon 660M is System Dependent.
• Shading Units: GTX 590 has 512; Radeon 660M has 384.
• TMUs: GTX 590 has 64; Radeon 660M has 24.
• ROPs: GTX 590 has 48; Radeon 660M has 16.
• Ray Tracing Cores: GTX 590 has 0; Radeon 660M has 6.
• Pixel Rate: GTX 590 is 19.46 GPixel/s; Radeon 660M is 30.40 GPixel/s.
• Texture Rate: GTX 590 is 38.91 GTexel/s; Radeon 660M is 45.60 GTexel/s.
• FP32 Performance: GTX 590 is 1,244.2 GFLOPS; Radeon 660M is 1,459.2 GFLOPS.
• FP16 Performance: GTX 590 has none listed; Radeon 660M is 2.918 TFLOPS (2:1).
• TDP: GTX 590 is 365 W; Radeon 660M is 40 W.
• Slot Width: GTX 590 is Dual-slot; Radeon 660M is IGP.
• Power Connectors: GTX 590 is 2x 8-pin; Radeon 660M is None.
• Suggested PSU: GTX 590 is 750 W; Radeon 660M has none.
• Bus Interface: GTX 590 is PCIe 2.0 x16; Radeon 660M is PCIe 4.0 x8.
• Display Outputs: GTX 590 is 3x DVI, 1x mini-DisplayPort; Radeon 660M is Portable Device Dependent.
• DirectX Support: GTX 590 is 12 (11_0); Radeon 660M is 12 Ultimate (12_2).
• Vulkan Support: GTX 590 has none listed; Radeon 660M is 1.4.
• Dimensions: GTX 590 is 279 mm x 111 mm x 40 mm; Radeon 660M has no dimensions.
• Release Date: GTX 590 is 2011-03-23; Radeon 660M is 2022-01-03.
• Launch MSRP: GTX 590 is 699 USD; Radeon 660M has none.

Head-to-Head Benchmarks

The only direct benchmark comparison is Geekbench OpenCL, where the AMD Radeon 660M wins 12,876 to 12,830, a 0.4% margin. This result is remarkably close given the architectural differences. The GTX 590's nearest rival, AMD Radeon Pro 455, scores 12,831—effectively identical to the GTX 590's 12,830. The Radeon 660M's nearest rival, NVIDIA RTX A2000 Mobile, scores 13,821, which is 0.1% above the Radeon's average but well above its OpenCL-only score.

The Radeon 660M also has a Geekbench Vulkan score of 14,748, which is 14.5% higher than its own OpenCL score. This indicates that the RDNA 2.0 architecture scales well with modern APIs, whereas the GTX 590 has no Vulkan result at all, limiting its relevance in Vulkan-based workloads.

The GTX 590 does hold advantages in memory bandwidth (164.0 GB/s vs system-dependent) and ROP count (48 vs 16), which could matter in high-resolution, fill-rate-bound scenarios. However, the Radeon 660M's higher pixel rate (30.40 GPixel/s vs 19.46 GPixel/s) and texture rate (45.60 GTexel/s vs 38.91 GTexel/s) suggest it handles pixel and texture throughput more efficiently despite fewer execution units.

The FP32 delta is also notable: the Radeon 660M's 1,459.2 GFLOPS is 17.3% higher than the GTX 590's 1,244.2 GFLOPS. This computational advantage, combined with the OpenCL benchmark win, suggests the Radeon has a genuine compute edge despite being an integrated part.

The Verdict

The data points to a clear split based on use case. The AMD Radeon 660M is the better choice for compute workloads and modern API support: it wins the OpenCL benchmark, has a higher FP32 throughput, offers FP16 support, includes ray-tracing cores, supports Vulkan 1.4 and DirectX 12 Ultimate, and consumes just 40 W. Its average benchmark score of 13,812 places it at the 55th percentile, and its nearest rivals include discrete GPUs like the AMD Radeon RX 570X and NVIDIA RTX A2000 Mobile.

The NVIDIA GeForce GTX 590 is the better choice only for scenarios requiring dedicated high-bandwidth memory and high ROP counts. Its 164.0 GB/s bandwidth and 48 ROPs are unmatched by the Radeon 660M, but its 365 W TDP, dual-slot size, 699 USD launch MSRP, and lack of modern API support make it an obsolete choice for current workloads. Its 52nd percentile ranking and single benchmark result reinforce that it is outperformed by the newer, more efficient AMD part.

For any modern user, the AMD Radeon 660M is the logical pick: it wins the only direct test, scores higher on average, supports current APIs, and does so at a fraction of the power draw. The GTX 590 remains a historical artifact—a powerful-for-its-time dual-GPU card whose architecture cannot keep pace with a 2022 integrated processor in compute performance.