AMD Radeon HD 8790M vs NVIDIA Quadro 4000 Comparison

The AMD Radeon HD 8790M and the NVIDIA Quadro 4000 make for an unusual pairing: a mobile GCN part from 2013 against a workstation Fermi card from 2010. The database shows them landing within less than one percent of each other in the one benchmark both have recorded, yet the hardware underneath those scores diverges sharply. One is a compact 77 mm² chip built for laptops, the other a 529 mm² workstation die pulling over a hundred watts. Which one the data favors depends almost entirely on what the workload asks for.

Head-to-Head Benchmarks

The only direct comparison in the database is Geekbench OpenCL, and it is remarkably close. The Radeon HD 8790M scores 5017 while the Quadro 4000 scores 4979, a gap of just 0.8 percent in AMD's favor. That is a statistical dead heat by most standards, but it is interesting nonetheless: a mobile chip with a small die and no external power connector edges out a full workstation card in a general-purpose compute test.

Context from each card's rival list sharpens the picture. The HD 8790M sits at an average benchmark score of 5691, placing it in the 33rd percentile of all GPUs in the database. Its Vulkan result of 6365 is the stronger of its two recorded scores, pulling that average above its OpenCL figure. The Quadro 4000 averages 4979 across its recorded results and sits in the 29th percentile. So while the single head-to-head test is a near-tie, the HD 8790M's overall standing is modestly higher.

Curiously, the two cards' rival neighborhoods barely overlap. The HD 8790M's nearest rivals include the Intel Iris Pro Graphics P6300 (5712, a 0.4 percent gap), the GeForce GTX 670MX (5721, 0.5 percent), the GTX 550 Ti (5731, 0.7 percent), and the Quadro M500M (5604, 1.6 percent). The Quadro 4000's rivals are a different crowd entirely: the RTX 5060 Ti 16 GB (4970, 0.2 percent), the Radeon R7 Graphics (4998, 0.4 percent), the R5 M430 (5018, 0.8 percent), and the R7 M360 (4931, 1.0 percent). Two cards with nearly identical OpenCL numbers end up rubbing shoulders with completely different tiers of hardware, which hints that the single shared benchmark understates their differences.

Architecture Differences

The architectural gap between these two is wide. The HD 8790M is built on TSMC's 28 nm process using AMD's GCN 1.0 architecture, codenamed Mars. It packs 950 million transistors into a 77 mm² die, a density of 12.3 million transistors per square millimeter. The Quadro 4000 uses the much older 40 nm node and NVIDIA's Fermi architecture, chip GF100, with 3,100 million transistors spread across 529 mm², yielding only 5.9 million transistors per square millimeter. The AMD part achieves roughly double the transistor density on a die nearly seven times smaller.

Their execution resources are structured differently too. The HD 8790M has 384 shading units, 24 TMUs, and 8 ROPs, with a base clock of 850 MHz and a boost clock of 900 MHz. The Quadro 4000 has 256 shading units, 32 TMUs, and a much larger complement of 32 ROPs, with no base or boost clock recorded in the database. The ROP difference is telling: the Quadro 4000 delivers a pixel rate of 7.600 GPixel/s versus 7.200 GPixel/s for the HD 8790M, despite the AMD card's higher throughput elsewhere.

Memory subsystems also diverge. Both carry 2 GB of GDDR5, but the Quadro 4000 pairs a 256-bit bus with memory running at 702 MHz (2.8 Gbps effective) for 89.86 GB/s of bandwidth. The HD 8790M uses a narrower 128-bit bus with faster 1000 MHz memory (4 Gbps effective), reaching 64.00 GB/s. The workstation card holds a clear bandwidth advantage of roughly 40 percent.

Software support separates them as well. The HD 8790M reports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. The Quadro 4000 reports DirectX 12 (11_0) and OpenGL 4.6, with no Vulkan support listed. Both are end-of-life products; the HD 8790M launched at the end of March 2013, the Quadro 4000 at the start of November 2010.

Where Each One Wins

The HD 8790M wins on raw compute throughput and efficiency-oriented metrics. Its FP32 output of 691.2 GFLOPS exceeds the Quadro 4000's 486.4 GFLOPS by a substantial margin, and that shows up in the Geekbench OpenCL result where the mobile part leads by 0.8 percent. Its texture rate of 21.60 GTexel/s also beats the Quadro's 15.20 GTexel/s, a 42 percent advantage driven by higher clocks on its shading hardware. Add the Vulkan API support and its higher 33rd percentile standing, and it is the stronger card for compute-style and texture-heavy workloads.

The Quadro 4000 wins on pixel throughput and memory bandwidth. Its 32 ROPs push it to 7.600 GPixel/s against 7.200 GPixel/s, and its 89.86 GB/s of bandwidth comfortably outpaces 64.00 GB/s. For workloads that are bandwidth-starved or fill-rate-bound, the workstation card holds the edge. It also offers fixed display outputs, 1x DVI and 2x DisplayPort, versus the HD 8790M's "portable device dependent" outputs, and a standard PCIe 2.0 x16 interface rather than MXM-A (3.0).

FAQ

Q: Which GPU is faster in benchmarks?

A: The AMD Radeon HD 8790M leads in the one shared test, Geekbench OpenCL, scoring 5017 versus 4979 for the Quadro 4000, a 0.8 percent gap. It also holds a higher percentile ranking, 33rd versus 29th, across the database.

Q: Which one has more memory bandwidth?

A: The Quadro 4000, with 89.86 GB/s on a 256-bit bus, compared to 64.00 GB/s on the HD 8790M's 128-bit bus. Both use 2 GB of GDDR5.

Q: Do these GPUs support Vulkan?

A: The HD 8790M does, with Vulkan 1.2.170 listed. The Quadro 4000 has no Vulkan support recorded in the database.

Q: How do their manufacturing processes compare?

A: The HD 8790M is a 28 nm chip with 950 million transistors on a 77 mm² die. The Quadro 4000 is a 40 nm chip with 3,100 million transistors on a 529 mm² die.

Q: How power-hungry is each card?

A: The Quadro 4000 has a recorded TDP of 142 W, a single-slot design, one 6-pin power connector, and a suggested 300 W PSU. The HD 8790M is an MXM module with no power connectors and no TDP recorded in the database.

Q: Are these GPUs still in production?

A: No. Both are listed as end-of-life. The Quadro 4000 launched in November 2010 with a launch MSRP of 1,199 USD; the HD 8790M followed in March 2013.

The Verdict

The data points to the HD 8790M as the stronger all-around performer. It wins the only head-to-head benchmark, posts higher theoretical compute (691.2 versus 486.4 GFLOPS FP32), delivers more texture throughput (21.60 versus 15.20 GTexel/s), and carries the higher database percentile. It is also the more modern design by a wide margin: a newer architecture, a newer process node, and Vulkan support the Quadro lacks.

The Quadro 4000 still earns its place in specific scenarios. Its 256-bit bus and 89.86 GB/s of bandwidth, its 32 ROPs and superior pixel rate, and its fixed DisplayPort and DVI outputs make it the pick for fill-rate and bandwidth-sensitive work in a desktop PCIe slot. Buyers choosing purely on benchmark scores, however, should follow the data toward the HD 8790M.

Specification Differences

| Specification | AMD Radeon HD 8790M | NVIDIA Quadro 4000 |

|---|---|---|

| Architecture | GCN 1.0 | Fermi |

| Chip | Mars | GF100 |

| Generation | Solar System (HD 8700M) | Quadro Fermi (x000) |

| Process node | 28 nm | 40 nm |

| Transistors | 950 million | 3,100 million |

| Die size | 77 mm² | 529 mm² |

| Transistor density | 12.3M / mm² | 5.9M / mm² |

| Base clock | 850 MHz | not recorded |

| Boost clock | 900 MHz | not recorded |

| Memory speed | 1000 MHz (4 Gbps effective) | 702 MHz (2.8 Gbps effective) |

| Bus width | 128 bit | 256 bit |

| Memory bandwidth | 64.00 GB/s | 89.86 GB/s |

| Shading units | 384 | 256 |

| TMUs | 24 | 32 |

| ROPs | 8 | 32 |

| Pixel rate | 7.200 GPixel/s | 7.600 GPixel/s |

| Texture rate | 21.60 GTexel/s | 15.20 GTexel/s |

| FP32 | 691.2 GFLOPS | 486.4 GFLOPS |

| TDP | not recorded | 142 W |

| Slot width | MXM Module | Single-slot |

| Power connectors | None | 1x 6-pin |

| Suggested PSU | not recorded | 300 W |

| Bus interface | MXM-A (3.0) | PCIe 2.0 x16 |

| Display outputs | Portable Device Dependent | 1x DVI, 2x DisplayPort |

| DirectX | 12 (11_1) | 12 (11_0) |

| Vulkan | 1.2.170 | not listed |

| Release date | March 2013 | November 2010 |

| Predecessor | London | Quadro FX Tesla |

| Successor | Gem System | Quadro Kepler |

| Launch MSRP | not recorded | 1,199 USD |

| Percentile vs all GPUs | 33 | 29 |

| Average benchmark score | 5691 | 4979 |