NVIDIA RTX PRO 4000 Blackwell vs NVIDIA Tesla M60 Comparison

FAQ

Q: How does the NVIDIA RTX PRO 4000 Blackwell compare to the Tesla M60 in the Geekbench Vulkan benchmark?

A: The RTX PRO 4000 Blackwell scores 194,168 in Geekbench Vulkan, while the Tesla M60 scores 31,473. The RTX PRO 4000 Blackwell wins the head-to-head matchup with a delta of -83.8%, meaning the M60 trails by approximately 84% behind the newer card.

Q: What are the average benchmark scores for both cards?

A: The Tesla M60 has an average benchmark score of 30,490 across its recorded tests. The RTX PRO 4000 Blackwell has an average score of 27,135 across its nine recorded benchmarks. The M60 sits at the 75th percentile of all GPUs in the database, while the RTX PRO 4000 Blackwell sits at the 72nd percentile.

Q: Which cards are closest to the Tesla M60 in average score?

A: The nearest rival to the M60 is the NVIDIA CMP 70HX with an average score of 30,476 and a delta of 0%. The AMD Radeon RX 6800 trails by 1.3% behind the M60's score of 30,433, while the GeForce RTX 3070 Ti trails by 1.8%.

Q: What is the memory configuration difference?

A: The Tesla M60 comes with 8 GB of GDDR5 memory on a 256-bit bus, delivering 160.4 GB/s of bandwidth. The RTX PRO 4000 Blackwell has 24 GB of GDDR7 memory on a 192-bit bus, offering 672.0 GB/s of bandwidth.

Q: What are the power requirements for each card?

A: The Tesla M60 has a TDP of 300 W, uses a single 8-pin power connector, and requires a suggested PSU of 700 W. The RTX PRO 4000 Blackwell has a TDP of 140 W, uses a single 16-pin power connector, and requires a suggested PSU of 300 W.

Q: What is the production status of the two models?

A: The Tesla M60 is end-of-life, with a release date of August 2015, and its successor is Tesla Pascal. The RTX PRO 4000 Blackwell is still active, released in March 2025, with its predecessor being Workstation Ada and no successor listed.

The Verdict

The database shows two entirely different product classes. The Tesla M60 is an end-of-life Maxwell-based accelerator with no display outputs, designed for compute deployments. The RTX PRO 4000 Blackwell is an active Blackwell workstation card with four DisplayPort 2.1b outputs. The recorded data indicates that for any modern workload, the RTX PRO 4000 Blackwell is the stronger pick, but the choice depends on the specific application.

For users who require a current-generation GPU with display output support, the RTX PRO 4000 Blackwell is the only viable option between the two, since the M60 has no display outputs at all. The RTX PRO 4000 Blackwell also offers a significantly higher FP32 throughput of 36.83 TFLOPS, compared to the M60's 4.825 TFLOPS. The RTX PRO 4000 Blackwell also includes 70 RT cores and 280 tensor cores, features entirely absent from the M60.

For legacy compute environments that rely on Maxwell-specific behavior or require the M60's dual-slot 300 W form factor, the older card still holds the 75th percentile in the database, slightly above the RTX PRO 4000 Blackwell's 72nd percentile. However, the head-to-head benchmark result is one-sided: the RTX PRO 4000 Blackwell wins the only shared test by 83.8%. The verdict is straightforward: choose the RTX PRO 4000 Blackwell for modern workloads, especially those needing ray tracing, tensor operations, or display connectivity. Choose the Tesla M60 only if a deployment is locked to the older Maxwell generation and needs the specific 8 GB GDDR5 configuration.

Head-to-Head Benchmarks

The only benchmark shared by both cards in the database is Geekbench Vulkan. In this test, the RTX PRO 4000 Blackwell scores 194,168, while the Tesla M60 scores 31,473. This gives the RTX PRO 4000 Blackwell a win with a delta of -83.8%, meaning the M60's score is about 84% below the Blackwell card's result. This is a decisive margin, and it aligns with the generational gap between Maxwell 2.0 and Blackwell 2.0.

Looking at the broader benchmark records for the RTX PRO 4000 Blackwell, the card scores 4,648 in 3DMark Steel Nomad DX12, 28,427 in PassMark G3D, 14,805 in PassMark GPU Compute, and 1,265 in PassMark G2D. The PassMark DirectX results show a spread across DirectX versions: 354 in DX9, 276 in DX11, 173 in DX10, and 97 in DX12. The M60's only other recorded score is Geekbench OpenCL at 29,506.

The M60's nearest rival data puts it in a competitive cluster: the CMP 70HX is at 0% delta, the RX 6700 is 0.2% behind, the RX 6800 is 1.3% behind, and the RTX 3070 Ti is 1.8% behind. This shows the M60's average score is very close to several modern consumer cards, despite its age. The RTX PRO 4000 Blackwell's nearest rivals show a slightly different picture: the RX 6700 XT is -1.1% behind, the RTX 4070 Mobile is -1.1% behind, the RTX 3090 is -1.6% behind, and the RTX A4000 is 1.7% ahead. The RTX PRO 4000 Blackwell sits in a similar performance tier as the RTX 3090 in average score, which is a notable result for a 140 W single-slot card.

Specification Differences

The two cards differ in nearly every measurable specification. The Tesla M60 uses the GM204 chip with a 28 nm process, while the RTX PRO 4000 Blackwell uses the GB203 chip on a 5 nm process. Transistor counts differ dramatically: the M60 has 5,200 million transistors on a 398 mm² die, giving a density of 13.1 million transistors per mm². The RTX PRO 4000 Blackwell packs 45,600 million transistors on a smaller 378 mm² die, reaching a density of 120.6 million per mm².

Clock speeds also diverge. The M60 runs at a base clock of 557 MHz and boosts to 1178 MHz, with memory at 1253 MHz (5 Gbps effective). The RTX PRO 4000 Blackwell runs at 1230 MHz base and 2055 MHz boost, with memory at 1750 MHz (28 Gbps effective). Memory capacity is 8 GB GDDR5 on the M60 versus 24 GB GDDR7 on the RTX PRO 4000 Blackwell. Bus widths differ as well: 256-bit for the M60 and 192-bit for the RTX PRO 4000 Blackwell, with bandwidth at 160.4 GB/s versus 672.0 GB/s, respectively.

The compute units show a large gap. The M60 has 2048 shading units, 128 TMUs, and 64 ROPs. The RTX PRO 4000 Blackwell has 8960 shading units, 280 TMUs, 96 ROPs, 70 RT cores, and 280 tensor cores. Pixel rate is 75.39 GPixel/s for the M60, compared to 197.3 GPixel/s for the RTX PRO 4000 Blackwell. Texture rate is 150.8 GTexel/s versus 575.4 GTexel/s. FP32 performance is 4.825 TFLOPS for the M60, while the RTX PRO 4000 Blackwell delivers 36.83 TFLOPS, with matching FP16 performance at 36.83 TFLOPS (1:1).

Power and physical specifications also differ. The M60 is a dual-slot card with a 300 W TDP, a single 8-pin connector, and a suggested 700 W PSU. It measures 267 mm (10.5 inches) in length. The RTX PRO 4000 Blackwell is a single-slot card with a 140 W TDP, a single 16-pin connector, and a suggested 300 W PSU. It measures 241 mm (9.5 inches) in length, 111 mm (4.4 inches) in height, and 20 mm (0.8 inches) in width. The bus interface is PCIe 3.0 x16 on the M60 versus PCIe 5.0 x16 on the RTX PRO 4000 Blackwell. The M60 has no display outputs, while the RTX PRO 4000 Blackwell has four DisplayPort 2.1b outputs. The M60 supports DirectX 12 (12_1), while the RTX PRO 4000 Blackwell supports DirectX 12 Ultimate (12_2). Both support OpenGL 4.6 and Vulkan 1.4.

Architecture Differences

The Tesla M60 is built on Maxwell 2.0 architecture with the GM204 chip, fabricated by TSMC on a 28 nm process. Maxwell 2.0 was released in the Tesla Maxwell generation with the Mxx naming scheme. The M60 does not include RT cores or tensor cores, as those features were not part of the Maxwell architecture. Its FP16 performance is not listed in the database, confirming that the card lacks dedicated half-precision throughput.

The RTX PRO 4000 Blackwell is built on Blackwell 2.0 architecture with the GB203 chip, also fabricated by TSMC but on a 5 nm process. It belongs to the Blackwell PRO W generation with the x000 naming scheme. The architecture includes 70 RT cores for ray tracing and 280 tensor cores for AI acceleration. FP16 performance is listed at 36.83 TFLOPS (1:1), meaning the card achieves the same throughput for half-precision as for single-precision operations.

The transistor density difference is a direct result of the process node change: 13.1 million transistors per mm² on 28 nm versus 120.6 million per mm² on 5 nm. This is a 9.2x improvement in density, enabling the RTX PRO 4000 Blackwell to fit over eight times the transistor count on a physically smaller die. The architecture also brings a newer DirectX 12 Ultimate feature set, which includes hardware ray tracing support, whereas the M60 is limited to DirectX 12 (12_1) without RT support. The display output situation also reflects architectural direction: Maxwell-era Tesla cards were compute-only, while Blackwell PRO cards are designed for workstation use with multiple DisplayPort outputs.

Where Each One Wins

The RTX PRO 4000 Blackwell wins in every modern compute and graphics scenario tracked in the database. It wins the only shared benchmark, Geekbench Vulkan, by 83.8%. Its FP32 throughput of 36.83 TFLOPS is over seven times the M60's 4.825 TFLOPS. Memory bandwidth is 672.0 GB/s versus 160.4 GB/s, a more than fourfold advantage. The RTX PRO 4000 Blackwell also has 70 RT cores and 280 tensor cores, which the M60 lacks entirely. For any workload involving ray tracing, AI inference, or modern DirectX 12 Ultimate features, the RTX PRO 4000 Blackwell is the only choice between these two. The card's 24 GB of GDDR7 memory also makes it suitable for large datasets that would exceed the M60's 8 GB capacity.

The Tesla M60 wins in a narrower set of circumstances. It holds a slightly higher percentile rank in the database at 75, compared to 72 for the RTX PRO 4000 Blackwell, indicating its average score in the database is positioned slightly better relative to all GPUs. The M60's average benchmark score of 30,490 also exceeds the RTX PRO 4000 Blackwell's average of 27,135, though this is due to the different mix of tests recorded for each card. The M60 also has a wider 256-bit memory bus, which can be an advantage in bandwidth-sensitive legacy workloads that are optimized for that bus width. For deployments that require the Maxwell architecture specifically, such as compatibility with older compute stacks, the M60 remains the applicable part. The M60's dual-slot form factor and 300 W TDP may fit in infrastructure designed for high-power accelerator slots, though the RTX PRO 4000 Blackwell's lower 140 W TDP and single-slot size make it easier to deploy in dense systems. The M60 has no display outputs, so it cannot drive monitors, while the RTX PRO 4000 Blackwell can output to four displays via DisplayPort 2.1b. In summary, the RTX PRO 4000 Blackwell wins for all modern, feature-rich workloads, while the Tesla M60 retains relevance only for legacy Maxwell-specific compute tasks or environments where its exact power, slot, and memory configuration is required.