Why Your Old GPU Might Beat Nvidia's New 50 Series

Tech Notice's latest benchmark testing reveals something the GPU marketing teams don't want you dwelling on: newer doesn't always mean faster, at least not for 3D rendering workloads.

Across hundreds of tests in Octane and Redshift—the render engines actually powering professional 3D work—the performance story is messier than the generation numbers suggest. The 4080 Super consistently beats the 5080 in Octane benchmarks. The 4070 Ti Super outpaces the 5070 Ti. In some cases, the gap is wide enough to make the older card the obvious choice, especially when you factor in the price difference between last-gen and current-gen hardware.

This isn't a quirk of one benchmark or a testing error. Tech Notice ran these tests across multiple scenarios, including 10-minute sustained runs to check for thermal throttling or performance degradation. The results held. As they note in the video: "The 4080 Super is actually better than the 5080. So, not all 3D performance will be increased with the newer 50 series."

What's Happening Here

The simplest explanation: VRAM matters more than architecture improvements for these workloads, and when the VRAM is identical between generations, the newer card's architectural tweaks don't translate to meaningful performance gains. Sometimes they translate to losses.

Octane, in particular, showed the biggest performance variance. The render engine scales nearly linearly with multiple GPUs—two RTX 4090s performed almost exactly twice as well as one—but the 50-series advantage over 40-series wasn't consistent across the product stack. The 5090 did pull ahead of the 4090 noticeably, but further down the lineup, the pattern broke.

Redshift told a similar story. Using both the older R24 benchmark (which doesn't support 50-series yet) and the newer R26 version, Tech Notice found that while the top-tier 5090 delivered modest improvements over the 4090, mid-range cards from the previous generation often matched or exceeded their supposed successors. The 4070 Ti Super beat the 3090 Ti handily, which shows how much the 40-series improved over 30-series. But the 4070 Super also beat the 5070, suggesting that particular jump didn't materialize.

The AMD Question

AMD's position in these benchmarks is unambiguous: they're not competitive for 3D rendering. The 7900 XTX, AMD's flagship from the previous generation, performed roughly on par with Nvidia's 4060 Ti in Redshift. The newer AMD 9000 series didn't improve the picture. In fact, Tech Notice found that "the best of the newest AMD... is actually slower than the last generation in terms of 3D rendering."

This isn't a matter of optimization or driver maturity. These render engines have had years to work with AMD hardware. The performance gap reflects fundamental architectural differences in how these GPUs handle the specific parallel workloads that 3D rendering demands. AMD has focused their recent efforts on AI capabilities—the AI Pro R9700 doubles VRAM specifically for those workloads—but for traditional 3D work, Nvidia maintains a substantial lead.

As Tech Notice summarizes: "If you're going with 3D, I hope at this point you've already noticed that going with Nvidia is the option for you."

What This Means For Your Wallet

The practical takeaway: if you're building or upgrading a 3D rendering workstation right now, you should be comparing specific benchmark numbers, not just generation numbers or product tier names. A 4080 Super purchased at current prices will likely outperform a more expensive 5080 in the software you actually use.

The naming confusion doesn't help. Nvidia's current lineup includes 4070, 4070 Super, 4070 Ti, and 4070 Ti Super—four distinct cards with meaningfully different performance profiles. Tech Notice cautions viewers to be "very careful with the namings of all of these GPUs" because the performance jumps between these variants can be larger than the jumps between generations.

Multi-GPU setups remain viable for Octane users, where the scaling is excellent. Two 4080s will substantially outperform a single 4090 in many scenarios, despite the 4090 being the higher-tier card. That math changes your cost-benefit analysis significantly.

The Benchmark Nobody Talks About

Tech Notice titled this video "The GPU Benchmark Nobody's Talking About" for a reason. Most GPU coverage focuses on gaming performance or general compute benchmarks. Manufacturer marketing emphasizes theoretical peak performance, new architecture features, or AI capabilities. Professional 3D rendering falls into a coverage gap—too specialized for consumer tech media, too hardware-focused for VFX industry publications.

Yet this is where actual purchasing decisions happen for a significant professional audience. A 3D artist choosing the wrong GPU based on generation hype rather than application-specific performance could easily waste a thousand dollars or more on a card that performs worse than the cheaper, older alternative.

The data also raises questions about how these cards are being designed and marketed. If the 5080 can't beat the 4080 Super in widely-used professional applications, what exactly is the value proposition? More efficient power consumption? Better ray-tracing performance in games? Those might matter to some buyers, but they're not what the product tier suggests you're paying for.

For anyone making GPU decisions based on 3D rendering performance, these benchmarks are worth more than the entire marketing budget behind the 50-series launch. The numbers are clear: check the specific workload performance before assuming newer equals better.