NVIDIA, AMD squabbling over whose GPUs are faster for DeepSeek's AI benchmarks

NVIDIA, AMD squabbling over whose GPUs are faster for DeepSeek's AI benchmarks

FP Staff February 10, 2025, 14:51:43 IST

AMD sparked the rivalry by releasing benchmark results that showed its RX 7900 XTX outpacing NVIDIA’s RTX 4090 and RTX 4080 Super in DeepSeek R1 tests. NVIDIA was quick to counter AMD’s claims showing a106% advantage using Distill Llama 8B

Advertisement
NVIDIA, AMD squabbling over whose GPUs are faster for DeepSeek's AI benchmarks
Image Credit: Reuters, Reuters

NVIDIA and AMD have locked horns in a heated debate over which of their GPUs can best handle DeepSeek’s AI benchmarks. Both companies have released competing test results for DeepSeek R1, each claiming dominance in speed and performance.  

AMD’s RX 7900 XTX has been pitted against NVIDIA’s high-end RTX 4090 and RTX 4080 Super, with both tech giants eager to prove their hardware superiority. However, beyond performance claims, AMD’s competitive edge may ultimately lie in its pricing.

STORY CONTINUES BELOW THIS AD

AMD kicks off the benchmark battle

AMD sparked the rivalry by releasing benchmark results that showed its RX 7900 XTX outpacing NVIDIA’s RTX 4090 and RTX 4080 Super in DeepSeek R1 tests. According to AMD’s David McAfee, their RDNA3-based GPU was up to 13 per cent faster than the RTX 4090 and 34 per cent faster than the RTX 4080 Super in certain tasks.

In tests using DeepSeek R1 Distill Qwen 7B, AMD’s GPU maintained a 13 per cent lead over the RTX 4090. Other benchmarks, such as Distill Llama 8B and Distill Qwen 14B, saw AMD’s card outperform by 11 per cent and 2 per cent, respectively. However, NVIDIA did have a slight edge in the largest task, Distill Qwen 32B, where the RTX 4090 led by 4 per cent. Against the RTX 4080 Super, AMD’s GPU showed even greater dominance, with performance boosts ranging from 22 per cent to 34 per cent across various benchmarks.

Of course, AMD’s tests, optimised for its own hardware, are unlikely to favour NVIDIA’s GPUs. Given this, scepticism around the accuracy of manufacturer-released benchmarks is warranted.

NVIDIA strikes back

NVIDIA was quick to counter AMD’s claims and a 106 per cent advantage using Distill Llama 8B. Even the RTX 4090, according to NVIDIA, outperformed AMD’s GPU by 47 per cent in both benchmarks.

NVIDIA, AMD squabbling over whose GPUs are faster for DeepSeek's AI benchmarks 2-2025-02-be94a7b639726dca065643f620977c54
Image Credit: NVIDIA

However, as tech outlet Tom’s Hardware noted, different test conditions, optimisations, and driver versions can influence these results. Benchmark battles between companies often involve cherry-picking tests and setups that favour their own hardware, leaving consumers to sift through the conflicting data.

AMD’s secret weapon: Price

While NVIDIA may boast superior performance in its own tests, AMD has a strong selling point: affordability. The RX 7900 XTX is significantly cheaper than both the RTX 4090 and RTX 4080 Super. For many buyers, especially those prioritising cost-efficiency, this price advantage may outweigh slight differences in benchmark performance.

Although AMD’s RDNA3 architecture and AI Accelerator technology are not yet as widely adopted in AI applications as NVIDIA’s offerings, the growing focus on price-performance balance could help AMD capture a larger share of the AI market.

As the GPU wars rage on, consumers and developers must navigate the hype, keeping in mind that manufacturer-released benchmarks often tell only part of the story. Ultimately, factors like pricing, software compatibility, and long-term support may play just as crucial a role as raw performance in shaping purchasing decisions.

End of Article
Latest News
Find us on YouTube
Subscribe
End of Article

Top Shows

Vantage Firstpost America Firstpost Africa First Sports