RTX 5050 FurMark leak shows slower performance than an AMD RX 6000 mobile GPU
Nvidia unveiled its entry-level RTX 5050 desktop GPU with an MSRP of $249 earlier this week. In its performance graphs, it compared the GPU to the two-generation old RTX 3050 and the RTX 4060 from the last generation. While it outperformed the RTX 30 series card, it was slower than the last-gen RTX 40 series card in raw performance. Now, a new third-party benchmark leak has shown that it is slower than AMD’s RX 6000 series mobile GPU and Nvidia’s own RTX 20 series card as well.
The screenshots shared by Richard Huynh via GPU Magick show FurMark benchmark results at 4K resolution. The Palit RTX 5050 running an older OpenGL driver scored 1,978 and boosted up to 2,940 MHz (2.94 GHz). As per Nvidia’s specs, the RTX 5050 has a boost clock of 2.57 GHz so the FurMark boost clock speed is pretty impressive. What’s not so impressive is the score that places the desktop GPU lower than AMD Radeon RX 6600M laptop GPU, as well as the desktop RTX 2060.
Notably, the RTX 3050, which Nvidia compares the RTX 5050 with, was much slower in the FurMark score database. That being said, the lower score compared to the two-generation old AMD laptop GPU and three-generation old Nvidia desktop GPU does not bode well for the RTX 5050 that is priced only $50 lower than the RTX 5060.
It should be noted that the FurMark test for the RTX 5050 was done with an older Intel CPU, the Core i7-9700, which does not even support PCIe Gen 4. Secondly, it was a 4K benchmark typically designed to test the cooling capabilities rather than true performance of a GPU. With that in mind, the thermal results seem quite inaccurate as the GPU reported only 39 degrees C max core temperature under 100% load.
Even though the FurMark leak can’t be fully trusted, it does raise some questions about the RTX 5050 and its value proposition. Though it is the cheapest card in the RTX 50 series lineup, board partners will price the GPU above MSRP, inching closer to the MSRP of the RTX 5060 that has a significantly higher CUDA core count and newer GRRD7 memory.
Source(s)
No comments for this article
