Within the last few weeks there has been quite a substantial amount of leaks regarding Nvidia’s upcoming RTX 30 series cards based on the Ampere GPU architecture, including the possible specs for the RTX 30 series cards, as well as leaked pictures of the graphics card’s design and even the cooler block design itself. Now it seems like some performance results have been leaked online, as some rumors have been going around online about some 3DMark benchmark results for one of the RTX 30 Ampere GPUs.
One thing that most of the rumors and leaks have in common are that there are currently 3 planned variants of the RTX 30 series, including the ¬¬gc_id:5003[RTX 3080]¬¬, ¬¬gc_id:5004[RTX 3080 Ti]¬¬, and the ¬¬gc_id:5005[RTX 3090]¬¬ which is set to replace the Titan variant instead. Each one is reported to be using the GA102 die, and according to these rumors, there are three different memory configurations: 10GB, 12GB, and 24GB respectively. These are pretty much exactly what was leaked before.
Whilst browsing some results in 3DMark, they stumbled across an interesting result for Time Spy by an undetermined Nvidia GPU. Time Spy is a DirectX12 benchmark test within 3DMark. Within these private results came some intriguing scores, including some strange clock speed numbers. None of this is officially confirmed, so keep a pinch of salt at the ready, but nevertheless we thought this was really interesting to look at, so we’d love to know your thoughts on it!
GPU Brand Name Nvidia Corporation GPU Clock Speed (Boost) 1935MHz Memory Clock Speed 6000MHZ
The information from the results above indicate that the GPU Vendor is indeed Nvidia Corporation, whilst the GPU core clock speed was marked at 1935MHz and the memory clock is shown as 6000MHz.
First of all, Time Spy only reports the boost clock speeds, which would explain the GPU core clock speed of 1935MHz above. Secondly, 6000MHz?? Apparently this usually happens when an early or internal driver is used instead, or possibly an unknown type of memory that the 3DMark detection software is unable to recognise.
Okay so now let’s get into all the juicy stuff, a list of scores was then compiled into the graph below, comparing this unknown Nvidia Ampere GPU to other graphics card results. From here we can get a good idea of how well this card actually performs, and the results are very interesting.
Breaking all that down, this means that this unknown Ampere graphics card is roughly 30.98% better than the stock Founders Edition of the ¬¬gc_id:4050[RTX 2080 Ti]¬¬, 21.07% better than the stock ¬¬gc_id:4149[RTX 2080 Ti Lightning Z]¬¬ from MSI, 22.14% better than the stock ¬¬gc_id:4075[Titan RTX]¬¬, 8.3% better than the top result for the ¬¬gc_id:4016[Titan V]¬¬ under LN2 (Liquid Nitrogen cooling), and only 2.18% less than an overclocked ¬¬gc_id:4060[RTX 2080 Ti XC]¬¬ by EVGA.
It’s most likely that this is the RTX 3080 tested here, which means even more performance gains with the higher variants like the RTX 3080 Ti or RTX 3090, and if that is the case then a nearly 31% increase in performance over the RTX 2080 Ti Founders Edition is pretty impressive. But let’s just hope that the boost clock speed is just an early version and the final speeds will be higher.
Either way, this year we will see quite possibly the most anticipated generation of graphics cards, and certainly the most anticipated PC hardware this year. So everyone is scrambling to find the most amount of info possible before Nvidia officially announces them. Take these results with a pinch of salt, but we thought they’d be an interesting look at what we can expect with Nvidia’s next generation of GPUs.
What do you think? What card is most likely tested here? Do you think the boost clock speeds will increase? When do you reckon we’ll hear something official from Nvidia? Let us know your thoughts!