While early experiences have already claimed that Nvidia will wait till 2025 to launch the RTX 5090, the primary rumours in regards to the upcoming GPU are already making the rounds on-line – together with experiences that it may very well be as a lot as 70% sooner than the RTX 4090.
The rumours had been first shared by Panzerlied (by way of VideoCardz), the identical leaker who shared the primary particulars on Nvidia’s next-generation of client GPUs and that NVIDIA intends to skip the xxx04-class GPUs of their GPU sequence. Following Panzerlied’s submit, VideoCardz requested kopite7kimi about these rumours, which the leaker corroborated.
According to Panzerlied (primarily based on DeepL translation), the RTX 5090 would have a 50% improve in scale, a 52% improve in reminiscence bandwidth, a 78% improve in L2 cache, and 15% greater clock speeds. All this is able to reportedly internet a 70% enchancment in efficiency.
Taking the RTX 4090 specs under consideration, a 52% reminiscence bandwidth improve (about 1.54 TB/s) means we might see the RTX 5090 utilizing both 24 GB of GDDR7 reminiscence working at 32 Gbps. Other alternate options could be utilizing a 512-bit reminiscence bus with 24 Gbps reminiscence or a 448-bit reminiscence bus with 28 Gbps reminiscence. As for the clock pace, it is doable the upcoming flagship card will increase near the three.0 GHz mark, and in some instances, even surpass it. Lastly, a 78% improve in L2 cache means we will anticipate 128 MB of L2 cache on the RTX 5090.
Although all this seems promising, it’s going to take time to get affirmation on all of this. After all, the RTX 5090 and different RTX 50 sequence GPUs are anticipated to be launched in 2025.
Discuss on our Facebook web page HERE.
KitGuru says: What sort of generational uplift are you anticipating from the RTX 50 sequence? Does a 70% enchancment over the already mighty RTX 4090 appear believable?
