The detailed specifications for NVIDIA’s next-generation GeForce RTX 3090, RTX 3080 & Titan RTX have been released. Many of the rumors about the new graphics card lineup were getting popular and it was confirmed that the lineup will make 4k gaming the mainstream in 2020. Well, with the introduction of the RTX 3090, it seems like NVIDIA might be taking Ampere to a whole new level.
RTX 3090 and RTX 3080 Specifications
With many rumors and speculations, the news that GeForce RTX 3080 and GeForce RTX 3090 will be the first cards featuring the Ampere Gaming GPUs was getting popular. The specifications came out from KatCorgi and Kopite7kimi (via Videocardz). Kopite7kimi has given accurate details in recent times. The A100 GPU leaks were also confirmed by Kopite7kimi. We should treat these leaks as a pinch of salt as they are not confirmed yet. We previously predicted that NVIDIA might be rebranding the RTX 3000’s Titan as the 3090, but this new leak seemingly disproves it.
NVIDIA GeForce RTX 3090
The NVIDIA GeForce RTX 3090 is rumored to be the fastest Ampere graphics card. NVIDIA can choose this nomenclature to name the graphics card. But it could be possible that the RTX 3090 is an internal codename of RTX 3080 Ti. However, this time the card is getting called GeForce RTX 3090. According to the specifications, the GA102-300-A1 GPU will be featured on the GeForce RTX 3090. The GPU will be having 5248 CUDA cores or 82 SMs. That will be a 20% increase in cores compared to RTX 2080 Ti. Other details like clock speeds, TMU/ROP counts are not provided yet. Coming to the memory, It will get up to 12 GB memory with speed up to 21 Gbps. We can expect up to 1 TB/s bandwidth because it is a 384-bit bus. The memory designed is referred to as GDDR6X. However, the information is not yet confirmed by the vendors.
NVIDIA GeForce RTX 3080
Coming to the Geforce RTX 3080, it will feature GA102-200-Kx-A1 SKU. Reports also say that the same 4352 CUDA cores as RTX 2080 Ti will be featured on the SKU that will be arranged in 68 SMs. The card will feature up to 10 GB memory with speed up to 19 Gbps. We can expect up to 760 GB/s bandwidth as it is a 320-bit bus. The memory designed is also referred to as GDDR6X.
The SKU is changed from TU104 to GA102 and is expected to give better performance. The SKU will also result in higher wattage and thermals.
NVIDIA 2nd Gen Titan RTX
The card is based on Unlocked Ampere GA102 GPU. The card will be the 2nd Gen Titan RTX card and it will be featuring the Ampere GA102-400-A1 GPU. The GPU will be having 5376 CUDA cores that will be arranged in 84 SMs. The 24 GB of GDDR6X memory is featured with a 384-bit bus interface. It will feature a bandwidth of 816 GB/s. The bandwidth is slightly lower than RTX 3090 which is due to a slower pin speed of 17 Gbps. For compensation, the users will be provided the enhanced memory compression architecture and twice the size of memory buffer featured in RTX 3090.
NVIDIA’s CEO Jensen has confirmed that all next-generation products will be featuring the same NVIDIA Ampere GPU architecture.
The lineup is expected to launch in September. Enthusiasts have many expectations from the lineup. We’ll have to wait for the launch to see if those are met or not.