The first picture of NVIDIA's Ampere GA102 GPU which will be powering the next-generation GeForce RTX 3090 & GeForce RTX 3080 graphics card has leaked out over at Chiphell. NVIDIA flagship gaming graphics cards will be making use of the GA102 GPU to deliver unprecedented performance gains over the Turing TU102 GPU core.
NVIDIA Ampere GA102 GPU For High-End & Enthusiast Gaming GeForce RTX 3090 & RTX 3080 Graphics Cards Pictured
[UPDATE 26/08/20] - Rogame has posted a more detailed analysis of the leaked GA102 GPU die. The die was mirrored to complete the entire package and it turns out that the GA102 GPU could measure around 628mm2 which is slightly smaller than the TU102 GPU die (754mm2). What is interesting about this die size projection is that it ends up near to what Kopite7kimi had reported back in July. The leaker reported a die size of GA102 to be exactly 627.12mm2.
Posted by Chiphell user PolyMorph, the die seems for the NVIDIA GeForce RTX 3090 graphics card which is more or less confirmed to be the flagship gaming offering in the Ampere GeForce RTX 30 series lineup. The specific chip is the GA102-300-A1 and this SKU has previously been hinted to be utilized by the flagship GeForce RTX 3090 graphics card. The GA102-300-A1 chip pictured is in the qualification state but we also got to see pictures of Qual-Sample GPUs on several pre-productions PCBs.
Several unique identifiers such as the date the chip was produced and the specific plant where it was fabricated have been blurred out. It's done to remove any traces that can get the leaker into trouble. We can expect different PCBs for the GeForce RTX 3090 and GeForce RTX 3080 based on the recent leaks. The GeForce RTX 3090 is going to feature a triple-slot design while the GeForce RTX 3080 will stick to a dual-slot cooler.
NVIDIA GeForce RTX 3090 - The Fastest Ampere Gaming Graphics Card
According to leaked details, the GeForce RTX 3090 will feature the GA102-300-A1 GPU. The GA102-300-A1 GPU is said to be comprised of 5248 CUDA cores or 82 SMs. In total, that's a 20% increase in cores over the GeForce RTX 2080 Ti. No details such as clock speeds, TMU/ROP counts are provided and we can't take the Ampere A100 die as a reference since gaming & HPC parts share a different configuration hierarchy.
Moving on to the memory side, things start to look interesting as it is stated that the card will get up to 24 GB memory with faster GDDR6X pin speeds. Since that's a 384-bit bus we are looking at, it will be able to offer nearly 1 TB/s bandwidth.
NVIDIA GeForce RTX 3080 - GA102 For High-End Gaming
Moving on to the GeForce RTX 3080, the rumor reports that the card will be featuring the GA102-200-KD-A1 SKU. This cut down SKU will feature the same 4352 CUDA cores as the RTX 2080 Ti that will be arranged in a total of 68 SMs. The card is reportedly going to feature up to 20 GB of memory that is also going to be GDDR6X. Assuming the memory is running at 19 Gbps across a 320-bit bus interface, we can expect a bandwidth of up to 760 GB/s.
The switch to GA102 from TU104 on the RTX 2080 is definitely an interesting one. A high-end SKU would also result in higher wattage and thermals but I believe this could be done to raise the bar up on the sub $500 segment which should comprise of GA104 based RTX 3070 and RTX 3060. The latest performance leaks also show a considerable performance improvement over the RTX 2080 and even the RTX 2080 Ti in synthetic benchmarks. If the same could translate over to the gaming performance, then Ampere does look like a super enticing product for gamers who have waited for almost 2 years now.
The NVIDIA GeForce RTX 30 'Ampere' Gaming graphics card lineup is expected to be announced on 1st September at a digital event which will be hosted by CEO of NVIDIA, Jensen Huang. We expect at the very least, three new graphics cards during the event which include the GeForce RTX 3080 Ti (or RTX 3090), GeForce RTX 3080, and a next-generation Titan. There's definitely a lot more to talk about regarding the Ampere GPUs but we will have to wait until more information comes our way.