Nvidia and AMD are currently working on their next generations of GeForce RTX and Radeon RX. New rumors emerge. They deal with various issues including that of video memory.
The next family of GeForce RTX graphics cards will not necessarily be called the GeForce RTX 40 series. This name has now become “temporary” through several leaks. Nvidia will likely adopt the GeForce RTX 50 series name for its Ada Lovelace generation. kopite7kim Mention these solutions as **90/**80 etc. just in case they aren’t called RTX 40 but RTX 50.
GeForce RTX Ada LoveLace
The new speculation about video memory does not evoke 24 Gbps, but 21 Gbps. We would be faced with a profile identical to that of the new GeForce RTX 3090 Ti. The thermal envelope is said to reach 600 watts. This number is plausible with regard to the PCIe 5.0 connection that the current 3090 Ti is equipped with.
GeForce RTX “Ada lovelace”.
| GPU’s | GA102 | AD102 |
| Architecture | amp | Ada Lovelace |
| engraving | 8nm | 5nm |
| Graphics Processing Cluster (GPC) | 7 | 12 |
| Texture Processing Cluster (TPC) | 42 | 72 |
| Streaming Multiprocessors (SM) | 84 | 144 |
| CUDA cores | 10,752 | 18,432 |
| L2 cache | 6MB | 96MB |
| Power TFLOPs | 37.6 | 90? |
| storage type | GDDR6X | GDDR6X |
| memory bus | 384-bit | 384-bit |
| storage capacity | 24GB (3090) | 24GB (4090 or 5090?) |
| TGP | 350W | 600W |
| release date | 09/20 | 2H 2022? |
Based on rumours
The Ada Lovelace AD102 GPU is said to consist of 144 SMs, a 71% increase over the current GA102 GPU! At the core of its mechanics would be no less than 18,432 CUDA cores at work, as well as even a 384-bit bus, enough to guarantee a bandwidth of 1004 Gb/s.
GeForce RTX 5090 or RTX 4090, a monster equipped with an AD102 GPU and 24GB of memory
Radeon RX RDNA 3 (Radeon RX 7950 XT?)
On the side of AMD it mumble that the next Navi 31, the flagship GPU of the RDNA 3 family (Radeon RX 7950 XT?) will be equipped with 7 chiplets. Not everything is clear on this topic. Normally, AMD should use an MCM contraction approach of multi-chip modules with 2 GCD (5nm) and 4 MCD (6nm), all accompanied by an IOD.
This hypothetical Radeon RX 7950 XT with its Navi 31 GPU should offer 15,360 stream processors clocked at 2.5 GHz. We do not know whether this frequency represents the maximum possible value or the base value. Meanwhile, that’s an 11% improvement over the Radeon RX 6900 XT, enough with further changes to make for a theoretical performance of 38.4 TFLOPs in the FP32, a nice +67% over the Radeon RX6900XT. The card is said to be accompanied by an impressive TDP of 500 watts, again an increase of 67% compared to its predecessor.
In terms of equipment, it cannot be ruled out that the 32 GB RAM offers twice as much as the current AMD flagship. The operation of this GDDR6 at 21 Gbps would be guaranteed by a 256-bit bus. The Infinity Cache should also double (512 MB).
Here’s an early review based on rumors from VideoCardz.
| GeForce RTX4090 | Radeon RX7950XT | |
| delicacy of engraving |
TSMC-N5 |
TSMC N5/N6 |
| Architecture | Ada Lovelace | RDNA3 |
| design | Monolithic | Multi-Chip Module (MCM) |
| dies | 1 | 2 GCDs + 4 MCDs + 1 IOD |
| MEGA cluster | 12 GPC (Graphics Processing Cluster) | 2×3 shader engines |
| SUPER cluster | 72 TPCs (Texture Processing Clusters) | 2×30 WGP (RDNA) |
| GPU cluster | 144 SMS | – |
| FP32 cores | 18,432 CUDA cores | 15,360 stream processors |
| GPU frequency | ~2.2GHz | ~2.5GHz |
| R.A.M. | 24GB GDDR6X | ? GB GDDR6 |
| memory frequency and memory bus | 21 Gbps – 384 bits | ? Gbps – 256 bits |
| Hidden | 96MB L2 cache | 256 or 512 MB Infinity Cache |
| consumption | 600W (TDP) | 500W (TBP) |
| release date | Q3/Q4 2022 | Q3/Q4 2022 |
| perfomance | 2x GA102 |
2.5x NAVI 21 |