Benchmarks of the now canceled NVIDIA GeForce RTX 4080 12 GB graphics card have made their way to Chiphell Forums, giving us a look at what could've been the performance of the AD104 GPU.
NVIDIA GeForce RTX 4080 12 GB Graphics Card Benchmarks Leak Out, Cancelled Card Was Slower Than An RTX 3090 Ti
The NVIDIA GeForce RTX 4080 12 GB graphics card was supposed to launch next month at a price starting at $899 US. However, given the vast disparity in performance between the 16 GB and 12 GB models, NVIDIA decided to cancel the 12 GB variant while keeping the 16 GB variant on track for a 16th November launch.
Given how close this cancellation occurred to the launch, AIBs definitely received a batch of the Ada AD104 GPUs to prepare their custom models and we know from our own sources that these cards were definitely ready to ship out to retailers and distributors. Now despite the cancellation, gaming and synthetic benchmarks from these samples are starting to leak out.
There are no details such as the model number for this specific GeForce RTX 4080 12 GB graphics card but it does feature a max GPU clock as high as 2820 MHz and consumes 262W of power. The GPU was able to deliver a peak FP32 compute power of 42 TFLOPs in the AIDA64 GPGPU test. We cannot say for sure if these are reference or overclocked specs but the card wasn't going to launch in FE variants anyways and since we are talking custom model, it is likely to be a factory overclock.
FP32 Compute Horsepower Comparisons (Higher is Better)
As for other synthetic benchmarks, the NVIDIA GeForce RTX 4080 12 GB graphics card scored 5382 points in 3DMark Speed Way, 10794 points in 3DMark Time Spy Extreme, 13472 points in 3DMark Fire Strike Ultra, and 66.98 FPS in the DXR Feature Test.
3DMark Speed Way
3DMark Firestrike Ultra Graphics
3DMark Time Spy Extreme Graphics
There's also the Cyberpunk 2077 in-game benchmark which was tested at 4K with Ray Tracing enabled. The GPU scored an average of 67.52 FPS.
NVIDIA GeForce RTX 4080 12 GB 'Official' Specifications (Now Cancelled)
The NVIDIA GeForce RTX 4080 will also feature a 12 GB variant that is expected to utilize the AD104-400 GPU configuration with 7,680 cores or 60 SMs enabled which is the full-fat SKU. The GPU will be packing 48 MB of L2 cache and features a 192-bit bus interface so that's why we are getting up to 12 GB of GDDR6X capacities. The memory is said to be running at 21 Gbps speeds for a memory bandwidth of 504 GB/s. The card is said to be clocked at a 2610 MHz boost clock however the peak frequency will be higher.
- NVIDIA GeForce RTX 4080 12 GB "Official" TBP - 285W
- NVIDIA GeForce RTX 3080 10 GB "Official" TBP - 320W
For power, the TBP is now set to be rated at 285W which is a 35W decrease versus the RTX 3080 10 GB model. The card is expected to offer better performance than the RTX 3080 but given the specs cut over the 4080 16 GB, there will be a big gap between both models. The maximum BIOS TGP is said to be rated at 366W.
The huge performance gap between the RTX 4080 16 GB and RTX 4080 12 GB was made evident when NVIDIA published its own gaming benchmarks showing us both native & DLSS performance numbers. As per the benchmarks, the NVIDIA GeForce RTX 4080 12 GB graphics card was around 11% slower than the RTX 3090 Ti in gaming and you can also see in the synthetic benchmarks that the card is barely on par with the RTX 3090 Ti while costing $900 US. So the RTX 4080 12 GB had to be canceled and now we can expect a new card to take its place and offer better value to gamers.
NVIDIA GeForce RTX 40 Series Official Specs:
Graphics Card Name | NVIDIA GeForce RTX 4090 | NVIDIA GeForce RTX 4090 D | NVIDIA GeForce RTX 4080 | NVIDIA GeForce RTX 4070 Ti | NVIDIA GeForce RTX 4070 | NVIDIA GeForce RTX 4060 Ti | NVIDIA GeForce RTX 4060 |
---|---|---|---|---|---|---|---|
GPU Name | Ada Lovelace AD102-300 | Ada Lovelace AD102-250 | Ada Lovelace AD103-300 | Ada Lovelace AD104-400 | Ada Lovelace AD104-250 | Ada Lovelace AD106-350 | Ada Lovelace AD107-400 |
Process Node | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N |
Die Size | 608mm2 | 608mm2 | 378.6mm2 | 294.5mm2 | 294.5mm2 | 190.0mm2 | 146.0mm2 |
Transistors | 76 Billion | 76 Billion | 45.9 Billion | 35.8 Billion | 35.8 Billion | 22.9 Billion | TBD |
CUDA Cores | 16384 | 14592 | 9728 | 7680 | 5888 | 4352 | 3072 |
TMUs / ROPs | 512 / 176 | TBD | 320 / 112 | 240 / 80 | 184 / 64 | 136 / 48 | TBD |
Tensor / RT Cores | 512 / 128 | 456 / 128 | 304 / 76 | 240 / 60 | 184 / 46 | 136 / 34 | TBD |
L2 Cache | 72 MB | 72 MB | 64 MB | 48 MB | 36 MB | 32 MB | 24 MB |
Base Clock | 2230 MHz | 2280 MHz | 2210 MHz | 2310 MHz | 1920 MHz | 2310 MHz | 1830 MHz |
Boost Clock | 2520 MHz | 2520 MHz | 2510 MHz | 2610 MHz | 2475 MHz | 2535 MHz | 2460 MHz |
FP32 Compute | 83 TFLOPs | TBD | 49 TFLOPs | 40 TFLOPs | 29 TFLOPs | 22 TFLOPs | 15 TFLOPs |
RT TFLOPs | 191 TFLOPs | TBD | 113 TFLOPs | 82 TFLOPs | 67 TFLOPs | 51 TFLOPs | 35 TFLOPs |
Tensor-TOPs | 1321 TOPs | TBD | 780 TOPs | 641 TOPs | 466 TOPs | 353 TOPs | 242 TOPs |
Memory Capacity | 24 GB GDDR6X | 24 GB GDDR6X | 16 GB GDDR6X | 12 GB GDDR6X | 12 GB GDDR6X | 8-16 GB GDDR6 | 8 GB GDDR6 |
Memory Bus | 384-bit | 384-bit | 256-bit | 192-bit | 192-bit | 128-bit | 128-bit |
Memory Speed | 21.0 Gbps | 21.0 Gbps | 23.0 Gbps | 21.0 Gbps | 21.0 Gbps | 18.0 Gbps | 17.0 Gbps |
Bandwidth | 1008 GB/s | 1008 GB/s | 736 GB/s | 504 GB/s | 504 GB/s | 288 GB/s (554 GB/s Effective) | 272 GB/s (453 GB/s Effective) |
TBP | 450W | 425W | 320W | 285W | 200W | 160-165W | 115W |
Price (MSRP / FE) | $1599 US / 1949 EU | 12,999 RMB (China-Only) | $1199 US / 1469 EU | $799 US | $599 US | $399-$499 US | $299 US |
Price (Current) | $1599 US / 1859 EU | 12,999 RMB (China-Only) | $1199 US / 1399 EU | $799 US | $599 US | $399-$499 US | $299 US |
Launch (Availability) | 12th October 2022 | 28th December 2023 | 16th November 2022 | 5th January 2023 | 13th April 2023 | 24th May / 18th July 2023 | 29th June 2023 |
News Source: Olrak