GeForce 40 series

From Justapedia, unleashing the power of collective wisdom
Jump to navigation Jump to search

GeForce 40 series
NVIDIA RTX 4090 Founders Edition - Verpackung (ZMASLO).png
An RTX 4090 Founders Edition in its packaging
Release dateOctober 12, 2022; 11 months ago (2022-10-12)
CodenameAD10x
ArchitectureAda Lovelace
ModelsGeForce RTX series
Transistors
  • 35.8B (AD104)
  • 45.9B (AD103)
  • 76.3B (AD102)
Fabrication processTSMC 4N[1]
Cards
High-end
  • GeForce RTX 4080
Enthusiast
  • GeForce RTX 4090
API support
Direct3DDirect3D 12.0 Ultimate (feature level 12_2)
OpenCLOpenCL 3.0
OpenGLOpenGL 4.6
VulkanVulkan 1.3
History
PredecessorGeForce 30 series
An aftermarket AIB variant of the RTX 4090 (Asus ROG Strix) on the bottom, next to an RTX 3090 (Gigabyte Gaming OC) for size comparison

The GeForce 40 series is a family of graphics processing units developed by Nvidia, succeeding the GeForce 30 series. The series was announced on September 20, 2022, at the GPU Technology Conference (GTC) 2022 event; the RTX 4090 was released on October 12, 2022,[1] and the RTX 4080 will be released the following month. The cards are based on the Ada Lovelace architecture and feature hardware-accelerated raytracing (RTX) with Nvidia's third-generation RT cores and fourth-generation Tensor Cores.

Details

Architectural highlights of the Ada Lovelace architecture include the following:[2]

  • CUDA Compute Capability 8.9[3]
  • TSMC 4N process (custom designed for NVIDIA)[1] – not to be confused with N4
  • Fourth-generation Tensor Cores with FP8, FP16, bfloat16, TensorFloat-32 (TF32) and sparsity acceleration
  • Third-generation Ray Tracing Cores, along with concurrent ray tracing, shading and compute
  • Shader Execution Reordering - needs to be enabled by the developer[4]
  • NVENC with 8K 10-bit 60FPS AV1 fixed function hardware encoding[5][6]
  • A new generation of Optical Flow Accelerator to aid DLSS 3.0 intermediate AI-based frame generation[7]
  • No NVLink support[8]

Products

  • Double-precision (FP64) performance of the Ada Lovelace chips are 1/64 of single-precision (FP32) performance.
  • All the cards feature GDDR6X video memory.
Model Launch Launch
MSRP
(USD) 		Code
name(s) 		Transistors (billion) Die size (mm2) Core
config[a] 		SM
count[b] 		L2
cache
(MB) 		Clock speeds[c] Fillrate[d][e] Memory Processing power (TFLOPS) TDP
(watts)
Core
clock
(MHz) 		Memory
(GT/s) 		Pixel
(Gpx/s) 		Texture
(Gtex/s) 		Size
(GB) 		Bandwidth
(GB/s) 		Bus
width
(bit) 		Half
precision
(boost) 		Single
precision
(boost) 		Double
precision
(boost) 		Tensor
compute
[sparse]

GeForce RTX
4080 (12 GB)[9][10] 		Cancelled $899 AD104-400 35.8 294.5 7680
240:80:60:240 		60 48 2310
(2610) 		21.0 184.8
(208.8) 		554.4
(626.4) 		12 504 192 35.482
(40.090) 		35.482
(40.090) 		0.554
(0.626) 		160.4
[320.8] 		285
GeForce RTX
4080[11] 		Nov 16, 2022 $1199 AD103-300 45.9 378.6 9728
304:112:76:304 		76 64 2210
(2505) 		22.4 247.5
(280.6) 		671.8
(761.5) 		16 716.8 256 42.998
(48.737) 		42.998
(48.737) 		0.672
(0.762) 		194.9
[389.8] 		320
GeForce RTX
4090[10] 		Oct 12, 2022 $1599 AD102-300 76.3 608.5 16384
512:176:128:512 		128 72 2230
(2520) 		21.0 392.5
(443.5) 		1141.8
(1290.2) 		24 1008 384 73.073
(82.575) 		73.073
(82.575) 		1.142
(1.290) 		330.3
[660.6] 		450
  1. ^ Shader Processors : Texture mapping units : Render output units : Ray tracing cores : Tensor Cores
  2. ^ The number of Streaming multi-processors on the GPU.
  3. ^ Core boost values (if available) are stated below the base value inside brackets.
  4. ^ Pixel fillrate is calculated as the lowest of three numbers: number of ROPs multiplied by the base core clock speed, number of rasterizers multiplied by the number of fragments they can generate per rasterizer multiplied by the base core clock speed, and the number of streaming multiprocessors multiplied by the number of fragments per clock that they can output multiplied by the base clock rate.
  5. ^ Texture fillrate is calculated as the number of texture mapping units (TMUs) multiplied by the base (or boost) core clock speed.

RTX 4080 12 GB controversy

When the 12 GB RTX 4080 was announced, numerous outlets, prominent YouTubers, reviewers, and the community criticized Nvidia for calling it an RTX 4080 instead of RTX 4070, given previous Nvidia GPU generations and the large gap in specifications and performance compared to the 16 GB card.[12][13][14][15] Unlike other cases of the same-named product with differing memory configurations that were otherwise very close in performance, the 12GB RTX 4080 uses a completely different chip and configuration: among other differences, the 12 GB 4080, using the AD104 chip, was to feature 27% fewer CUDA cores, along with a cut down 192-bit memory bus, typically used for the xx60 cards[citation needed]. This made the card up to 30% slower than the 16 GB RTX 4080 in raw performance while being priced significantly higher than previous xx70 cards ($900 vs. $500 for the RTX 3070, approximately 80% more expensive).

On October 14, 2022, Nvidia announced that due to the confusion caused by the naming scheme, it would be "unlaunching"—i.e. pausing the launch of—the 12 GB RTX 4080, with the 16 GB RTX 4080's launch remaining unaffected. Future marketing plans for the card have not been revealed.[16][17]

See also

References

  1. ^ a b c "NVIDIA Delivers Quantum Leap in Performance, Introduces New Era of Neural Rendering With GeForce RTX 40 Series". NVIDIA Newsroom. September 20, 2022. Retrieved October 8, 2022.
  2. ^ "NVIDIA Ada Lovelace Architecture". NVIDIA.
  3. ^ "I.7. Compute Capability 9.x". docs.nvidia.com.
  4. ^ Palumbo, Alessio (September 23, 2022). "NVIDIA Ada Lovelace Follow-Up Q&A - DLSS 3, SER, OMM, DMM and More". Wccftech. Retrieved September 25, 2022.
  5. ^ "Creativity At The Speed of Light: GeForce RTX 40 Series Graphics Cards Unleash Up To 2X Performance in 3D Rendering, AI, and Video Exports For Gamers and Creators". NVIDIA.
  6. ^ "Nvidia Video Codec SDK". August 23, 2013.
  7. ^ Chiappetta, Marco (September 22, 2022). "NVIDIA GeForce RTX 40 Architecture Overview: Ada's Special Sauce Unveiled". HotHardware. Retrieved September 25, 2022.
  8. ^ "Jensen Confirms: NVLink Support in Ada Lovelace is Gone". TechPowerUp. September 21, 2022.
  9. ^ "NVIDIA GeForce RTX 4080 Graphics Cards for Gaming". NVIDIA. Archived from the original on September 28, 2022. Retrieved October 30, 2022.
  10. ^ a b "NVIDIA Ada GPU Architecture" (PDF). Nvidia. Retrieved October 1, 2022.
  11. ^ "NVIDIA GeForce RTX 4080 Graphics Cards for Gaming". Nvidia. Retrieved October 14, 2022.
  12. ^ Laird, Jeremy (September 22, 2022). "We've run the numbers and Nvidia's RTX 4080 cards don't add up". PC Gamer. Retrieved September 27, 2022.
  13. ^ "Why the RTX 4080 12GB feels a lot like a rebranded RTX 4070". Digital Trends. September 21, 2022. Retrieved September 27, 2022.
  14. ^ Walton, Jarred (September 23, 2022). "Why Nvidia's RTX 4080, 4090 Cost so Damn Much". Tom's Hardware. Retrieved September 27, 2022.
  15. ^ Guyton, Christian (September 22, 2022). "Buyer beware: the 12GB RTX 4080 is hiding a dirty little secret". TechRadar. Retrieved September 27, 2022.
  16. ^ "Unlaunching The 12 GB 4080". NVIDIA. Retrieved October 14, 2022.
  17. ^ Warren, Tom (October 14, 2022). "Nvidia says it's 'unlaunching' the 12GB RTX 4080 after backlash". The Verge. Retrieved October 14, 2022.

External links

Retrieved from "https://justapedia.org/index.php?title=GeForce_40_series&oldid=1119682501"