NVIDIA’s GeForce RTX 5080 “Blackwell” GPU is an enthusiast option and replaces the RTX 4080.
NVIDIA GeForce RTX 5080 GPU: Blackwell is designed for serious gamers.
NVIDIA GeForce RTX 5080 is the company’s upcoming GPU based on the Blackwell architecture. The RTX 5080 is the second fastest GPU in the RTX 50 lineup after the RTX 5090 and is said to be the successor to the current generation RTX 4080 and RTX 4080 SUPER GPUs.
NVIDIA’s upcoming GeForce RTX 50 series is based on the Blackwell architecture, and the RTX 5080 will be the second fastest in the series after the flagship RTX 5090 GPU. NVIDIA’s Blackwell architecture is used in both data centers and gaming GPUs. Blackwell GPUs utilize a new custom 5nm process node, specifically N5B.
The NVIDIA GeForce RTX 5080 GPU is expected to take over the Ada-based GeForce RTX 4080 GPU by offering a noticeable performance advantage in gaming and general graphics workloads. NVIDIA has continued with 80-class cards for generations, and they were the fastest GPUs in the entire lineup until the introduction of the Turing-based RTX 30-series GPUs. Since then, 90-class GPUs have become the fastest in the world. Lineup.
GeForce RTX 5080 – Second fastest in the RTX 50 lineup
The NVIDIA GeForce RTX 5080 is expected to feature the GB203-300-A1 die, which is the successor to the RTX 4080’s AD103-300-A1. It is expected to feature a PG144/147-SKU45 PCB and carry 84 SMs.
There is currently no detailed information about the die other than the basic GPU configuration, but there are some leaked specs for the GPU. We’ll update this section as soon as we have information about the GPU die, but for now we expect NVIDIA to follow up with the same tier GPU die used in previous generations of 80-class cards.
NVIDIA GeForce RTX 5080 specifications
The NVIDIA GeForce RTX 5080 will reportedly feature 10,752 Cuda cores and 84 SMs, an increase of about 10.5% compared to the RTX 4080’s 9,728 Cuda cores and 76 SMs. We expect NVIDIA to maintain the same memory size. It’s an RTX 5080, but it comes with the new GDDR7 memory type compared to GDDR6X. RTX4080.
The NVIDIA GeForce RTX 5080 GPU is expected to feature 16 GB of VRAM on a 256-bit bus and deliver the fastest GDDR7 memory speeds of up to 32 Gbps, for total memory bandwidth of approximately 1 TB/s. Masu. The 1024 GB/s bandwidth is better than the RTX 4090 GPU, which offers 1007 GB/s bandwidth.
Evolution of GDDR graphics memory:
Graphics Memory GDDR7GDDR6XGDDR6GDDR5X Workload Gaming / AIGaming / AIGaming / AIGaming Platform (Example)GeForce RTX 5090?GeForce RTX 4090GeForce RTX 2080 TiGeForce GTX 1080 Ti Die Capacity (Gb)16-648-328-328-16 Number Arrangement 12?121212 Gb/s/Pin 32-42.519-2414-1611.4 GB/s/Arrangement 128-14476-9656-6445 GB/s/System 1536-1728 912-1152672-768547 Configuration (Example) 384 IO (12pcs x 32 IO) 384 IO (12 x 32 IO package)384 IO (12 x 32 IO package)384 IO (12 x 32 IO package) Typical system frame buffer 24 GB (16 Gb)
36 GB (24 Gb)24 GB12GB12GB Module Package266 (BGA)180 (BGA)180 (BGA)190 (BGA) Average Device Power (pJ/bit)TBD7.257.5 8.0 Standard IO Channel PCB (P2P SM)PCB (P2P SM )PCB (P2P SM)PCB (P2P SM)
The memory speeds will be even faster than what the RTX 5090 brings, with about 43% more bandwidth than the RTX 4080 has. There were also rumors recently that the 5080 would come in a 24 GB version, or another edition offering 3 GB per module. However, the launch version will only have 16 GB VRAM.
That said, NVIDIA GeForce RTX 5080 brings higher TBP than RTX 4080 at 400W. This is 80W more than the RTX 4080 (320W). TBP is simply the peak power available to the board and is not compared to the actual power consumption which is lower than the actual wattage rating when looking at the RTX 40 GPU. The RTX 5080 features a single 12V-2×6 connector for power requirements as the connector can deliver up to 600W.
SpecificationsGeForce RTX 5080GeForce RTX 4080 ArchitectureBlackwellAda Lovelace Process Node4NP(TSMC’s 5nm)4N(TSMC’s 5nm) GPUGB203-400-A1AD103-300-A1 Cuda Core 107529728 SMTBD76 Tensor Core TBD304 RT Core TBD76 Clock Speed (Base/Boost) TBD2205/2505 MHz VRAM Size and Type16 GB/24 GB GDDR716 GB GDDR6X Memory Speed32 Gbps22.4 GbpsMemory Bus Width256 bits256 bitsMemory Bandwidth1024 GB/s716.8 GB/sTBP400W320W Power connector 1x 12V-2x61x 12VHPWR
NVIDIA’s GeForce RTX 5080 – Expected Performance
The NVIDIA GeForce RTX 5080 replaces the RTX 4080, and recent rumors say it will offer 10% better performance than the RTX 4090. According to the specs, the RTX 5080 may be better than the RTX 4090 in ray tracing alone, and the RTX 4090 appears to be better. It has an advantage over the 5080.
Nevertheless, some theories may prove both scenarios and some concrete data will be needed to determine whether RTX 5080 is better than RTX 4090 . Compared to the RTX 4080, the spec upgrades are almost guaranteed to make the 5080 significantly faster. This means most titles will display 60+ fps at 4K resolution. You should also keep in mind that the RTX 4080 is 20-25% faster than the RTX 3090, and you can expect similar, if not more, gains from the RTX 5080.
The NVIDIA GeForce RTX 4080 and 4080 SUPER offer solid performance in modern titles for both rasterization and ray tracing, and the RTX 5080 brings higher memory bandwidth, making it even better at high resolutions, especially 4K. We can expect them to continue to maintain their dominance.
NVIDIA GeForce RTX 5080 price
There are no reliable leaks regarding the price of the RTX 5080 GPU. Similarly, there is no reliable pricing information for other RTX 50-series GPUs, including the RTX 5090 and RTX 5070. It’s not always wise to talk about prices before launch, as prices can change very slowly and sometimes after launch (but that’s very rare).
The RTX 4080 SUPER launched for $1,000, a slight price reduction compared to the original model, which retailed for $1,199. With that in mind, the price will probably be around $1000, but it would be great if you could get it for less since it gives you the performance of the 4090 class graphics cards currently on sale. About 2000 USD at half price.
NVIDIA GeForce GPU segment/tier pricing
Graphics Segment20242023-20242022-20232021-20222020-20212019-20202018-20192017-2018 Titan TierGeForce RTX 5090GeForce RTX 4090GeForce RTX 4090GeForce RTX 3090Ti
GeForce RTX 3090GeForce RTX 3090Titan RTX (Turing)Titan V (Volta)Titan Xp (Pascal) Price TBD$1599 US$1599 US$1999 US
$1499 US$1499 US$2499 US$2999 US$1199 US Super Enthusiast TierGeForce RTX 5080GeForce RTX 4080 SUPERGeForce RTX 4080GeForce RTX 3080 TiGeForce RTX 3080 TiGeForce RTX 2080 TiGeForce RTX 2080 TiGeForce GTX 1080 Ti Price TBA $999 US$1199 US$1199 US$1199 US$999 US$999 US$699 US Enthusiasts TierGeForce RTX 5070 Ti?GeForce RTX 4070 Ti SUPERGeForce RTX 4070 TiGeForce RTX 3080 12 GBGeForce RTX 3080 10 GBGeForce RTX 2080 SUPERGeForce RTX 2080GeForce GTX 1080 Price TBD$799 US$799 US$799 US$699 US$699 US$699 US$549 US High-end TierGeForce RTX 5070GeForce RTX 4070 Super
GeForce RTX 4070GeForce RTX 4070
GeForce RTX 4060 Ti 16GBGeForce RTX 3070 Ti
GeForce RTX 3070GeForce RTX 3070Ti
GeForce RTX 3070GeForce RTX 2070 SUPERGeForce RTX 2070GeForce GTX 1070 Price TBD$599
$549 $599 US
$499 $599
$499$599
$499$499 US$499 US$379 US Mainstream TierGeForce RTX 5060 TiGeForce RTX 4060 Ti
GeForce RTX 4060GeForce RTX 4060Ti
GeForce RTX 4060GeForce RTX 3060 Ti
GeForce RTX 3060 12GBGeForce RTX 3060 Ti
GeForce RTX 3060 12GBGeForce RTX 2060 SUPER
GeForce RTX 2060
GeForce GTX 1660Ti
GeForce GTX 1660 Super
GeForce GTX 1660GeForce GTX 1060GeForce GTX 1060 Price TBD$449
$299 $399 US
$299 $399
$329 $399
$329 $399
$349
$279
$229
$219 US$249 US$249 US Entry TierGeForce RTX 5060RTX 3050 8 GB
RTX 3050 6 GBRTX 3050RTX 3050GTX 1650 Super
GTX1650GTX1650 super
GTX1650GTX1050Ti
GTX1050GTX1050Ti
GTX 1050 price TBD$229
$179$249 US$249 US$159 US
$149 $159
$149 $139
$109 $139
$109
NVIDIA’s GeForce RTX 5080 – Release Date and Availability
NVIDIA finalized the RTX 5080 design along with the RTX 5090 in September 2024, but several reliable reports suggest that the RTX 5080 will not launch this year. NVIDIA is expected to announce the RTX 5080 along with the RTX 5090 and RTX 5070 in early 2025, likely at CES in January.
