As was foretold, Nvidia has revealed its new RTX 50-series desktop graphics playing cards at its CES 2025 keynote, which occurred within the early hours UK time. The present lineup contains 4 playing cards: a $1999/£1939 RTX 5090, a $999/£979 RTX 5080, a $749/£729 RTX 5070 Ti and a $549/£539 RTX 5070. Every “Blackwell” graphics card comes outfitted with GDDR7 reminiscence, some spectacular frame-rate claims and DLSS 4 multi body era. The 2 top-end playing cards are arriving on thirtieth January, whereas the upper-mid-range choices are scheduled for February.
The “relative efficiency” claims for a few of these graphics playing cards are wild, so let’s begin with an thought of anticipated frame-rates for every card and a fast explainer of DLSS 4 multi body era earlier than we cowl off the architectural enhancements.
The $1999 RTX 5090 is unsurprisingly positioned as the last word graphics card, a prosumer mannequin with 32GB of GDDR7. It is based mostly across the GB202 GPU with 92 billion transistors, versus 76 billion on the 4090, and contains 21760 CUDA cores. Nvidia’s slides promise double the frame-rates of the RTX 4090, or a snug 4K 240Hz with full RT and maxed settings in video games that help the multi body era (MFG) function.
To clarify earlier than we go additional, MFG provides in as much as three generated frames for every historically rendered body. This leads to increased frame-rates and due to this fact visible fluidity, however not essentially increased efficiency as we would usually label it, as latency is reliant on the “base” frame-rate relatively than the ultimate “output” frame-rate. The idea is much like the unique body era (FG) function on RTX 40-series playing cards, and as such depends on developer integration and Nvidia’s Reflex 2 latency mitigation tech to work properly.
Surprisingly, whereas the RTX 5090 has a rated complete graphics energy of 575W – proper up in opposition to the PCIe 5.0 600W restrict and 125W increased than the RTX 4090 – the Founders Version design is definitely slimmer than the RTX 4090 FE. It is a two-slot card that will truly slot in small kind issue PC circumstances (with requisite 1000W PSUs), with two flow-through followers and the PCB situated within the centre. It’s going to be fascinating to see how the brand new design performs, provided that early 5090 prototype coolers have been reportedly quad-slot behemoths!
The $999 RTX 5080 is designed to double the frame-rates of the RTX 4080, with Nvidia promising past 2x leads to video games like Cyberpunk 2077, Alan Wake 2 and Black Fantasy: Wukong – all heavy RT titles. The 5080 is predicated on the smaller GB203 die and comes with 16GB of GDDR7, 10752 CUDA cores and a rated energy of 360W – with an 850W PSU really useful. The pricing matches that of the RTX 4080 Tremendous, so it’s going to be fascinating to see how these fashions evaluate by way of apples-to-apples efficiency when FG or MFG is not used.
Nvidia
RTX 5090
RTX 5080
RTX 5070 Ti
RTX 5070
GPU
GB202
GB203
GB203
GB205
Cores
21760
10752
8960
6144
Enhance clock
2.41GHz
2.62GHz
2.45GHz
2.51GHz
Tensor core TOPS
3352
1801
1406
988
RT core TFLOPS
318
171
133
94
Reminiscence
32 GB GDDR7
16 GB GDDR7
16 GB GDDR7
12 GB GDDR7
Reminiscence bus width
512-bit
256-bit
256-bit
192-bit
Reminiscence bandwidth
1792GB/s
960GB/s
896GB/s
672GB/s
Complete Graphics Energy
575W
360W
300W
250W
PSU suggestion
1000W
850W
750W
650W
Energy connector
600W PCIe 5.0 (4x 8-pin)
450W PCIe 5.0 (3x 8-pin)
300W PCIe 5.0 (2x 8-pin)
300W PCIe 5.0 (2x 8-pin)
Worth
$1999/£1939
$999/£979
$749/£729
$549/£539
Launch Date
January thirtieth
January thirtieth
February
February
The 2 February playing cards, the RTX 5070 Ti 16GB and 5070 12GB, hit on the $749 and $549 value factors. The 5070 Ti makes use of a cut-back model of the GB203 GPU with 8960 CUDA cores, whereas the 5070 makes use of the GB205 with 6144 cores. It is disappointing to see the 12GB of GDDR7 on the $549 RTX 5070 card within the 12 months of our lord 2025, however Nvidia would have wanted to equip the 5070 with a 256-bit bus (or use a blended VRAM allocation) to hit 16GB so the transfer isn’t sudden.
By way of efficiency, it is the identical theme: the 5070 Ti is supposed to double the frame-rate of the 4070 Ti, whereas the 5070 does the identical for the 4070. Nevertheless, Nvidia additionally identified that because of this the 5070 ought to due to this fact provide the identical frame-rates because the RTX 4090, which might be an enormous coup at a couple of third of the worth! Provided that this depends an uneven MFG vs FG comparability, we’ll want to attend for correct third-party testing to see if that declare is actually borne out and what the differential seems like in non-FG and certainly non-DLSS/non-RT titles.
In addition to confirming its desktop lineup, Nvidia additionally introduced RTX 50-series laptop computer GPUs. These will arrive from March and embody smaller reminiscence allocations. These embody the RTX 5090 24GB, RTX 5080 16GB, RTX 5070 Ti 12GB and RTX 5070 8GB.
Handle cookie settings
Now that we have got the line-up lined, it is price shortly mentioning among the architectural adjustments that contribute to each quicker general efficiency and the important thing good points by way of RT and AI efficiency. These enhancements are derived from beefier SMs with extra throughput and tighter Tensor core integration, enhancements to shader execution reordering and acceleration for FP4 compute. There’s additionally enhanced {hardware} flip metering wanted for multi body gen, new RT cores with double the ray triangle intersection fee and higher compression. The playing cards additionally help the PCIe 5.0 interface and include DisplayPort 2.1b UHBR20 ports, matching AMD’s prior-gen playing cards.
When you’re interested by studying extra concerning the Blackwell structure, Nvidia’s weblog has extra element on these architectural particulars, and we anticipate the standard tech press like Players Nexus or TechSpot will provide extra detailed breakdowns of those adjustments as soon as they’ve had an opportunity to digest the knowledge.
By way of our output, we anticipate to publish extra details about DLSS 4 multi body era within the close to future, so look out for that, and naturally we’re eager to place the brand new GPUs by way of our refined check suite as quickly as doable.
