Nvidia RTX 5000 vs RTX 4000: Graphics Cards Technical Data Compared 74 comments
Image: Nvidia
With the GeForce RTX 5090, RTX 5080, RTX 5070 Ti and RTX 5070, Nvidia has introduced four new consumer graphics cards. But it is not only the comparison of the new models with each other that is interesting, but also the direct comparison with its predecessors of the RTX 4000 generation.
In individual tables, Techastuce compares the GeForce RTX 5090 with the GeForce RTX 4090, the RTX 5080 with the RTX 4080 Super and the RTX 4080, the RTX 5070 Ti with the RTX 4070 Ti Super and the RTX 4070 Ti as well as the RTX 5070 with the RTX 4070 Super and RTX 4070.
RTX 5090 vs. RTX 4090
The flagships begin with a comparison of the GeForce RTX 5090 with the RTX 4090. The RTX 5090 can boast more shaders, an increased VRAM from 24 to 32 GB, the GDDR7 memory type plus fast and a significantly wider memory interface. As for clock speeds, Blackwell is a bit slower than Ada Lovelace, but the RTX 5090’s 575-watt TDP compares to the RTX 4090’s significantly lower 450-watt TDP.
However, when it comes to AI TOPS, the new flagship beats the old one by more than a factor of 2.5. However, it is still unclear to what extent the values are comparable and at what level of precision they were set by Nvidia. Lower precision initially makes performance appear higher without specifying it for comparison. One possibility would be that the new tensor cores could operate with half the precision compared to the RTX 4000, for example with INT2 instead of INT4. If you then take into account the difference between the CUs and the slightly lower clock speed, nothing could have changed with the same precision. Nvidia itself compared FP4 on RTX 5000 and FP8 on RTX 4000 in AI benchmarks and confirmed an expected increase in performance by a factor of 2.
GeForce RTX 5090 vs. RTX 4090 model GeForce RTX 5090 Architecture GeForce RTX 4090 Blackwell Ada Lovelace GPU GB202 AD102 Active CUs (chip) 170 (192) 128 (144) active shaders 21,760 16,384 Boost/base clock speeds 2,410/2 010 MHz 2520/2230 MHz FP32 ALU 104.9 TFLOPS 82.6 TFLOPS Memory 32 GB GDDR7 24 GB GDDR6X Memory interface 512-bit 384-bit Memory throughput 28 Gbps 21 Gbps Bandwidth 1792 GB/s 1008 GB/s PCI Express 5.0 x16 4.0 ×16 TDP 575 W 450 W Founders Edition Yes 2 slots 3 slots Power connection 1x 12V-2×6 (600 watts) 1x 12VHPWR (600 watts) I/O 3× DisplayPort 2.1b
1 × HDMI 2.1 3 × DisplayPort 1.4a
1× HDMI 2.1 DLSS DLSS 4 DLSS 3.5 AI TOPS 3,352 1,321 UVP (incl. MwSt.) 2,329 euros 1,949 euros
GeForce RTX 5080 vs. RTX 4080 Super vs. RTX 4080
Comparing the GeForce RTX 5080 with the two previous generation models, the RTX 4080 Super and RTX 4080, we notice that the new model is superior or at least equal in all aspects. In the RTX 5080, the clock speeds of the GB203 GPU are also significantly higher than those of the B202 of the RTX 5090. Although the TDP increases from 320 to 360 watts, the TDP increase percentage of 12.5 percent is significantly lower than that of almost 28 percent for the RTX 5090.
GeForce RTX 5080 vs. RTX 4080 Super vs. RTX 4080 model GeForce RTX 5080 GeForce RTX 4080 Super Architecture GeForce RTX 4080 Blackwell Ada Lovelace GPU GB203* AD103 Active CUs 84 (84) 80 (84) 76 (84) active shaders 10,752 10 240 9728 Clock Speeds Boost/Base 2620/2300 MHz 2550/2295 MHz 2510/2210 MHz FP32 ALU 56.3 TFLOPS 52.2 TFLOPS 47 TFLOPS Memory 16 GB GDDR7 16 GB Memory Interface GDDR6X 256 bits Memory Throughput 30 Gbps 23 Gbps 22.4 Gbps Bandwidth 960 GB/s 736 GB/s 717 GB/s PCI Express 5.0 x16 4.0 ×16 TDP 360 W 320 W Founders Edition Yes 2 slots ( 304 mm) 3 locations Electrical connection 1x 12V -2×6 (min. 450 watts) 1x 12VHPWR I/O 3× DisplayPort 2.1b
1 × HDMI 2.1 3 × DisplayPort 1.4a
1× HDMI 2.1 DLSS DLSS 4 DLSS 3.5 AI TOPS 1 801 836 780 Recommended price (VAT included) 1,169 euros 1,109 euros 1,329 euros * not confirmed
GeForce RTX 5070 Ti vs. RTX 4070 Ti Super vs. RTX 4070 Ti
For the Ti models of the x70 series, which have to do without Founders Edition for RTX 4000 and now also for RTX 5000, the specifications also speak clearly about the new model, which also has a slightly higher TDP than its predecessors. Unlike comparing the RTX 4070 Ti (Super) to the RTX 4080, Nvidia is taking a slightly different approach to clock speeds with the new generation. If the RTX 4070 Ti (Super) has a higher frequency than the RTX 4080 (Super), the RTX 5070 Ti now has a lower frequency than the RTX 5080.
GeForce RTX 5070 Ti vs. RTX 4070 Ti Super vs. RTX 4070 Ti Model GeForce RTX 5070 Ti GeForce RTX 4070 Ti Super GeForce RTX 4070 Ti Architecture Blackwell Ada Lovelace GPU GB203* AD103 AD104 Active CUs (chip) 70 (84) 66 (84 ) 60 (60) active shaders 8960 8448 7680 Clock Speeds Boost/Base 2450/2300 MHz 2610/2340 MHz 2610/2310 MHz FP32 ALU 43.9 TFLOPS 44.1 TFLOPS 40.1 TFLOPS Memory 16 GB GDDR7 16 GB GDDR6X 12 GB Memory interface GDDR6X 256-bit 192-bit Memory throughput 28 Gbps 21 Gbps Bandwidth 896 GB/s 672 GB/s 504 GB/s PCI Express 5.0 x16 4.0 ×16 TDP 300 W 285 Watt Founders Edition No – Electrical connection 1x 12V-2×6 (min .300 watts) 1x 12VHPWR I/O? 3 × DisplayPort 1.4a
1× HDMI 2.1 DLSS DLSS 4 DLSS 3.5 AI TOPS 1 406 706 641 Recommended price (VAT included) 879 euros 889 euros * not confirmed
GeForce RTX 5070 vs. RTX 4070 Super vs. RTX 4070
The duel between the RTX 5070 and the RTX 4070 Super will be interesting if no new software functions like DLSS 4 are used. Because then the new graphics card must assert itself against its super predecessor with fewer shaders but higher memory bandwidth. Compared to its direct predecessor, the GeForce RTX 4070, the comparison is clear again.
GeForce RTX 5070 vs RTX 4070 Super vs RTX 4070 model GeForce RTX 5070 GeForce RTX 4070 Super Architecture GeForce RTX 4070 Blackwell Ada Lovelace GPU GB205* AD104 Active CUs (chip) 50 (50) 56 (60) 46 (60) active shaders 6,144 7 168 5888 Clock Speeds Boost/Base 2510/2160 MHz 2475/1980 MHz 2475/1920 MHz FP32 ALU 30.8 TFLOPS 35.5 TFLOPS 29.1 TFLOPS Memory 12 GB GDDR7 12 GB Memory Interface GDDR6X Memory throughput 192-bit 28 Gbps 21 Gbps Bandwidth 672 GB/s 504 GB/s PCI Express 5.0 x16 4.0 ×16 TDP 250 W 220 W 200 W Founders Edition Yes 2-slot power connection 1x 12V-2×6 (min. 300 watts) 1x 12VHPWR 3× I/O DisplayPort 2.1b
1 × HDMI 2.1 3 × DisplayPort 1.4a
1× HDMI 2.1 DLSS DLSS 4 DLSS 3.5 AI TOPS 988 568 466 Recommended price (VAT included) 649 euros 659 euros * not confirmed Themes: CES 2025 GeForce GeForce RTX 40 GeForce RTX 50 graphics cards Nvidia Nvidia Ada Lovelace Nvidia Blackwell
An engineer by training, Alexandre shares his knowledge on GPU performance for gaming and creation.
