With The Release Time Approaching, The Next Generation Of AMD And NVIDIA Graphics Cards Not Only Have More And More Specifications, But Also Have Clearer Performance. The Performance Of The Tfrlo31 , Which Is Expected To Be 32 Times Higher Than That Of The Tfrlo31 , Is Now Almost 28 Times Higher Than That Of The Tfrlo31 , Which Is Expected To Be The Next Generation Navix*
Greymon55 Disclosed Earlier, amd Is Expected To Adjust The Specification And Performance Of Navi 31, But I Didn't Expect Such A Big Change.
If There Are 60 WGPs (working Group Processors) And 15360 Stream Processors, In Order To Achieve The Computing Performance Of 92tflops Fp32, The Core Frequency Must Be As High As 2995mhz . Of Course, There Is A High Probability That A Series Of Different 32-bit Instruction Sets Will Be Introduced To Obtain Higher Floating-point Performance.
According To @ Kepler_ L2, Amdgpu Driver Information Shows That Rdna3 Architecture Will Still Retain The Traditional Cu Computing Unit Architecture , And Will Not Cancel It.
[](//img1.mydrivers.com/img/20220502/5976220268c8447cbdb7492d668ac250.png)
For NVIDIA, Kopite7kimi Claimed That Ad102 Core Is Expected To Provide Floating-point Performance Of 100tflops, Which Is Not Only 2.5 Times Higher Than RTX 3090 Ti And 2.8 Times Higher Than RTX 3090, But Also Slightly Better Than Its Opponent Again**
If There Are 18432 CUDA Cores, The Core Frequency Is About 2.7ghz , And According To Convention, The RTX 4090 Will Not Be Full Of Blood, And The Frequency Will Be Higher.
Of Course, Floating-point Performance And Game Performance Cannot Be Equated. In Fact, It Remains To Be Seen Which Is Stronger Or Worse, But It Can Also Reflect A Considerable Part Of GPU Performance. It Seems That This Is Another Very Fierce Competition.
[](//img1.mydrivers.com/img/20220502/4a476d0e1b2247ee8a3a24d2da1eb142.png)