The GeForce RTX 3090 Ti is the first and single gaming geforce based on a full active GA102 processor. Core with the identifier “350” not only wait with all 84 shader multiprocessors and thus 10.752 FP32 alus, but also with high stroke. Meanwhile, the original GeForce RTX 3090 must be content with a slightly slimmedized GA102, but can already fall back on the full 384-bit interface. The latter also offers the RTX 3090 Ti, of course, including added value: For the first time, GDDR6x memory is used with an effective data rate of 21 gigatransfers / s. Thus, the RTX 3090 Ti as the first GeForce takes the 1 TByte / S hurdle. The no less impressive 40 TFLOPS on FP32 computing power is achieved not only by the small units of the RTX 3090, but above all by a clock boost: 1,860 MHz are officially advised, with many years of experience teaching that for the vast majority of scenarios Conservatively appreciated and at best in high-load scenarios will be undercut. Consequently, RTX 3090 Ti has 12.4 percent more computing power and 7.7 percent higher memory throughput than the RTX 3090.
To ensure deployment, NVIDIA packs all this in a 450 watt package. This does not just consistently powerful coolers, but also a powerful power supply. In fact, RTX 3090 Ti represents a kind of test run for the immediate future with all these superlatives. The Board partners rehearse the increased signal quality challenges, because the high currents and rapid storage require a skilful handling of physics. In parallel, a new standard is introduced during the power supply, which is not a solo on by NVIDIA this time, but follows an industrial standard. The accompanying habituation of ever higher performance values can be considered as a preliminary game for the next GPU generation.
To home page