الأربعاء، 19 أكتوبر 2022

Nvidia Is Reportedly Ditching Ada Titan, RTX 4090 Ti Still on Track

(Photo: RTX Turing Titan, back when SLI was still a thing)
Now that the RTX 4090 has cemented its standing atop the GPU field (for the time being), we turn our attention to its inevitable successor. Previously it was reported that Nvidia had a Titan-class GPU waiting in the wings. This was somewhat of a surprise, as the company never offered a Titan with its previous Ampere architecture. After all, that’s kind of what the RTX 3090s were. The same can be said for the RTX 4090, as no gamer needs 24GB of GDDR6 memory. Still, Nvidia was said to be planning a behemoth GPU to simultaneously destroy AMD’s dreams and gamers’ bank accounts. Now it’s being tipped that those plans have gone up in smoke, literally. Instead, Nvidia will be focusing on a true Ada Lovelace flagship GPU: the RTX 4090 Ti. Our PCIe slot hurts just thinking about its potential heft.

The newest rumors are courtesy of famed GPU informant kopite7kimi. This particular leaker has been pretty accurate in the past. This time, they’re saying the Titan GPU is a no-go. Interestingly, this same leaker was the one who previously reported on “the beast.” This was reportedly an 800W GPU with an outrageous 48GB of GDDR6X memory. If you recall, previous Titans were differentiated from GeForce cards by their excessive memory allocation.

That became a moot product with the RTX 3090’s 24GB of VRAM. It was previously reported this new Titan GPU was being tested by Nvidia, but it was experiencing issues due to its audacious power consumption. It was rumored to have been cut down to a mere 600 to 700W, but even that was too much for some environments. Moore’s Law Is Dead reported the prototype card was melting power supplies and tripping breakers due to its power consumption.

The RTX Titan from the Turing era looks quaint these days, despite its prodigious power (at the time). (Image: Nvidia)

Instead of launching a Titan, Nvidia is reportedly focusing on the inevitable RTX 4090 Ti. When it launched the 4090, it was obvious Nvidia had left some room for a future GPU. Jensen stated the full AD102 die has over 18,000 CUDA cores. Despite that, the RTX 4090 has “only” 16,384. This is a typical strategy for Nvidia; leaving some room in case it needs to battle an AMD GPU.

The rumored 4090 Ti will likely be around 10 to 20 percent more powerful than the RTX 4090. According to Wccftech, it’ll sport faster 24Gb/s GDDR6X memory, more CUDA cores, and higher clock speeds due to binned dies. This could grant it a boost clock approaching 3GHz. Power consumption will probably only increase 25W though, making it a 475W GPU. Some board partners might push this number up beyond 500W though.

As always, the 4090 Ti exists as Nvidia’s ace card, tucked away in case of emergency. It might not need it. New reports suggest Nvidia has outgunned AMD this time around, even though AMD’s cards are yet to launch. New leaks suggest its RDNA3 GPUs will land somewhere in between the RTX 4090 and 4080 16GB. Nvidia’s spies might have gathered some intelligence on AMD’s plans, which could have motivated it to cancel its RTX 4080 12GB. That GPU was offered at $899, which might have been a bridge too far considering what AMD might have on offer. We’ll find out soon enough, as AMD will announce its next-gen GPUs on Nov. 3.

Now Read:



sourse ExtremeTechExtremeTech https://ift.tt/BunteYQ

ليست هناك تعليقات:

إرسال تعليق