NVIDIA GeForce RTX 4090 Is The First Gaming Graphics Card To Ship 100 TFLOPs of Compute Efficiency
On: October 12, 2022
NVIDIA’s GeForce RTX 4090 is the primary gaming graphics card to realize over 100 TFLOPs of compute efficiency. You may also learn our full evaluation of the cardboard right here.
Breaking The 100 TFLOPs Barrier! NVIDIA GeForce RTX 4090 Turns into The Quickest Gaming Graphics Card For Compute & Quickest Gaming Graphics Card, Interval!
Breaking the 100 TFLOPs barrier is not any simple feat. Earlier than at present, NVIDIA’s quickest gaming graphics card, the GeForce RTX 3090 Ti, solely delivered 40 TFLOPs of compute horsepower. With the launch of the GeForce RTX 4090, we get near the 100 TFLOPs barrier however not formally. NVIDIA states that the GeForce RTX 4090 Founders Version provides 83 TFLOPs at default settings. Which means the cardboard is 17 TFLOPs shy of that 100 TFLOPs mark.
So we determined it was time to check how far we will push the NVIDIA GeForce RTX 4090 Founders Version with some overclocking. To get to 100 TFLOPs, we first pushed the ability restrict and temp restrict slider all the best way to the max and upped the Core and Reminiscence clocks by +275 and +1100 MHz, respectively. This wasn’t sufficient as the cardboard was being restricted by its energy design. That’s once we landed our fingers on MSI’s newest Afterburner which allowed us to boost the core voltages. At 100%, we noticed some efficiency regression so we needed to keep on with +55% which confirmed us some good outcomes.
With the overclock utilized on our NVIDIA GeForce RTX 4090 graphics card, we noticed a most GPU core clock of 3150 MHz on the AD102 Ada GPU, a most energy draw of 547W and our temps peaked at 69C. All of this was accomplished on air and with no unique liquid cooling, chillers or LN2 had been used.
And behold, we noticed the magical variety of not 100 however nearly 101 TFLOPs right in entrance of our eyes. To place issues into perspective, it is a 22% compute increase over the inventory RTX 4090 and a 2.5x compute efficiency increase over the RTX 3090 Ti. The AD102 GPU additionally ripped aside the data-center-focused Hopper H100 GPUs by providing over 50% higher FP32 efficiency. Ada Lovelace is really a game changer and we will undoubtedly see it develop into a well-liked compute and AI graphics card when Quadro variants of the stated chip launch because the RTX 6000 ADA and L60.
NVIDIA GeForce RTX 4090 ‘Official’ Specs – $1599 US Pricing
The NVIDIA GeForce RTX 4090 will use 128 SMs of the 144 SMs for a complete of 16,384 CUDA cores. The GPU will come full of 72 MB of L2 cache and a complete of 176 ROPs which is solely insane.
As for reminiscence specs, the GeForce RTX 4090 will function 24 GB GDDR6X capacities that will likely be clocked at 21 Gbps speeds throughout a 384-bit bus interface. This can present as much as 1 TB/s of bandwidth. This is similar bandwidth as the prevailing RTX 3090 Ti graphics card and so far as the ability consumption is anxious, the TBP is rated at 450W. The cardboard will likely be powered by a single 16-pin connector which delivers as much as 600W of energy. Customized fashions will likely be providing larger TBP targets.
The NVIDIA GeForce RTX 4090 GPU formally hits retail cabinets tomorrow when NVIDIA and customized card companions’ designs develop into accessible to the general public. You possibly can try our evaluation right here.
