⋮  

NVIDIA GeForce RTX Series Is Purpose Built For OC – Will Be Quiet As A Whisper

Sep 14, 2018 06:37 EDT
Submit

Another day, another leak by Videocardz, this time about a design feature of RTX-series that is sure to resonate well with enthusiasts getting ready to part with their money. It looks like NVIDIA has drastically improved power delivery systems and voltage curves with their new Turing lineup.  If the leaked slides are anything to be believed, the card's PCB will deliver a much cleaner voltage supply to the GPU - allowing for greater OC headrooms and cooler operation.

NVIDIA's RTX series has overclocking in its soul: GeForce RTX 2080 Ti, RTX 2080 and 2070 will be overclocking beasts

Let's start with the sound levels and acoustics. The GTX 1080 - which is what NVIDIA is using for comparison against the RTX 2080 - can become as loud as 36 dBA under load but the RTX 2080 will only hit a maximum peak of 29 dBA. This is a pretty significant reduction and can be the difference between "annoying-whining-hum" and "a-comfortable-hum-that-you-can-barely-hear". To put this into perspective a standard whisper is around 30 dBA. You can officially say that these things are as quiet as a whisper now.

NVIDIA GeForce RTX 3070 Graphics Card To Have Considerably Larger Supply Compared To RTX 3080 & RTX 3090, APAC Retailers Report

Up next, we have the slides that overclockers will actually be interested. NVIDIA has given new voltage curves of the cards and the VRMs can supply an unholy level of amperege to the GPU if required. Not only that but the voltage signal is now super clean and a lot tighter than before. Noise and ripple has been virtually eliminated and the company has managed to add on an additional 20 Watts of TDP envelope for a total of 55 Watts (this was 35 Watts with the GTX 1080). These additional few wattts will allow for more overclocking headroom - which I am sure their own Turbo Dynamic boost will utilize as well.

This brings me to the second point. While the slides clearly state that the GPU is meant for overclocking, these features will impact the average joe a lot more than actual professional OC-ers - wanna know why? Because they will undoubtedly be utilized by their dynamic turbo boost service which automatically applies the highest stable overclock possible to the card given the thermal conditions. This is also why Pascal cards run a lot higher in terms of frequency than their minimum guaranteed turbo boost - this is also why I suspect Turing will run a lot higher than the guaranteed turbo boosts imply.

NVIDIA GeForce RTX/GTX "Turing" Family:

Graphics Card NameNVIDIA GeForce GTX 1650NVIDIA GeForce GTX 1650 D6NVIDIA GeForce GTX 1650NVIDIA GeForce GTX 1660NVIDIA GeForce GTX 1660 SUPERNVIDIA GeForce GTX 1660 TiNVIDIA GeForce RTX 2060NVIDIA GeForce RTX 2070NVIDIA GeForce RTX 2080NVIDIA GeForce RTX 2080 Ti
GPU ArchitectureTuring GPU (TU117)Turing GPU (TU117)Turing GPU (TU116)Turing GPU (TU116)Turing GPU (TU116)Turing GPU (TU116)Turing GPU (TU106)Turing GPU (TU106)Turing GPU (TU104)Turing GPU (TU102)
Process12nm FNN12nm FNN12nm FNN12nm FNN12nm FNN12nm FNN12nm FNN12nm FNN12nm FNN12nm FNN
Die Size200mm2200mm2284mm2284mm2284mm2284mm2445mm2445mm2545mm2754mm2
Transistors4.7 Billion4.7 Billion6.6 Billion6.6 Billion6.6 Billion6.6 Billion10.6 Billion10.6 Billion13.6 Billion18.6 Billion
CUDA Cores896 Cores896 Cores1280 Cores1408 Cores1408 Cores1536 Cores1920 Cores2304 Cores2944 Cores4352 Cores
TMUs/ROPs56/3256/3280/3288/4888/4896/48120/48144/64192/64288/96
GigaRaysN/AN/AN/AN/AN/AN/A5 Giga Rays/s6 Giga Rays/s8 Giga Rays/s10 Giga Rays/s
Cache1.5 MB L2 Cache1.5 MB L2 Cache1.5 MB L2 Cache1.5 MB L2 Cache1.5 MB L2 Cache1.5 MB L2 Cache4 MB L2 Cache4 MB L2 Cache4 MB L2 Cache6 MB L2 Cache
Base Clock1485 MHz1410 MHz1530 MHz1530 MHz1530 MHz1500 MHz1365 MHz1410 MHz1515 MHz1350 MHz
Boost Clock1665 MHz1590 MHz1725 MHz1785 MHz1785 MHz1770 MHz1680 MHz1620 MHz
1710 MHz OC
1710 MHz
1800 MHz OC
1545 MHz
1635 MHz OC
Compute3.0 TFLOPs3.0 TFLOPs4.4 TFLOPs5.0 TFLOPs5.0 TFLOPs5.5 TFLOPs6.5 TFLOPs7.5 TFLOPs10.1 TFLOPs13.4 TFLOPs
MemoryUp To 4 GB GDDR5Up To 4 GB GDDR6Up To 4 GB GDDR6Up To 6 GB GDDR5Up To 6 GB GDDR6Up To 6 GB GDDR6Up To 6 GB GDDR6Up To 8 GB GDDR6Up To 8 GB GDDR6Up To 11 GB GDDR6
Memory Speed8.00 Gbps12.00 Gbps12.00 Gbps8.00 Gbps14.00 Gbps12.00 Gbps14.00 Gbps14.00 Gbps14.00 Gbps14.00 Gbps
Memory Interface128-bit128-bit128-bit192-bit192-bit192-bit192-bit256-bit256-bit352-bit
Memory Bandwidth128 GB/s192 GB/s192 GB/s192 GB/s336 GB/s288 GB/s336 GB/s448 GB/s448 GB/s616 GB/s
Power ConnectorsN/AN/A6 Pin8 Pin8 Pin8 Pin8 Pin8 Pin8+8 Pin8+8 Pin
TDP75W75W100W120W125W120W160W185W (Founders)
175W (Reference)
225W (Founders)
215W (Reference)
260W (Founders)
250W (Reference)
Starting Price$149 US$149 US$159 US$219 US$229 US$279 US$349 US$499 US$699 US$999 US
Price (Founders Edition)$149 US$149 US$159 US$219 US$229 US$279 US$349 US$599 US$799 US$1,199 US
LaunchApril 2019April 2020November 2019March 2019October 2019February 2019January 2019October 2018September 2018September 2018
How high do you think RTX-Series graphics cards (like the 2080) will OC?
Submit