Video Cards & GPUs - Page 322

AMD has just created massive buzz for CES 2019 by announcing that it will "catapult computing, gaming, and visualization technologies forward with the world's first 7nm high-performance CPUs and GPUs, providing the power required to reach technology's next horizon".

We should expect the world's first 7nm CPU to be fully unveiled and shown on stage with the next wave of EPYC processors, as well as the world's first 7nm GPU that should take form as Vega 20 inside of a new Radeon Instinct accelerator. The new EPYC 'Rome' 7nm CPUs should come with 64C/128T of processing power, utterly demolishing Intel with its flagship 28C/56T processor that is still stuck on the 14nm node.

10nm won't be a thing for Intel until late 2019 giving AMD a process node leadership it hasn't had in a decade. Meanwhile, NVIDIA has just unleashed its new Turing GPU architecture on the 12nm process, with 7nm GPUs coming from NVIDIA in 2019. AMD will be ahead of both of their main competitors for process node technology, and this is something that AMD should be shouting from the roof tops, or in this case the Venetian in Las Vegas on January 9, 2019 at 9AM.

NVIDIA has opened up the pre-orders of the new Quadro RTX 5000 and RTX 6000 graphics cards in the US, while the flagship Quadro RTX 8000 isn't available for pre-orders right now and once it does, it'll sell for a huge $10,000.

Quadro RTX 5000 costs $2300 while the Quadro RTX 6000 costs $6300 and there are vast differences between them, with the RTX 5000 packing the TU104 GPU while the RTX 6000 and RTX 8000 have the TU102 GPU. One of the large differences between the Quadro RTX 5000 and RTX 6000 is that there is 16GB of GDDR6 is on the RTX 5000 while the RTX 6000 packs 24GB of GDDR6.

If you want to pre-order NVIDIA's new Quadro RTX 5000 or RTX 6000 graphics cards, you can do that here.

You can't even buy a new GeForce RTX 2080 Ti graphics card right now, and while you can get the RTX 2080 it's something I wouldn't recommend over the GeForce GTX 1080 Ti. That isn't stopping companies from unveiling their custom NVLink bridges, with ASUS unveiling the new ROG-NVLink bridge.

As you can see, ASUS has put some amazing design work into the ROG-NVLink bridge. There's not much to work with, and while NVIDIA's stock NVLink bridge looks good, ASUS takes it to another level.

ASUS has their new ROG-NVLink bridge in both 3- and 4-slot for bigger motherboards as well as Aura Sync technology with its RGB lighting.

MSI has announced two new GeForce RTX 20 series graphics cards with its next-gen Sea Hawk EK X series with the new flagship MSI GeForce RTX 2080 Ti Sea Hawk EK X that features a beyond gorgeous EKWB water block.

MSI has also revealed the new RTX 2080 Ti Sea Hawk X, with both of the cards featuring identical specs in terms of GPU base and boost clocks, and GDDR6 clocks. We have Turing GPU boost clocks of up to 1755MHz (13.6% faster than RTX 2080 Ti Founders Edition) on both cards, while both cards have a 300W TDP (up from 250W on the FE). Both cards have the same clocks as the air-cooled MSI GeForce RTX 2080 Ti Gaming X TRIO.

It's a damn tall card, but man does that EK water block look amazing.

AMD will unleash their next-gen Navi GPU architecture in 2019, something the engineers at Radeon Technologies Group worked directly with Sony on for the next-gen PlayStation 5... but what about post-Navi? We're now hearing about a new GPU architecture for 2020 and beyond called Arcturus.

AMD's purported next-gen Arcturus GPU architecture has been teased by an AMD employee on the Phoronix Forums, with this person adding that a new line up of graphics cards will continue using the older Radeon naming scheme versus the one used for Vega. This means we shouldn't expect the new cards to be called Radeon RX Navi or Radeon RX Arcturus, and rather Radeon RX 680 and Radeon RX 780, I guess?

The new Arcturus codename continues the star theme that started with Polaris and continued with Vega, right into Navi and now the purported Arcturus GPU architecture codename. The AMD staffer said: "We are going back to real codenames where possible (ie where it's not already too late to change), and trying to keep the old "the name vaguely suggests HW generation so it will be easier to remember in the future" model. The first new codename should be Arcturus".

There are plenty of custom GeForce RTX 2080 and RTX 2080 Ti graphics cards that are announced, available to buy, on their way, or coming very soon... but one of which isn't here yet are the custom GIGABYTE and AORUS branded cards, something we're now seeing for the first time and man do they look GOOD.

GIGABYTE has started teasing its upcoming AORUS GeForce RTX 2080 XTREME 8G graphics card on social media, so now we know it rocks a gorgeous triple-fan Windforce cooler. We have huge 100mm fans, direct heat pipes, and 12+2 power phases. On the back, GIGABYTE is using a metal backplate with an AORUS logo that has RGB lighting.

One thing that GIGABYTE has done differently to other AIB partners with its upcoming AORUS GeForce RTX 2080 XTREME 8G is that it has 7 display outputs... 5 of which can be used at once. We have 3 x DisplayPort 1.4, 3 x HDMI 2.0b, and 1 x USB-C with VirtualLink.

NVIDIA has launched its two higher-end RTX graphics cards with both the GeForce RTX 2080 and RTX 2080 Ti released, but the RTX 2080 Ti isn't being made available for another week or so for consumers. The company has now come out and announced an October 17 launch for its GeForce RTX 2070.

The upcoming GeForce RTX 2070 graphics card uses the Turing TU106 GPU, TU106 features 10.8 billion transistors, 2304 CUDA cores, 288 Tensor Cores, and 36 RT Cores. The transistor count on TU106 is much greater than its Pascal counterpart in the GP104, with the new TU106 GPU featuring 10.8 billion transistors versus the 7.2 billion inside of GP104.

TU106 has 3 x GPCs, 36 SMs, and 8 x 32-bit memory controllers (for a 256-bit memory bus, the same as the RTX 2080). NVIDIA's new TU106G GPU sees each GPC with a raster unit and 6 x TPCs, with each TPC packing a PolyMorph Engine and 2 x SMs. Each of the SMs inside of TU106 will, like TU102 and TU104, feature the new RT Core for ray tracing. Each SM packs 64 CUDA cores, 256KB register file, 96KB L1 data cache/shared memory cache, and 4 x texture units.

AMD refreshed the Polaris 10 GPU a couple of months ago when we exclusively revealed the Radeon RX 500X series was going out to OEMs with the Polaris 20 GPU inside, but now there's rumors that AMD will refresh the refresh with the purported Polaris 30.

The rumors have started on Chiphell and we haven't confirmed them with any AMD sources, but Polaris 30 will reportedly be made on the 12nm process compared to the 14nm node that AMD used on the Polaris 10/20 and Vega 10 GPUs. The rumors are hot, stating that AMD could launch the 'new' Polaris 30-based graphics cards as soon as next month, ahead of the new Vega 20 GPUs made on 7nm.

We are being told to expect a 15% performance improvement over the RX 500 series, which should be enough to get the purported Radeon RX 680 (that's what I'll call it for now) to be getting close to GeForce GTX 1070 Ti and GTX 1080 performance. Hell, it'll start competing with the Radeon RX Vega 56 and 64 at that point, which won't be good for AMD unless Vega is discontinued - which it should be at this stage.

NVIDIA has just unleashed their new GeForce RTX series graphics cards led by the flagship GeForce RTX 2080 Ti (our review here) as well as the new GeForce 411.63 drivers (that you can get here) and now it's AMD's turn for some Radeon news.

AMD has just released their updated Radeon Software Adrenalin Edition 18.9.2 drivers (download them here), which include or add support for games like F1 2018 with up to 3% more performance on the Radeon RX Vega 64 at 1440p. Shadow of the Tomb Raider is one of the latest games out, with the new RSAE 18.9.2 drivers being up to 4% faster on the RX Vega 64 at 1440p.

Even Fortnite receives some optimization under the new drivers, with up to 5% more performance at 1080p on the Vega 64 while Star Control: Origins is up to 14% fastrer on the Vega 64 when on native 4K.

NVIDIA's new GeForce RTX series graphics cards are finally here and it did NOT take overclockers very long to get into them and start pushing the new Turing GPU and super-fast GDDR6 memory technology to their (current) limits. I'm sure we'll see more records broken as time goes on, and as more custom cards with custom PCBs and power delivery systems arrive.

Legendary overclocker KINGPIN has used the new GeForce RTX 2080 Ti Founders Edition with EVGA's Precision X1 overclocking software and pushed the cards to limits that I thought we wouldn't be at for a while. The new Turing TU102 GPU was pushed up to 2415MHz (I could only do 2100MHz on stock cooling) while the 11GB of GDDR6 was ramped up from its already fast stock clocks of 14Gbps up to 17Gbps (I could reach 15-16Gbps stable, depending on the bench). All of this was done with LN2 cooling, so the Turing GPU and GDDR6 is being kept as cool as humanly possible.

KINGPIN volt modded the card himself, passing the voltage limits that NVIDIA put in place, so that he could overclock the hell out of the RTX 2080 Ti. Using LN2, KINGPIN was able to take the #1 spots in all 3DMark benchmarks: TimeSpy Extreme, TimeSpy, FireStrike Ultra, FireStrike Extreme, and FireStrike. The system packed an Intel Core i9-7980XE processor cooled with LN2 and overclocked to 5.5GHz on all 18C/36T, an EVGA X299 Dark motherboard, and 32GB of G.SKILL Trident Z DDR4-3800 RAM.