Graphics Cards - Page 56

AMD's mobile-focused Radeon RX 7800M has been tested over USB4 and OCulink, with up to 28% more performance using OCulink versus USB4.

The Radeon RX 7800M is a Navi 32-based GPU with 3840 Stream Processors, which is the full configuration of the Navi 32 GPU for the desktop. The big change there is that it doesn't feature the full memory bandwidth, limited to a 192-bit memory bus and just 12GB of GDDR6 memory.

Our friends over at HotHardware tested the Radeon RX 7800M over both USB4 and OCulink, as the 7800M supports both of those connectivity standards. USB4 is limited to Thunderbolt 4/3 speeds at 40Gbps, while OCulink pumps at the full PCIe 4.0 x4 bandwidth of 64Gbps... so how does that additional bandwidth help the 7800M? Pretty decently, actually.

NVIDIA will reportedly launch 4 different GeForce RTX 50 series GPUs in Q1 2025, with new information teasing we'll see the RTX 5090, RTX 5080, RTX 5070 Ti, and RTX 5070 all released in early 2025.

The new information is coming from Benchlife, which reports that its sources from the AIC and "various passive component agents" that NVIDIA will "launch a complete GeForce RTX 50 series graphics card in the first half of 2025". The list included a rundown of the launches, with the 4 higher SKUs launching in Q1 2025, and the others are "TBC" right now.

NVIDIA GeForce RTX 50 series "Blackwell" GPU release rumors:

NVIDIA is reportedly working on its GeForce RTX 5070 Ti graphics card, with new leaks teasing some information on CUDA core counts and power consumption.

The purported GeForce RTX 5070 Ti will rock and roll with a rumored 8960 CUDA cores, 70 Streaming Multiprocessors, and the expected GB203 "Blackwell" GPU. The leaked data suggests the card is called PG147 and SKU 60 internally at NVIDIA, meaning it uses the same board as the RTX 5080 (which is SKU 45) which means we'll see a cut-down GB203 that's used on the RTX 5080.

Leaker "kopite7kimi" has said that the RTX 5070 Ti should feature a 300W TDP, which is a 50W increase over the RTX 5070 and 100W less than the RTX 5080... so right in the middle. We don't know if we're looking at 300W of TGP (maximum) or TDP (thermal design power).

NVIDIA's next-generation GeForce RTX 5090 graphics card will use the GB202 "Blackwell" GPU, which is reportedly 744mm2 making it around 22% larger than the AD102 "Ada Lovelace" GPU inside of the current-gen flagship RTX 4090.

In a new post on X by leaker "MEGAsizeGPU" we hear that the GB202 GPU has a die size of 24x31, which measures out to 722mm2, with the GPU featuring a package size of 63x56, which is roughly 3528mm2. The GeForce RTX 4090 features the AD102 GPU which has a die size of 609mm2, and a package size of 2601mm2 (51x51mm). This means the GB202 GPU has a 22% increase in die size, and 35% increase in packaging area.

NVIDIA's next-gen GeForce RTX 5090 and its GB202 GPU might sound huge, but the TU102 "Turing" GPU featured a 754mm2 die, and was the first to introduce the Tensor and RT Core technology on a GeForce Gaming GPU, with a reduction in die size during Turing, with improved process node optimization with each generation.

NVIDIA's RTX 5090 has popped up on pre-order, kind of - not as a standalone product, of course, but as part of seriously powerful liquid-cooled servers and workstations from a vendor called Comino.

The revelation came courtesy of a press release from Comino, spotted by Tom's Hardware, offering 'priority access' to would-be buyers of these workstations, so you can "be among the first to harness RTX 5090 power."

Comino's Grando server, which is equipped with eight RTX 5090 graphics cards, can be 'pre-ordered' via email or an online form, but as you might expect, this is more about registering your interest at this point - and providing details of the project you plan to use the system with. You'll then get an ETA for the arrival of your server (and stock will be limited to begin with, unsurprisingly).

Yesterday, we reported on NVIDIA's third-quarter financial earnings report, in which the company recorded $35.1 billion in total revenue and a gross margin of 74.6%. Granted, a big chunk of this comes from NVIDIA's Data Center business, where its AI GPUs and services are leading a worldwide computing boom.

Still, its GeForce RTX Gaming and AI PC sector also delivered impressive revenue growth, $3.3 billion for the quarter, up 15% from a year ago. However, this figure is set to drop, as NVIDIA CFO Colette Kress strongly hinted there will be supply constraints for the current quarter and that there could be a gaming GPU shortage until the new calendar year.

"Our gaming supply, given what we saw selling through, was moving quite fast," Colette Kress responded when asked for further clarification. "The challenge that we have is, how fast could we get that supply into the market for this quarter? Not to worry, I think we'll be back on track with more supply as we turn the corner into the new calendar year."

NVIDIA spoke to investors during a recent call, which revealed that despite potential cooling issues with Blackwell GPUs, NVIDIA will continue to push "full steam" ahead.

NVIDIA just wrapped up its Q3 2025 earnings call. The company touted the power of its coming Blackwell AI chips and how the massive popularity of its AI-centered chips will only fuel the company's revenue and profits further. Notably, NVIDIA moved to the top of the list of the world's most valuable publicly traded companies, passing the likes of Microsoft, Apple, and Amazon.

NVIDIA was able to achieve this through its AI data center business, which accounted for a staggering $30.7 billion of its quarterly earnings. By comparison, NVIDIA's gaming sector 'only' pulled in around $2-3 billion. Additionally, much of that quarterly revenue is profit for NVIDIA, with the company posting profit figures for the following quarters: Q1, $14.8 billion, $16.6 billion in Q2, and $19.3 billion in Q3. By comparison, Apple and Microsoft made $21.4 billion, and $24.7 billion profit this past quarter, respectively.

NVIDIA has posted its financial results for the third quarter that ended on October 27, 2024. With $35.1 billion in revenue, the company once again exceeded expectations. This figure is 17% higher than the previous quarter and 94% higher than the same period of last year, driven once again by AI and the company's Data Center segment. And with a gross margin of 74.6%, NVIDIA shows no signs of slowing down.

"AI is transforming every industry, company, and country," said Jensen Huang, founder and CEO of NVIDIA. "Enterprises are adopting agentic AI to revolutionize workflows. Industrial robotics investments are surging with breakthroughs in physical AI. And countries have awakened to the importance of developing their national AI and infrastructure."

That's a lot of "AI" in one statement. Jensen Huang also said that the entire planet is shifting toward "NVIDIA computing," which is an incredibly bold thing to say, while confirming that its highly anticipated Blackwell chips are in full production for Data Center, Gaming, and AI PCs.

Although we're still waiting for Intel to launch its second-gen Arc desktop 'Battlemage' graphics cards for desktop PCs, with the latest rumors pointing to a December 2024 launch, the Xe2 graphics architecture is already out there as part of the company's latest Lunar Lake Core Ultra 200V mobile chips.

According to a new benchmark listing, Battlemage-next or Xe3 is already being tested internally as part of Intel's upcoming Panther Lake chips. No matter what happens with Intel's ambitions in the gaming GPU space, it certainly looks like Arc graphics will continue to be developed as APUs become more prevalent in mobile devices and gaming handhelds.

As part of its latest Q3 2024 Earnings Call, Intel CEO Pat Gelsinger noted that the future of Arc would involve fewer desktop SKUs and a greater focus on integrated graphics. It will be interesting to see if Intel releases an Xe3-powered desktop GPU or if the technology remains in the mobile space.

AMD is abandoning the RDNA name after the (soon to arrive) RX 8000 family of GPUs, with the following generation after RDNA 4 set to be called UDNA.

That's what we're told by a regular leaker on the Chiphell forum, Zhangzhonghao, who explains that CDNA and RDNA, Instinct AI (data center) and Radeon gaming GPUs respectively, will both be brought under the same UDNA umbrella sooner than we expected.

AMD had already announced its planned transition to UDNA - one unified (hence the 'U') architecture - a couple of months ago. Hints were dropped that made us think RDNA 5 would probably still happen - and the move to UDNA GPUs would come after that - but if this rumor is right, the change will come through more quickly. RDNA 5 won't exist, with RX 9000 taking the UDNA name.

Auras Technology is a GPU cooling supplier. Its cold plates and manifolds are used in GeForce RTX GPUs for desktop users. The company is also on NVIDIA's recommended supplier list. It will supply parts to its partners for the upcoming GeForce RTX 50 'Blackwell' Series of cards, which are expected to launch early next year.

During the company's recent financial earnings call, Auras chairman Yu-Shen Lin said that the upcoming GeForce RTX 50 Series might "seize the whole market starting in December." This will increase the demand for its cooling products, covering both the desktop PC and server markets.

Cooling is an essential component of any consumer GPU used for PC gaming. In fact, the bulk of modern GPU shrouds are all about housing fans, heatsinks, and vapor chambers. With the word that the next-gen flagship GeForce RTX 5090 will draw more than 450 Watts, you can be sure that cooling will be essential for the upcoming RTX 50 Series.

Nothing is official, and it only is in the tech world once the hardware is ready to go out the door. However, several credible sources are pointing to NVIDIA launching its next-gen gaming-focused GeForce RTX 50 Series 'Blackwell' GPUs in January 2025, with the big reveal happening at CES 2025.

We know the specs and details about the new flagship GeForce RTX 5090, RTX 5080, and RTX 5070, right down to chip names and memory configurations. Again, nothing is official, but it looks like NVIDIA is preparing to launch all three in early 2025 - within a few weeks or a month. And with that, Team Green is reportedly already ramping down GeForce RTX 40 Series production to make way for the shiny new GeForce RTX 50 Series.

According to the latest rumors from the Chinese-based Board Channels forum (via Gazlog), which cite AIB partners as the source, the supply of everything from the GeForce RTX 4070 to the GeForce RTX 4090 is about to dry up, and retail stock will be potentially hard to find as we head into the new year.

Usually, packaging isn't the first thing that leaks for an unannounced GPU, but that seems to be the case for what is expected to be Intel's new mainstream and budget Arc B580 Graphics Card. As spotted by @josefk972 on X, sleeves or packaging for 'INTEL BMG B580' have appeared on a shipping manifest.

Rumors suggest that Intel will launch its next-gen Battlemage desktop graphics cards in December next month to get ahead of the GeForce RTX and Radeon RX deluge that we're all expecting to see at CES 2025. The Intel Arc B580 will be a desktop card based on the company's Battlemage or Xe2 architecture and a direct successor to the Intel Arc A580.

We reviewed the first-gen Intel Arc A 580 last year, taking the stylish SPARKLE Intel Arc A580 ORC OC Edition for a spin. At the time, the sub-$200 GPU was a great choice for entry-level 1080p gaming - outperforming the GeForce RTX 3050. A new $200 GPU from Intel would help differentiate it from the high-end and mid-range GPUs on the way from NVIDIA and AMD.

Another hint has been dropped that NVIDIA's next-gen GeForce graphics cards could be serious power hogs.

The latest on the incoming RTX 5000 series GPUs, which suggests that they could up the wattage requirements from the already power-hungry levels of RTX 4000 boards, comes from Corsair in a rather unexpected manner.

As part of a press release (noticed by VideoCardz) about next-gen desktop graphics cards expected to launch in 2025 - there are three new generations, in fact, from AMD, Intel and NVIDIA - underlines that Corsair's power supplies will fully support these new GPUs.

NVIDIA is celebrating the 20th anniversary of Valve's game-changing Half-Life 2, with an exclusive custom-wrapped GeForce RTX 4080 SUPER graphics card with the Half-Life 2 logo and Gordon Freeman... you can win it, too! Check it out:

NVIDIA explained in its post on X: "Welcome! Welcome to City 17! To celebrate the 20th anniversary of Half-Life 2, we're giving away an exclusive custom wrapped GeForce RTX 4080 SUPER GPU, highlighting the iconic Gordon Freeman. Want it? Comment #HalfLife2RTX + like this post to enter!"

The giveaway is awesome, but I don't know if I'd ever want to install that into a PC... it would be an amazing collector's item to sit on a shelf (in glass, of course). But, NVIDIA's GeForce RTX 4080 SUPER graphics card packs 10240 CUDA cores and 16GB of GDDR6X memory, which will be able to easily handle Half-Life 2 RTX when it drops with its huge upgrade in graphics and lighting.

According to Jon Peddie Research, the global graphics processor unit (GPU) market will be worth almost $100 billion by the end of the year - an astonishing figure that factors in the current AI boom and highly sought-after GPUs from NVIDIA.

The report notes that GPUs are becoming increasingly ubiquitous in technology and "can be found in almost every industrial, scientific, commercial, and consumer product made today." This includes game consoles, PCs, data centers, handhelds, VR, wearables, smart TVs, vehicles, and more. The report also acknowledges the rise of integrated GPUs, as seen in the latest Apple, AMD, Intel, and Qualcomm processors.

The report quickly points out that GPUs for data centers are "low-volume compared to other market segments," but when an NVIDIA GPU can cost tens of thousands of dollars, and demand for them is through the roof - AI is a big part of the GPU market that "will exceed $98.5 billion in 2024."

NVIDIA has reportedly shifted production to its new GeForce RTX 50 series "Blackwell" GPUs, with just a single RTX 40 series "Ada" GPU still in production.

In a new report from Board Channels, we're hearing that NVIDIA is reallocating its production lines to the new RTX 50 series, leaving the AD107 production line in operation. This means that NVIDIA's fleet of Ada Lovelace GPUs -- AD102, AD103, AD104, and AD106 -- have ceased production, the GeForce RTX 40 series is close to being retired (at least at a production level).

This doesn't mean NVIDIA's current-gen GeForce RTX 40 series "Ada Lovelace" GPUs have been discontinued, it just means that no NEW cards will be produced. All NEW cards produced will be the next-gen GeForce RTX 50 series "Blackwell" GPUs, which will be unleashed at CES 2025.

NVIDIA has revealed a fresh incentive to buy an RTX 4000 graphics card, which is a free copy of Indiana Jones and the Great Circle.

As ever, this free game comes with qualifying Lovelace GPUs, and that's not just desktop RTX 4000 models, but also laptops with one of the mobile flavors of the current-gen range.

What are those qualifying GPU models? They are as follows: NVIDIA RTX 4090, 4080 Super, 4080, 4070 Ti Super, 4070 Ti, 4070 Super, and 4070 desktop cards. With laptops, they are: NVIDIA RTX 4090, 4080, and 4070 mobile GPUs.

NVIDIA has released a new graphics driver that comes with support for S.T.A.L.K.E.R. 2 and Microsoft Flight Simulator 2024, and it also packs the fix for a bug that has been plaguing those with Corsair peripherals.

The GeForce Game Ready Driver version 566.14 brings in support for the mentioned games, which are also getting DLSS 3 Frame Generation - good news for your frame rates (those with RTX 4000 graphics cards, of course).

The driver is the first to be released following the debut of the official NVIDIA App, which just exited from its beta test, pushing GeForce Experience into obscurity (for those who used it, anyway). You can use that official app to grab the new v566.14 driver, of course.

The NVIDIA App launched into Beta earlier this year, incorporating all the top features from GeForce Experience, the NVIDIA Control Panel, and new RTX-powered tools and AI features. It is a one-stop shop for all things GeForce RTX, serving as the latest place to update drivers, optimize games, change display settings, or install and access RTX apps like GeForce NOW, NVIDIA Broadcast, ChatRTX, and more.

After several updates that have improved the app's performance and added new and highly requested features, NVIDIA is ready to announce that the NVIDIA App is officially released and no longer in beta. So, if you've been holding out on jumping from GeForce Experience, now's the time.

From the enhanced overlay and game recording features to the excellent tools for optimizing game performance to the fast and responsive performance of the app itself - it's a must-install for the millions of GeForce RTX owners out there.