As expected, Nvidia is going ahead with the launch of a 12GB version of its vaunted RTX 3080 graphics card, which is 2GB more than the previous version. Nvidia mildly bumped up some other specs as well, so it’s not just a simple “slap some more memory on that bad boy” scenario, but such a modest upgrade to an existing model is rare in the GPU world. And just like with previous launches in the pandemic, it comes with the asterisk that most people won’t even be able to purchase one, even if they are willing to pay an inflated MSRP for it, making it another GPU launch that elicits more sadness than jubilation. On a similar note, good luck finding a review of it, because it seems like Nvidia didn’t allow any day-one reviews for press, not even from its partners, but more on that later.
Despite that dour intro, the new GPU’s specs (mostly) look good on paper, thanks to its small-but-meaningful upgrades. The bump to 12GB should help the card run AAA games at high resolution that need just a smidge more RAM than 10GB, either to handle future high-resolution games or to juggle the VRAM demands of rasterization and ray tracing simultaneously. It was a bit weird for Nvidia to hold the RTX 3080 to just 10GB of VRAM at launch given how long GPUs have sat at or near 8GB. Increased GDDR6X costs and limited availability have reportedly impacted Ampere’s VRAM loadouts since it launched.
Nvidia has opened up the pipe a bit, upgrading the card’s memory bus from 320-bit in the previous model to 384-bit like the rest of its flagship cards. This is a big upgrade, and it provides 19 percent more memory bandwidth, from 760GB/s to 920GB/s. Nvidia also bumped up the CUDA cores, slightly, from 8,704 to 8,960. That’s about a three percent boost.
Clock speeds have also been given a nudge. Base clock is up 180MHz and boost clock nudges up 40MHz. The increase in base clock is substantial, but many factors govern clock speed during gaming. Base clock improves by 14 percent but boost clock is only up 2 percent. How much the RTX 3080 12GB gains over the RTX 3080 10GB will depend on which clock improvement has more of an impact in real-world gaming.
All these moderate changes have increased the card’s power requirements, taking it from 320W to 350W. That’s the same as the RTX 3080 Ti.
Nvidia’s website lists both the OG 3080 and the new variant, but it’s unclear if it will be selling a Founder’s Edition of the card, or for what price. Regardless, the 12GB model has begun popping up on several of the company’s partner sites, including EVGA which has a full landing page for it. The company is listing two versions: a super-thicc, triple-slot FTW version for $1,299, and a dual-slot XC3 Ultra for $,1249. The chonky FTW card requires three 8-pin PCIe power connectors, which is unchanged from the previous model. Galax is also promoting its new GPU on Twitter, with the unintentionally funny moniker “Ex Gamer.” Seriously Galax, at these prices we may all be ex-gamers before too long.
Overall, the new RTX 3080 12GB is a mild upgrade. We will have to wait and see if the extra memory and additional bandwidth will allow for big gains in high-resolution gaming, especially with ray tracing. The underlying issue is the GPU at the heart of the card also needs to be able to deliver the necessary performance this extra memory might unlock.
For example, Nvidia released an RTX 3060 GPU with 12GB of RAM last year to compete with AMD’s 6700 XT’s 12GB, but nobody would argue those cards can drive AAA gaming on a 4K panel with ray tracing. On the RTX 3080 however, that might actually be possible in some titles. Maybe by giving it a bit more RAM, the 3080 will actually deliver the performance it should have had at launch, so long ago.
Finally, as far as reviews for the card go, here’s where things get strange. According to TechSpot, Nvidia sampled a card to them to have a review ready for launch day, which is today, but didn’t give them a beta driver to actually test the card. Later in the process, they were told there actually wouldn’t be a driver at all, so instead they would need to wait until the card releases, and then just download the publicly available driver. That means that all the reviewers out there with GPUs in-hand will download the driver today, and start their testing, delaying reviews by at least several days, assuming there’s no issues in testing and so forth. That means Nvidia specifically did not want reviews to come out before people could consider buying the card, which is very odd.
TechSpot translates this “dodgy” behavior into Nvidia wanting to avoid the bad press that will likely accompany the reviews, as this card’s launch can easily be seen as a mere “price correction” for the original card.
As you recall, back in the mid-pandemic when the RTX 3080 officially launched, people were shocked at the relatively affordable $699 price point, but as we found out those cards barely existed. Nvidia itself stopped selling them on its website due to crushing demand, and offloaded that responsibility to Best Buy, which caused gamers to wear out the F5 button on their keyboards hoping to find one in stock. All of that is to say each launch since then has been met with instantaneous Out of Stock notifications, scalpers and bots buying every single card that is for sale, and a lot of pissed off gamers.
Nvidia seems to be trying to avoid that same wave of bad press this time around, or at least avoid it long enough to sell some cards in the channel before the inevitable arrives. We did not ask Nvidia for comment on this topic, but searching the web there are indeed no reviews of the card, despite it officially being launched today. Obviously Nvidia knows it will sell each and every one of them, at whatever price it dictates, regardless of any medical coverage, good or bad. Obviously, nobody will be able to find this card in stock anywhere, which was the same situation with its last launch: a 12GB version of the RTX 2060.
This leads us to gamers’ number one gripe with Nvidia right now – put your money into making more of the existing cards instead of launching newer, more expensive cards. We know Nvidia can’t just snap its fingers and triple production, however, and to its credit the company has stated publicly it is investing billions of dollars in securing supplies it needs to increase and maintain its production, but this situation will take awhile to resolve. Its CFO stated recently that help may be coming, but not until the second half of 2022.
Source From Extremetech
Author: Josh Norem