The tale of Nvidia’s 2018 could probably be told in pure pictographic form by simply putting up a 12 month snapshot of the company’s share price. It looks kind of like that classic Windows XP background: a rolling green hillock, but one with rather steep drops on either side. Yes, it’s been a trying year for team green, despite things looking unbelievably rosy for much of the year.

Jen-Hsun’s gang was selling all the old graphics cards it could make, cashing in on the fact the mining gold rush was seemingly going on indefinitely. Because of that it was holding back its next generation of GPUs, waiting to drop the new Turing generation with a resounding thud on both the professional and gaming sides of the market once nobody was buying Pascal cards anymore.

And then the bottom dropped out of the mining market, like the morning after a bad curry, arguably leaving much the same general soreness around the Santa Clara HQ. Nvidia has subsequently been left with a whole heap of old cards, and a bunch of expensive new ones that people aren’t neccessarily 100% convinced about.

So, stock-wise, technologically, and then also from a marketing standpoint, 2018 has been a tricksy ol’ 12 months for the GeForce faithful. And we haven’t even spoken of the whole GPP furore yet either…

NVIDIA 2018 REVIEW: JANUARY – MARCH
As every year in the PC market starts, 2018 was born in the glitz and faded glamour of Las Vegas at CES. All Nvidia really had to show off at the event was the brand new Big Format Gaming Displays. These were 50 or 60-inch panels, with built in Shield TV boxes, 4K G-Sync HDR capabilities, and the ability to refresh at 120Hz.

Sadly the panels seemed a little ropey when I had a chance to play with them at the show, and that was likely part of why everyone concerned has since gone back to the drawing board…

In graphics card terms this was right in the midst of the GPU mining good times, with Nvidia making some noises about doing right by gamers and making sure they could still buy graphics cards. It started talking about an ineffectual one-card-per-person limit… which retailers had already mostly implemented themselves to little or no success.

February was when we heard the first rumours of a Q3 2018 release for Nvidia’s new range of GPUs, though at that point we still didn’t know if it would be Volta-based or something else.

But people stopped caring as soon as Nvidia launched its GeForce Partner Program and AMD went on the offensive encouraging writers to question its ethics. It was suggested that some of the wording of GPP meant manufacturers, such as MSI and Asus, weren’t going to be allowed to sell AMD cards as well as keep on selling Nvidia GPUs.

There was a huge outcry from the industry and the general public, mostly because of a misinterpretation of a pretty standard marketing partnership which aimed at stopping manufacturing partners using Nvidia marketing money to promote the competition alongside its cards.

But never let facts get in the way of a good controversy. Realistically Nvidia’s partners didn’t want to have separate brands for their GeForce and Radeon cards as that would have meant spending their own money marketing the AMD brands rather than use Nvidia’s. And so they kicked up a fuss.

Away from the GPP furore, Nvidia once more tried to help funnel gaming GPUs away from miners by producing specific mining SKUs of its cards. And they just bought all those up as well.

But at least March also brought us the first peak at real-time ray tracing, with Microsoft and Nvidia announcing the DXR/RTX goodness at the Game Developers Conference. Then Nvidia went on to show how good hybrid ray tracing could look with an incredible Star Wars ray tracing demo… though it needed a $122K PC to run in real time.

However the Titan V was the only released desktop GPU capable of real time ray tracing, but it was also found to be a little dyscalculic, getting its maths wrong in actual professional scientific tasks. You could maybe forgive it getting a few rays traced incorrectly, but actual scientific maths?

NVIDIA 2018 REVIEW: APRIL – JUNE
It was very much a case of ‘out with the old and wait for the new’ in the middle of this year for Nvidia. First off it killed driver support for the old 400 and 500-series Fermi cards, and for any 32-bit operating systems, then summarily executed its unpopular GeForce Partner Program, “rather than battle misinformation” about it.

Though I kinda feel if Nvidia had been a bit more vocal about battling that misinformation it wouldn’t have come off looking so Machiavellian in its proposed partner dealings. But by that point it was too late to win back the partners or the public.

In another bid to help boost the lack of graphics cards in the market Nvidia publicised a ‘supply run’ signalling the company working to get more GPUs onto the shelves and into the PCs of gamers. Though that quickly backfired…

The cryptocurrency mining market’s demand for GPUs quickly dropped off a cliff, with Nvidia announcing that it was expecting a huge drop off in revenue from that sector, to the tune of around $190m, later in the year. And then acted surprised when that actually happened…

With reports that an unnamed Taiwanese graphics card manufacturer had sent back 300,000 unwanted GPUs to Nvidia, there was also evidence of a huge number of Pascal chips just sat in the channel. The miners didn’t want them anymore and gamers were holding out hope for a price drop as the next-gen cards were expected at any time. Nvidia tried to encourage more sales by dropping 27,648 GPUs into the Summit supercomputer, the fastest in the world, but even that wasn’t enough.

NVIDIA 2018 REVIEW: JULY – SEPTEMBER
To compound Nvidia’s growing concerns over the number of its GPUs still languishing in the channel, AMD released an independent report into driver stability suggesting the red team’s software was far more reliable than team green.

But Nvidia still resisted the urge to try and force down the pricing of its old Pascal GPUs, only trying to boost sales by packaging weak SSDs with aging graphics cards. Even after announcing the next generation of GPU, Turing, at SIGGRAPH in August prices remained artificially high for Pascal.

Though that announcement started the 20-series ball rolling, as quickly a video teaser was released which made the RTX 2080 naming scheme obvious for its new range. Then, at Gamescom in Cologne it officially unveiled the RTX 2080 along with the RTX 2080 Ti and RTX 2070, the first time a new generation was set out with three distinct GPUs.

Oh yeah, and ray tracing.

Nvidia spent much of its 20-series energies talking about real-time ray tracing, showing Battlefield V and Shadow of the Tomb Raider using the technologies, to greater or lesser effect. There was some consternation that a $1,200 GPU was only just capable of gaming at 60fps at 1080p with ray tracing enabled, and that set the tone for the new generation.

With everyone talking about the 20-series GPUs, about how expensive they were, how they weren’t really offering a huge amount of extra traditional gaming performance despite that extra cost, people missed Nvidia delaying the Big F***** Gaming Displays it first crowed about at CES in January. They’re meant to be released in 2019, though I’ve got a sneaky feeling we’ll still be missing them this time next year.

The general ill-feeling towards the 20-series cards followed when the RTX 2080 and RTX 2080 Ti were actually released in September. The RTX 2080 was a little pricier at launch than a GTX 1080 Ti, and just about beat it in standard gaming performance, while the RTX 2080 Ti was more expensive than any standard GeForce card ever released, but with higher performance.

But neither had any access to ray traced games as Microsoft was still working to get its October Update for Windows 10 and DirectX 12 ready for the public. And that wait went on, and on.

NVIDIA 2018 REVIEW: OCTOBER – DECEMBER
Finally, in October, Microsoft released its ray tracing enabling update… and pretty quickly removed it as it seemed to be randomly deleting peoples’ data. It briefly came back to life, before the company discovered some more bugs and found it had more work to do before it could have a full, uninhibited debut.

That left the Turing cards twiddling their thumbs, and meant the RTX 2070 release was a bit of a damp squib from a ray tracing point of view.

At least it wasn’t self-immolating like some Founders Edition RTX 2080 Ti cards reportedly were, however. Initially Nvidia said it didn’t see any issues with the cards after these quality concerns came to fiery light, but later released a fix aimed at solving at least some of the GPU errors.

Microsoft did eventually get its October, sorry, November Update released, which meant DICE could release its Battlefield V update and Nvidia could finally start the real-time ray tracing era for gaming in earnest. And it looks fantastic. I’m a big fan of the BF V ray tracing update, and Nvidia and DICE worked together to improve performance and fidelity with an update a few weeks later too. And they’re still going, having only scratched the surface so far. Sadly, it’s still the only ray tracing capable game available this year.

That was the only late 2018 bright spot for Nvidia as the Q3 earnings call signalled a massive drop in confidence and in its share price. From a high of $293 earlier in the year the share price is now around half that, wiping a huge amount off the value of the company. This was all blamed on the collapse of cryptocurrency mining, the fact that Nvidia had produced a huge number of GPUs which over-saturated the channel, and could find no way to get the prices down and stock shifted.

Later in December it was rumoured that one of Nvidia’s biggest shareholders, SoftBank, was set to sell off $3bn worth of stock in the new year because of it. It’s tough as all the analysts who were still screaming ‘BUY!’ even at the top of its share price are the same ones yelling ‘SELL!’ now that things are looking as tricky.

Never mind. Nvidia figured that launching the Titan RTX, a $2,500 ultra-ultra-enthusiast card was the way to salve its woes. It’s not, but fingers crossed the rumours of the first genuinely affordable (hopefully) 20-series GPU, the RTX 2060, launching early in 2019 are true, because mainstream gamers need something to look forward to.

Though there is a very good chance that ‘mainstream’ is now going to mean a $400 GPU. Happy new year indeed…

But as 2018 breathed its last, Nvidia had one more surprise for us. In using some super-smart deep learning machines the company’s researchers have been able to teach a computer to create whole new human beings. Well, pictures of humans anyway. The photorealistic images are like the deep fakes of your nightmares, creating believable people who have never existed.

It’s okay though, because the AI is unable to create perfect replicas of cats. So, in the future, if you need to make sure an image is real you need to make sure that it has an accurate image of a cat in it. Not the nightmarish beings the AI spits out. Seriously, click on the image and check out the full-scale picture. It’s seriously creepy.

Well, that’s been 2018, so what does 2019 hold for us and Nvidia? Hopefully less of the controversial, fewer GPU explosions, no more terrifying cats, and just more great graphics cards. We’re expecting the RTX 2060 to drop early in the new year, and the actual mainstream Turing GPU to land a little later in the year, potentially when AMD’s Navi arrives.

Beyond that… those Big F***** Gaming Displays might finally appear, or be nixxed for good, and the mobile versions of the 20-series GPUs will arrive in a whole slew of laptops too.

And there will be surprises, we’re sure of that. But just what form they’ll take right now is anyone’s guess.