It's interesting to see what Huawei is doing with their Harmony devices. I thought the Pura X was awesome as it unfolds into the size of a Kindle which is great for reading and such. Sadly their banned in the States so no one sees the need to try to compete.
Listen to what they've said about the future of AI and games. It would be very stupid for them to abandon the consumer segment and gift the market of gamers and other plebs to their competitors.
I think it’s more of a risk analysis thing that they haven’t completely dropped gaming as part of their business. They don’t really know how long this AI craze will go or if it’s just a temporary fad that will see demand for all this machine learning tech drop over time.
I think it's part that and part not wanting to make the mistake of giving up their "terminals" business and go all in on the lucrative "mainframes" because 10-20 years from now that might turn out to have been a big mistake. AI is here to stay in whatever forms it takes.
It used to be the case that every generation the 60 series would be around the same performance as the last gen 70 series. Now it's every 2 generations huh?
The article mentions an expected gaming improvement of around 5%. That puts it still behind the 4090 correct? I believe the 5080 was on average around 10-15% slower than the 4090. I'm curious what price Nvidia will launch that card. It's an 80 class that will likely be slightly slower than the last gen 90 series. Are they seriously going to try and price an 80 class card at over $1,500?
The last time NVIDIA released two successive generations on the same process node was the 7 series and the 9 series, and in that case, the GTX 960 was significantly less powerful than the GTX 770, with about half the memory bandwidth, 3/4 the compute, and half the texture fill.
I forgot a lot. Glancing at Tom's, 4090 was 20% ahead of 5080 in 4K raster back in January. So overclocking and adding some memory will not do much to bridge that gap. Months before launch, I theorized that the 5080 was intentionally nerfed (half the cores and bus width of 5090, little change from 4080/Super) to keep it sellable in China, but then the 5090D was conceived... only to face order cancellations and possible additional nerfing recently:
Spec leak before Computex presentation in a few hours. Not much new here but PCIe x16 is nice to have on the lower die. You could probably get away with sticking it in a PCIe 3.0 system. Will certainly be cut to x8 if they make a 9050/9040.
The only advantages it really has is that you theoretically can by one brand new, the GDDR7 memory, and the questionable value of MFG.
Good thing I didn't say they would. But AMD still hasn't been able to really pressure them to try so they feel like they can just sit on their laurels and not lose market share. Who knows, maybe after this whole 50-series disaster and AMD celebrating their 9070 sales we could see a strong comeback in the 60-series.
I don't think AMD can get a clean kill against the 5060 Ti 16 GB, but they can easily undercut the 5060 Ti 8 GB and outperform the 5060 8 GB with their 8 GB model. (Whatever excitement they generate has to be accompanied by a good supply to change anything, obviously.)
. That's probably what the 9060 XT 16 GB price will rise to after they give it a $350-360 MSRP. No pricing leak yet, but if they announce it tonight we could know within the next 3 hours.
Why the fuck are half the comments under this article about Trump and egg prices?
Wccfkek users love to banter on whatever the popular article of the moment is. They bully each other for being poor/rigless/brown/whatever too, it's pretty funny.
can you at least point in the direction of things that there are potential anonymous sources for?
16GB and FSR4 for $350 is great value. Most models are going to be $50-100 more than that. If there are supply issues they’ll also be scalped to hell and back. So hopefully AMD at least has decent supply ready to ship.
I'm excited to see how it performs. It has exactly half the cores of a 9070XT, but no idea what that means for guesstimating the performance.
