Radeon "RADV" Vulkan Driver Adds Experimental Support For Sienna Cichlid

[TemplarGR]

It used to be that transitioning to a denser node resulting in cost savings. That changed around the 28 nm mark. This means that it's cheaper to keep on manufacturing older GPUs on older nodes, with reduced prices, than making new GPUs on more advanced nodes which are targeted at the lower end of the market.

This is not true and it is extremely easy to debunk, almost effortless. Low tier CPUs @7nm=exist.

Yes, newer nodes are -initially- expensive, but on the other hand you get more transistors per wafer. So it relatively evens out. All they have to do to support the lower end of the market is release fewer core/CU count designs, it is as simple as that. The "it is expensive and not worth it" excuse, is just that, an excuse, because they made a metric crapton of 14nm gpus for the cryptocurrency market and regulated the supply in order to keep prices sky high, and when that bubble got burst they needed to empty the inventory, so they decided to "service" the low end by selling us the remaining stock. That is the truth, that is why they didn't deliver a low end variant of RDNA. I understand that AMD is a business and need to make money, but i don't have to like it. I refuse to buy a 5x0 gpu because although it is satisfying for the amount of money i am willing to spend (i am not a heavy AAA gamer), i just don't like the power consumption and noise, plus the PSU requirements (which bring more noise, size, cost and power consumption). That is why i have been waiting for RDNA2 in the hopes of getting a proper mainstream variant that won't keep the other people in the house awake when i game at nightime. I have a feeling AMD will deliver such variants, after the Bigger parts hit. At some point there won't be any Polaris to sell. LOL.

[skeevy420]

What the hell I just bought an RX 570 6 months ago.

As much as I like my 580, and mind you that I'm perfectly happy with my 580 in regards to performance and Linux support, I'd like something that performs about the same or slightly better, with newer tech and features, and doesn't draw as much power or use as much heat. I also made a mistake and bought a 4GB model since it was only $125. Most of the time, damn near all the time, 4GB is good enough, but it isn't on some newer games with certain settings set higher than medium or high.

Vega didn't have that GPU. Navi didn't have that GPU. That's two generations without a mid-range desktop GPU. Big Navi/Navi 2 need mid-range and low end tiered GPUs. A both downside and upside to that is they can wrap up the GCN generations and focus on the RDNA generations. Depends on who you are and what you can afford if it's an up or down side.

[pal666]

AMD's Day 1 (Linux) GPU support has always been an iffy at best prospect, especially for an informed consumer. All these articles lately are making me cautiously optimistic for SC.

afair polaris was sorted out in few weeks after release, but this indeed looks promising to become best day 1 so far

I just hope this generation finally has an RX 580 replacement in the $150-$250 range.

my 580 isn't that old to demand refresh, i'm eagerly waiting for big navi. by latest rumors large cards will come first, rest will follow in next year

[TemplarGR]

Lower power consumption can be achieved by lowering the maximum GPU clock or by restricting the set of permitted GPU clock states via files in /sys/class/drm/card0.

So, you are telling me i am supposed to buy a higher TDP card and gimp it, instead of the vendor actually supplying a proper chip? When you are buying, the vendor is expecting to be paid for the max performance he is delivering to you, not for the gimped one....