I find it amusing how this nonsense morphed. First it was the scheduler, then it was Async, now apparently Nvidia emulates all of DX12. Pretty soon we're going to find out that Nvidia emulates our reality as well. We're just floating in a giant Nvidia K80 that collapsed into a singularity, all rendered entirely in software.

Plausible

Plausible

I was wondering about something like this... As far as Vega Desktop, last I heard there were going to be four GPUs.

This forum turned into top lol On topic: I'm not going to wait for Vega, it's going to be 1080 price point so I can't afford it. I'm looking for a buyer for my R9 270x and then I'll get a RX570.

tbh i will be happy if Vega lands somewhere between a 1080 and a 1080ti in terms of performance and power consumption, but at a lower price point than the competition.

No wonder nvidia is already preparing Volta.

Why should Nvidia care about vega? Really? Both Titan XP and 1080ti will be faster than Vega.

I find it amusing how this nonsense morphed. First it was the scheduler, then it was Async, now apparently Nvidia emulates all of DX12. Pretty soon we're going to find out that Nvidia emulates our reality as well. We're just floating in a giant Nvidia K80 that collapsed into a singularity, all rendered entirely in software.

It does seem to have issues with drawcall submissions from multiple threads though. Which would make sense, given the internal structure. The scheduler remains the core strength and core weakness of the architecture.

Why should Nvidia care about vega? Really? Both Titan XP and 1080ti will be faster than Vega.

Please share the leak review performance specs

It does seem to have issues with drawcall submissions from multiple threads though. Which would make sense, given the internal structure. The scheduler remains the core strength and core weakness of the architecture.

Yeah I know. I just think it's funny how we've had a 6 month long discussion on this forum with tons of details and insight into the strengths and weakness of both architectures only to have people boil the entire thing down to "Nvidia emulates DX12".

Good news that it is still on track. Hopefully the performance is going to justify the wait.

Good news that it is still on track. Hopefully the performance is going to justify the wait.

Usually being late is not a good sign. And to be honest I don't believe anything AMD says until I see proof from trustworthy third party as Hilbert. There are also other things to consider like game tech and driver software. That side has certainly not been strong points of AMD either. This being said, I really hope there will be more competition this time around, as NVidia has been able to ask a bit too much money from their products. Not because they are jerks but because they have been too much ahead.

This forum turned into top lol On topic: I'm not going to wait for Vega, it's going to be 1080 price point so I can't afford it. I'm looking for a buyer for my R9 270x and then I'll get a RX570.

Consider a RX 480 instead ... if you stick with 4GB RX480 you'll get more performance for the same price. If you go 8GB RX480 you pay 10-20 more without running into any annoying VRAM limit (yes also games in 1080p and 1440p can have ultra big texture sets) AND you can OC to 580 levels in both cases, even more safely so if you do a reapplication of thermal compound actually you may even get a used 480 for 120-150 by now

Usually being late is not a good sign. And to be honest I don't believe anything AMD says until I see proof from trustworthy third party as Hilbert. There are also other things to consider like game tech and driver software. That side has certainly not been strong points of AMD either. This being said, I really hope there will be more competition this time around, as NVidia has been able to ask a bit too much money from their products. Not because they are jerks but because they have been too much ahead.

I tentatively agree, the last time AMD, or rather, ATi were late with a gpu design, the 2900XT was released and while it wasn't bad... it wasn't good either but it did provide the foundation for the VLIW Terrascale line of hardware that became very successful. The other thing is, remember how much hype that card got? So much, people were calling it the DX10 nVidia killer, that it was late because of how good it was going to be and dribbles of information kept on getting revealed such as it's innovative use of the memory ring bus, the shader clusters and it's incredible parallel design. This time on the other hand, we've only been given a few facts about Vega. It's name HBM2 Major change in architecture - either GCN5 or NCU Tile based rendering Memory controller able to access other parts of system memory other than it's own Possibly have ~12Tflops of computational power (~24Tflops in half-precision) That it's coming 2017 Q2 - possibly revealed at computex Demonstrations at previous computer events of engineering samples with diagnostic boards rendering a limited number of games (Doom, Battlefield 1, Battlefront) at a minimum of 60fps in 4K And this is all the information we have, the rest is pure speculation, such as basing performance of Vega, by scaling it against Fiji and applying the efficiencies of polaris. All we can really do is wait and there isn't really any point speculating because we don't know how good AMDs implementation of those new features will be and wont be found out until tested by reliable and unbiased sources.

remember how much hype that card got?

Maybe AMD is applying a different marketing strategy this time to prevent over-hyping and as a result of that disappointment.

Consider a RX 480 instead ... if you stick with 4GB RX480 you'll get more performance for the same price. If you go 8GB RX480 you pay 10-20 more without running into any annoying VRAM limit (yes also games in 1080p and 1440p can have ultra big texture sets) AND you can OC to 580 levels in both cases, even more safely so if you do a reapplication of thermal compound actually you may even get a used 480 for 120-150 by now

Already suggested the same thing to him. He is dead set on buying a 8 month old card with a new name.

Vega to be a dud and nvidia smash them again as alway amd to slow to release anything good

Usually being late is not a good sign...

Yep. Last time AMD was this late and tight lipped about a GPU release was the HD2900, and we know how that turned out. Last time Nvidia was so late with a card release was Fermi (GTX480), and we know how that turned out (the card to replace your egg skillet).

Yep. Last time AMD was this late and tight lipped about a GPU release was the HD2900, and we know how that turned out. Last time Nvidia was so late with a card release was Fermi (GTX480), and we know how that turned out (the card to replace your egg skillet).

So but that Volta should be horrible or something. Indeed Vega was delayed from Q1 to Q2.

So but that Volta should be horrible or something. Indeed Vega was delayed from Q1 to Q2.

Question is if it is due to fine driver tuning or building sufficient volume as they expect shopping spree.

Question is if it is due to fine driver tuning or building sufficient volume as they expect shopping spree.

Considering they were using debug PCB's in Q1 neither of these options look to be the truth.

Considering they were using debug PCB's in Q1 neither of these options look to be the truth.

Lets be honest. Both options sound hilarious. I honestly can't tell which one is funnier: a) Shooping spree, or b) AMD doing "fine tuning" on the unreleased product. The product which is supposed to be based on the brand new arch. When we full know 6 months from release they were still doing basics like occupying shaders on the much better known arch (Fiji) And that's the whole idea of - getting better with the time. Ie not being able to optimize the product right from he launch. Fine tuning lmao

Question is if it is due to fine driver tuning or building sufficient volume as they expect shopping spree.

😀