Yep and not only do you need to make the money back spent on all of the above you also need to make enough to keep moving forward for the next few generations.
Yep and not only do you need to make the money back spent on all of the above you also need to make enough to keep moving forward for the next few generations.The equipment, and future equipment, and the R&D involved needs to be funded.
Costs include many things allowing these fabs to keep afloat and on the bleeding edge.
Not sure if it was the image used for the calculation but it was the only image I saw here, and when I first saw it it was definitely not impressive with the math some people churned out:
Basically a door, some shadows and one enemy.
AMD pays for wafers from Global Foundries/TSMC/Samsung, and as such pays for the privilege of having their chips on a bleeding edge node. The wafer costs aren't just silicon, they're a fraction of that. Most of the wafer cost is there to cover R&D and equipment for these companies.What equipment are you guys talking about? AMD doesn't manufacture cards. The the AIB partners make the cards and the fabs take care of the rest. So AMD has to pay for the equipment so the AIB's and the fabs have the privilege of charging AMD to make their products? Doesn't sound logical.
Just saw this on reddit:
https://youtu.be/A9Dpb9j6E8U?t=256
You can see even an OC'd (2000+) 1080 Ti drops under 60 fps when staring at that door without an enemy and down to 51-52 fps during the main part of the fight.
A 1080 Ti normally runs @ 1800ish and often goes under 60 fps: https://youtu.be/KTnJvEL4RzA?t=1097
So that's about an 11% OC and pretty much max for a 1080 Ti and even then it drops to 51-53 fps during the fight (and with a shotgun not the GLOO gun causing more geo as seen in the AMD presentation)
So yeah... still a terrible presentation from AMD, but at least we now know that even a max OC 1080 Ti drops to low 50s in that same area.
They should just stop giving presentations. They suck at it.Still got to factor in CFX Vega. Depending on what math set you use (60hz projector or 30hz projector).
I really wish AMD would take a second to look at their presentation info before presenting it.
Oh well, the wait for July/August is real. Regardless the outcome, it has been a crazy ride!
AMD presentations are only as good as what they have to show. The Ryzen launch was they best they've had in years precisely because there was product and concrete details and demonstrations.
They need to get better at managing expectations as a lot of people were led to believe that there would be more about RX Vega at Computex but instead people were massively blue-balled. If there won't be anything for a while, just let people know. It may be disappointing, but far less so than when people keep building up expectations.
I told people not to expect news at Computex about RX Vega. They didn't listen. The hype train reached moon orbit with people expecting a reveal and pricing at Computex.
Reminder:
Radeon 3xx series launch: June 18, 2015
Radeon Fury X launch: June 24, 2015
Polaris RX4xx series launch: June 29, 2016
Irritating though it may be, a July launch for RX Vega is fine with me... even better if it is priced right, and performs within 10-15% of a 1080 Ti. I will purchase at least two Vega cards to support competition, even if it doesn't dethrone the 1080 Ti for single-card 4K gaming.
Perhaps because Raja specifically said that they would be "showing off" Vega at Computex.
It's AMD's problem that they poorly manage expectations.They did, though.
The fact that Vega is launching in August means that the Vega generation will be competing with Volta for the majority of its lifetime. I think we will see GV104 in Q4 2017 or early 2018. If Vega 10 loses to GP102 then it would be a mighty embarassment as GV104 will beat GP102 for lesser power. Nvidia is achieving all this with GDDR5X and next gen GDDR6 which are manufacturable in high volume, good yields and reasonable cost while AMD is having to resort to HBM2 which is higher cost, lower yields and much lower volume. GV102 will be another 35-40% faster than GV104 and close to 50% faster than GP102. If Vega fails some heads will roll at RTG. AMD cannot expect to even maintain 20% market share in 2018 if GV104 beats Vega 10. The way I see it there are only 2 ways -
1. Vega is better than Pascal and can remain competitive with vast majority of Volta product stack with some price cuts and refreshes on 14nm+.
2. Vega is worse than Pascal and AMD's GPU market share goes to their lowest ever in 2018 (probably around 15%). In this case drastic decisions would need to be taken to overcome the multi generational performance lead that Nvidia would sport.
Agreed.
People who say "so what if Vega is 10% slower than the 1080ti? It can be cheaper!" are missing the point. The problem is that, according to the current roadmap, a refreshed Vega is all AMD has in 2018 as well. Thus, Vega MUST be able to at least somewhat compete with Volta, which means it must be a good deal faster than GP102.
Sadly, AMD is not inspiring confidence in me with the way they act when it comes to Vega, and the late launch. I think of how they showed Ryzen exchanging blows with the 6900k months before launch, and then I look at Vega with shoddy CrossFire benchmarks and vague statements about its performance. As you say, I think 2018 may be a disastrous year for Radeon and that we won't see it bounce back to finally being competitive across the whole product stack until Navi+1 or Navi+2...
That's why I'm pissed about the delay, because if it's delayed and slower... and I waited THIS long for the chip, why not wait for a more power efficient chip? A delay is the worst thing AMD could have done for the gaming VEga, because at this point, if Nvidia did release news that a new release was coming soon, I'd buy a GTX 2070. Then sell it and get a GTX 2080Ti. Or maybe it'll be 1170 and 1180Ti.
AMD should probably work with their monitor partners on freesync and getting more news about freesync monitors out, and in general just doing their job.
Main reasons I'm excited about Nvidia are 4K 144hz GTX Volta/1080Ti.
AMD should just tell me that a 4K 144hz freesync monitor is coming.... Go do your job and promote your company and make your whole featurestack look attractive.
Honestly, that's about the only sane strategy atm. They have released in cadence with Nvidia for years only to be outdone. Release cards in between Nvidia generations. We get Vega in 2 months. We get Volta 6-8 months after that. We get a Vega refresh 8ish months after that. Nvidia's post volta after that, then navi and so on. No more head to head. Keep selling the cutting edge people new cards every 8-12 months from both sides.Well we could think of it this way. AMD decided to skip high end last gen and target Volta instead. So relatively speaking they're releasing before Nvidia instead of really late.
Mostly a post in jest
That's why I'm pissed about the delay, because if it's delayed and slower... and I waited THIS long for the chip, why not wait for a more power efficient chip? A delay is the worst thing AMD could have done for the gaming VEga, because at this point, if Nvidia did release news that a new release was coming soon, I'd buy a GTX 2070. Then sell it and get a GTX 2080Ti. Or maybe it'll be 1170 and 1180Ti.
AMD should probably work with their monitor partners on freesync and getting more news about freesync monitors out, and in general just doing their job.
Main reasons I'm excited about Nvidia are 4K 144hz GTX Volta/1080Ti.
AMD should just tell me that a 4K 144hz freesync monitor is coming.... Go do your job and promote your company and make your whole featurestack look attractive.
Vega is FreeSync 2