Originally Posted by delboy67
I dont agree with much you say, I'm just not one for bashing people though, I detect emotion in your posts and bias
some disagreed with me as well when I said it was a pitcairn replacement and would be priced that way (which I said at the start). btw In your link you say 150w and between 390 and 390x not 120 and 390x, plus you cant even say thats right now without real benches/reviews/drivers! My prediction for polaris is, there's still some small surprises to come, in fact, there has to be as leaks have suggested all gains come from the shrink rather than arch improvements and we know there has been lots of changes so surely some improvement, I bet there will be parts of games where old gcn struggled but where polaris will shine and much have better minimums. Maybe not 1070 killing review winning things but I dont think this will be a simple 390x at 120w. I just hope amd havent gone down the 'business is war' road with their pricing, it didnt work out too well for the only company I'd ever be emotional or biased toward!
Looks like the least optimistics of my ranges came true
Originally Posted by magnek
If they made a 500mm² chip and it didn't beat an OC'd 1080, we won't hear the end of it, and they'd have to price it cheap with next to zero margin again.
If they made a 500mm² chip and it blew away an OC'd 1080, it will be priced accordingly and we won't hear the end of it. That is until nVidia discounts the 1080, then that will become the best buy until 1080 Ti comes out.
If they made a 500mm² chip and it traded blows with an OC'd 1080, 75% of the population would still go with the 1080 and recommend it to their friends anyway.
Conclusion: The dumbest thing AMD could do right now is to release a 500mm² and "take nVidia by surprise".
I agree, i have been liking your more pragmatic posts.
To add on to this, 500mm2 die are the highest risk chips there are. Not only are they the most R and D heavy, they have the highest potential losses because of wafer yields. They only make sense if you have the professional market where you have 90% gross margins to make up for all the discarded dies. Or at the end of a node's lifecycle when the yields are as good as they are going to get and the namesake to have a crap load of sales.
Fury X I have said from the beginning was a waste of money and a vanity project that made them squander their last chance to capture marketshare at 28nm. To successfully build monolithic chips, you need to have a secondary market to make it worth it. For Nvidia, this is their 1.6 billion annually professional and data center market which is pretty steady and growing.
When you have limited resources, you go for the safest bets which are the most likely to yield the greatest rewards.
Right now, AMD doesn't have the resources to compete in this arena, particularly now that Nvidia's made a 100% single focused chip unlike their dual purposed high end of the past. As is AMD doesn't have much of a chance in the professional market. Atleast the super computing and 4000+ market. This is particularly because Intel has stepped up their game and a three way battle between Intel, Nvidia and themselves is not something they want to be a part of since both of these companies have put AMD in dire straights just taking one of them on at a time. AMD wants to run, run as fast as they can from this battle.