Originally Posted by SpeedyVT
AMD is doing great and people should be buying them on the merit of an alternative big company.
They are not doing great, their current CPU offerings are atrocious, unless you are interested in buying super cheapo budget builds. Otherwise Intel's offerings across the board are far superior at every price point. Then on the GPU side the only products they have that are relevant are the 380, 380X, and 390 everything else is outperformed by NVIDIA or has a better overall value when overclocking is included in the potential of the product. I'm sure there will be people that will argue this concept, but this isn't "Stockclock.net" so I couldn't care less about default clockspeed benchmark comparisons regardless of resolution because it is meaningless when overclocking comes into play (eg. "buh the Fury X is 5% bettah than the 980 Ti at 4K brah").
Originally Posted by coupe
Amd has a good track record on anything but bulldozer
, and we know why that is.
"Overclocker's Dream" ring a bell?
Originally Posted by PiOfPie
If it does actually reach Haswell levels, I'd say it'll recapture maybe 5-10% of the desktop market, mostly from two sources.
1) "Moderate" AMD fans like myself who either bought Intel for a generation or two or have held off buying new CPUs since Vishera. Skylake's performance uplift over Haswell is within the margin of error for many applications and regresses in some areas, so unless Kaby Lake or Cannonlake bring huge improvements, being ~10-15% behind in IPC isn't a huge liability, especially if priced appropriately. It'll be like the Core 2/Nehalem vs. Phenom II days.
2) Zen-based APUs, while coming a bit later, may also be a nice play for HTPCs and all-in-ones that casual consumers make use of. However, AMD may be screwed over by OEMs and Intel's rebates like they usually are.
Being backed into a corner brings focus. I think there's an understanding throughout the organization right now that it's either produce or perish. Lisa Su's an engineer instead of a marketing guy like RR was, and she's made pretty good decisions so far: narrowing the strategic focus, splitting off RTG, granting the Zen teams more autonomy in design, so forth.
Skylake has shown to have a 10% increase in IPC clock for clock
over Haswell, and 11% in Multi-threaded performance:
Kaby Lake is only going to be a refresh of Skylake, therefore it will only bring higher clockspeeds like we saw from the 4770k to 4790k (6700k to 6790k). However, regarding stock benchmarks and overclocking this could increase the percentage difference even further. Then of course Cannonlake will bring another 10% increase in IPC on top of that. This means, the only strategy AMD can have is offering more cores for less money compared to Intel's offerings (and again at Haswell levels of IPC). This game plan can without a doubt in my mind help them retake some of the market back, I project even more than 5-10% if this is actually what happens.
I only see Zen-based APUs being a saving grace, if they are reintroduced in the mobile market, since as it stands today, that market is currently dominated by Intel and Intel's on-die graphics are getting quite impressive.Edited by BiG StroOnZ - 1/23/16 at 9:22pm