The relationship between AMD model numbers and actual CPU generation is getting increasingly confusing. You have these which are 3000-series but 14nm Zen, and meanwhile they've started selling 12nm Zen+ chips as Ryzen 1600s on the desktop.
Indeed. And AMD's counterpart to ark.intel.com is horrendously opaque. I don't mind having confusing part numbers, as long as I have a simple way to get clear information about the part.
I think these chips go in really low end computers like possibly chromebooks or sub 300$ laptops where intel falls short on pricing. But from what I can understand AMD is already selling 300$ 3200U laptops which maybe perform similar to these.
Nonetheless I find it really amazing that only 3 years ago 2c/4t was normal for high end flagship laptops like xps 13 with 7200u and now those chips are lowest end chips being offered.
All tribalism apart, that's all AMDs merit. If it was up to Intel we would still be getting the same old overpriced dual core crap 3-5% faster than their previous one.
quad-core workstations have been available for years, I literally have one from 2010 and I doubt it was the first even then. There were probably Core2 Quad or Phenom X4 workstations.
AMD had their own quad-core laptops available for a long time too, the Excavator/Piledriver series goes back many years as well.
Cheap dual-cores have been around forever as well, this is not something that AMD has brought on with Zen. This is not really a distinctive product in most ways, it's just a replacement for those old cheapo Excavator products for bargain-basement laptops. So I'm not sure what you're trying to say here other than 'AMD good'.
dual-cores have always been the standard for ultrabooks, but those laptops are literally defined as eschewing performance in favor of making it thinner and lighter. You make a processor that pulls half the power, the workstations will take twice the cores and the ultrabooks will slice off another 2mm of thickness and stay with dual cores. Zen did not change this dynamic.
Yes, Zen2 is a good product, this is not a Zen2 chip, nor is it really notably higher core count or anything else. It's a replacement for A8 piledrivers.
>Nonetheless I find it really amazing that only 3 years ago 2c/4t was normal for high end flagship laptops like xps 13 with 7200u and now those chips are lowest end chips being offered.
MacBook Pro has been shipping Quad Core SandyBridge since Early 2011. Along with many others in the PC world.
At the low end, in Chromebooks, I think Arm processors are a better way to go. These days Chromebooks run Android apps and Android on x86 hasn't been great.
Gemini Lake is really good in the $250 to $350 segment. They have great hardware video decoding/encoding and the general CPU performance is between Core2 and Nehalem level, while offering Chromebook level battery life.
There are Chinese laptops like the Chuwui Lapbook Pro that get you a 1080p IPS screen, a quad-core Gemini Lake, and 8G of RAM for $320. I'd rather do that and be able to run standard x86 software at semi-reasonable speed than to mess with a chromebook. (I can see the draw for a momputer that you don't want to mess with though)
I'd love something in the 7nm or 10nm class, of course. If Dali was on Zen2 on 7nm it would be fantastic. It's just not possible yet in this price range. AMD is still launching 14nm in this segment, not even Zen+. Next-gen Atom (Skyhawk Lake) is going to be on Intel 14nm as well.
I'm curious how Dali/Barred Kestrel does on the extreme battery life/chromebook thing though. Raven Ridge (and the Zen+ successor, Picasso) did not have great idle power and this hurt it there. If they could get the idle power down, it would be a good alternative to Gemini Lake.
In many segments, low-end included, the processor is a small part of the total BoM. Display, memory, clamshell, keyboard, assembly and shipping are non-negligible.
One of the fun things about ARM is that you get to play with asymmetric multicore machines that don't exist anywhere else in the desktop space. At least for now.
For me, the worst flaw of Chromebooks is the keyboard and the lack of proper Control, Super and Meta keys.
Is this what's driving AMD to bang up against the 50.00 ceiling this morning? I'd be surprised if it's just the Mizuho rating on Intel's potential 10nm server chip delays.
No, this is just AMD doing the bare minimum to meet contractual obligations with Global Foundries. None of their other product lines make sense to fab on GF’s older process.
I suppose the current generation consoles are still on 12nm, but that must be ramping down at this point. PS5 and Xbox-something are a going to be big chunk of the 7nm capacity (30,000 wafers/mo) AMD took over from Apple, since they’re moving to TSMC’s next node for the next iPhone.
AdoredTV recently did a rough breakdown of how many wafers it’ll take for each product line. Epyc only needs about 1,500 wafers from TSMC since the chiplets are so high yield. Desktop and the new high end laptop parts don’t need much more. What boggles my mind is that between the new consoles they’ll need something like 10,000 wafers per month if this generation’s sales are on par with last generation. I guess it’s a nice way for TSMC to keep 7nm production up as Apple, AMD, and the other big ARM players move onto better nodes.
Edit: the consoles take so many wafers because they’re going to be giant monolithic APU designs somewhere in the 400mm^2 range.
Ramping down, sure, but they'll keep going a long time for production for the developing world. Consoles have extremely long lifespans in places like South America, India, and Africa, where $600 might be your entire yearly income. You're not going to race out and buy a PS5 on that.
It can be somewhat surprising but PS3 production was only discontinued in 2017, Xbox 360 production was discontinued in 2016.
Really interestingly, the IO die and the X570 chipset are both the same piece of silicon running different firmware. That's creative and an interesting vector to increase your yields.