If they can use their boost from Ryzen sales to bolster their graphics division, they'll be in a great position.
Yesterday I was talking with some coworkers about a new computer build and one of them had recently purchased Ryzen and the other was thinking about it.
So at least anecdotally, AMD does indeed appear to be on fire.
Yeah, I'm in the same boat as you, however I want to add a second graphics card since my monitor is 4k and my motherboard only has one full speed PCI express slot.
The only thing keeping me from pulling the trigger is the obscene memory pricing right now.
I deliberately delayed building my current desktop PC until Zen CPUs became available because I wanted to see if it was a reasonable alternative to Intel. And it was totally worth it. I got me a Ryzen 1700, so I have 8 cores, but at 65W it is not excessively energy-hungry. It purrs like a kitten and runs like cheetah.
Ryzen has indeed seized AMD a lot of mindshare among people who build their own rigs. The problem is that people who build their own rigs are a vanishingly small slice of the overall population of people who buy computers.
Really? From my perspective the amount of people building their own rigs has stayed the same. All my friends that game on PC haven't slowed down buying parts/upgrading their rig/building a brand new one.
However, casual PC users seemed to have dropped off the face of the planet, that entire market segment has shifted to laptops and tablets. So much so that the big players (Dell, HP, etc.) have severely cut back consumer PCs and focused on business workstations.
I'd be interested to see some real numbers on this!
AMD has taken the lead in performance per watt??? Do you have a source? Everything I've seen shows Intel is either matched or better. If you also take capitol cost into account, maybe AMD is in the lead; I don't really know.
Either way, Intel is soon to release the first 10nm process product, which should provide a significant boost, as long as the architecture can take advantage of it.
You mean the 10nm that has been delayed for 6 years now? That has so bad yields Intel has internally just renamed 12nm to 10nm and use that? The 10nm that has 50% yields while being 12nm THAT 10nm?
Not sure where you heard that Intel renamed 12nm to 10nm, but you're probably thinking of another fab. Intel's 10nm cell density is comparable to Global Foundry's 7nm [1]. And yes, Intel's 10nm process is very late; that doesn't mean it's not coming.
As far as yields go, they're constantly improving, and they depend on the processor, so your claim that the yield is 50% is either ignorant or intentionally misleading.
I checked out your link, but all I saw was price per performance numbers, not performance per watt. Price per performance is an important metric for gamers, but not so much for servers, which is Intel's bread and butter. In the server space, performance per watt is what matters. If you have some evidence that AMD has taken the lead in performance per watt, I would love to see it.
The 10nm parts due for Q4/19 (!) are 12nm more than 10nm.
The report also mentions the real 10nm yields, 8-10%, targets at 50-60%. I admit that I'm not sure what the current yield on their 12nm is but I would guess it would have to be the target.
Eh, I don't think it's capitalism per se, but more a direct consequence of the state of the art of the technology.
Moore's law is coming to an end, Intel institutionally has always relied on their process advantages to stay ahead. But now it's only expected that others are catching up as we see that moore's law is really an s curve. Intel will get another bump with 10nm. They're taking on EUV 'early' where no one else is (which is what's taking so long) and it'll probably mean that they have a head start there once they figure it out, but that'll be one of the last bumps for them. BEUV is sort of intrinsically flawed.
AMD has been undervalued for years due to short sellers. They release a better than expected earnings report and the stock goes down 10 percent. It's always been the whipping boy on Wall Street. But a great stock to profit from on a very cyclical basis when those shorts go nowhere.
I am secretly hoping that at least for Desktop Mac, Apple could switch to AMD. They could leave the Mac Pro and iMac Pro for Xeon ( Or they could choose AMD and that will be EPYC ) But right now it is very hard to argue for purchasing Intel when AMD offer many more cores for lower price at the expense of ~20% peak single threaded performance.
Unless Intel offer Apple very good deal but I thought part of the Intel / AMD agreement was Intel can not use any tactics to undercut AMD with incentives or rebate or something similar.
At the same time Intel is like assembling old Apple alumni from CPU to GPU.
My ideal would be if AMD gives an HDL license to Apple, and they make semi custom chips with the ex-Intrinsity guys going to town with power micro optimizations making a nice AMD mobile chip. It'd be kind of the best of both worlds for Apple with an x86 mobile chip that they control. And it'd be a nice way for AMD to differentiate itself as I'm sure that Intel would never in a million years give full HDL to Apple.
Glad to see this. Used AMD back about 10 years ago, but last two builds I've done were intel given they couldn't keep up with performance or price of intel. I have stuck with ATI/AMD cards though since they have seemed to stay competitive with nvidia and timing wise they seems to be the best bang for the buck.
25 comments
[ 3.0 ms ] story [ 76.3 ms ] threadI mean, the good form of "killing it".
Yesterday I was talking with some coworkers about a new computer build and one of them had recently purchased Ryzen and the other was thinking about it.
So at least anecdotally, AMD does indeed appear to be on fire.
The only thing keeping me from pulling the trigger is the obscene memory pricing right now.
However, casual PC users seemed to have dropped off the face of the planet, that entire market segment has shifted to laptops and tablets. So much so that the big players (Dell, HP, etc.) have severely cut back consumer PCs and focused on business workstations.
I'd be interested to see some real numbers on this!
Intel has been 'phoning it in' for nearly a decade now, because their products had a significant lead over AMD, particularly in performance-per-watt.
Now AMD has taken the lead, and it's going to take years for Intel to catch up. (Because CPU innovation moves at a glacial pace.)
The winners are the consumers.
Either way, Intel is soon to release the first 10nm process product, which should provide a significant boost, as long as the architecture can take advantage of it.
For Performance per Watt you can check this article: http://www.pcgameshardware.de/CPU-CPU-154106/Tests/CPU-Test-...
The current Intel top dog against AMD in terms of perf per watt is the Core i3 8300.
As far as yields go, they're constantly improving, and they depend on the processor, so your claim that the yield is 50% is either ignorant or intentionally misleading.
I checked out your link, but all I saw was price per performance numbers, not performance per watt. Price per performance is an important metric for gamers, but not so much for servers, which is Intel's bread and butter. In the server space, performance per watt is what matters. If you have some evidence that AMD has taken the lead in performance per watt, I would love to see it.
[1] https://www.semiwiki.com/forum/content/7191-iedm-2017-intel-...
That was Q4/17. Latest information: https://www.semiaccurate.com/2018/08/02/intel-guts-10nm-to-g...
The 10nm parts due for Q4/19 (!) are 12nm more than 10nm.
The report also mentions the real 10nm yields, 8-10%, targets at 50-60%. I admit that I'm not sure what the current yield on their 12nm is but I would guess it would have to be the target.
Moore's law is coming to an end, Intel institutionally has always relied on their process advantages to stay ahead. But now it's only expected that others are catching up as we see that moore's law is really an s curve. Intel will get another bump with 10nm. They're taking on EUV 'early' where no one else is (which is what's taking so long) and it'll probably mean that they have a head start there once they figure it out, but that'll be one of the last bumps for them. BEUV is sort of intrinsically flawed.
Unless Intel offer Apple very good deal but I thought part of the Intel / AMD agreement was Intel can not use any tactics to undercut AMD with incentives or rebate or something similar.
At the same time Intel is like assembling old Apple alumni from CPU to GPU.