Desktop volume down 16%, Notebook up 16% but ASP down 7%.
Q4 forecast is bad too, they are expecting 14% decline in revenue compared to Q4 last year.
You do know, that Moore's law, was not about technological limits, but rather, economic in nature, right? It is/was an economic law for the industry.The bolded doesn't make any sense. Physical limits are the absolutes we can reach in our universe and no amount of economic resources even if infinite in theory will push us past those limits ...
You do know, that Moore's law, was not about technological limits, but rather, economic in nature, right? It is/was an economic law for the industry.
You do know, that Moore's law, was not about technological limits, but rather, economic in nature, right? It is/was an economic law for the industry.
There are linear limits that are pretty close, but we haven't even started getting close to any vertical limits. Stacking is likely the next big advancement that will have years of improvements before those limits are reached.Moore's Law is neither about technological limits or economic systems. It's not even a law at all, it's just an observation of a system we know to be the industry ...
My point ultimately still stands that CMOS technology has a limit and that we can't surpass those physical limits. We have to accept those limits as a part of our reality even if it is detrimental to the tech industry's growth so being in deniable about the physics behind the subject isn't helpful ...
The so called 'limit' for CMOS transistors is actually bigger than just a single atom according to recent literature and even then we have to move off of silicon and use different materials to reach gate lengths as little as 1nm ..
Moore's Law is neither about technological limits or economic systems. It's not even a law at all, it's just an observation of a system we know to be the industry ...
My point ultimately still stands that CMOS technology has a limit and that we can't surpass those physical limits. We have to accept those limits as a part of our reality even if it is detrimental to the tech industry's growth so being in deniable about the physics behind the subject isn't helpful ...
That, and it wasn't a law as so much an observation/prediction that didn't last too long as the actual time period for doubling kept changing year after year.
Not that I would care for transistors <1nm wide, give me a device 100 nm wide that acts on 100 different values at once at 1THz rate and then we talk. Doesn't matter if it has to use light, quantum effects or some other (today) nice phenomenon, as long as computational power keeps growing. With decent performance/power ratios of course.
We had eras in computing, this doom and gloom on physical limitations approaching is moot when seeing only linear scaling as the focus.
Remember we are still far from emulating that 1kg of brain matter at 20 W with 1 kg of silicon... gives an idea how far the limitations truly are.
Gate lengths are not 5nm or even 40nm right now, so worrying about what happens when gate lengths actually reach 1nm is not a concern.
There are linear limits that are pretty close, but we haven't even started getting close to any vertical limits. Stacking is likely the next big advancement that will have years of improvements before those limits are reached.
i used to think the same about stacking, but tsmc and intel are seemingly pushing ahead on this. both mentioned peltier solutions to solve heating. but...The laws of thermodynamics will eat any vertical logic stacking designs alive. The highest end CPUs and GPUs runs over hundreds of watts on unstacked logic dies ...
Stacked memory modules ? Sure but it's not happening with logic dies since at that point MCM designs would make more sense ...
Directly from the gate to what? Anything other than a refrigerant system is going to go above ambient. Those would add to costs, energy use, complexity, size, etc.directly from the gate
To the IHS. Moving the hotspot heat through liquid would be faster and more directed than though the silicon itself.Directly from the gate to what? Anything other than a refrigerant system is going to go above ambient. Those would add to costs, energy use, complexity, size, etc.
found a link, might need to bone up on your french or find a translation. dont think this where i originally heard about it.
Transistor-integrated cooling for a more powerful chip
Researchers have created a single chip that combines a transistor and micro-fluidic cooling system. Their research should help save energy and further shrink the size of electronic components.www.sciencedaily.com
I don't believe anyone has suggested Intel is doomed. They are in for a pretty rough patch is all. And I doubt they will ever have that huge lead in everything again. But that's good for all of us. Now they have to actually try and innovate instead of just print money.Intel is in a 1000% better position in every aspect of financial measures than AMD was 12 years ago.
Intel is also very adept at financial engineering. Unfortunately Intel doesn't sell financial products, maybe it should consider joining that market.Intel is in a 1000% better position in every aspect of financial measures than AMD was 12 years ago.
Yes, Intel is in no way doomed. And the same thing that prevented AMD to get more marketshare 15 years ago, prevents it doing it now, and this is the quantity of silicon it can produce. If Intel disappeared now from the face of the Earth, AMD could not supply all the server, desktop and mobile chips needed worldwide.
So Intel will get on by at least 5-10 more years even if it did not retake the performance crown. there is no-one at this moment to supply all that demand. But if nothing changes, the fabs will became obsolete for top-of-the-line CPUs and if they will not perfect their 7-5-3nm nodes, they will be in trouble. At the moment they are ok. Global Foundries disappeared in 3 years from the front line of CPU manufacturing when they could not make their advanced nodes. How long will it take Intel if they cannot make another advanced node and TSMC can, and grows their wafer output to eventually full the AMD needs?
I was going to suggest the very same thing. Sears and GM, both had "Sears Financial" and "GM Financial". Why not "Intel Financial" or "Intel Capital" or something.Intel is also very adept at financial engineering. Unfortunately Intel doesn't sell financial products, maybe it should consider joining that market.
With all the issues with Intel and its foundries, what are the chances of NVIDIA buying Intel at a low PE? This will make them a direct competitor to AMD in most areas.
Intel is also very adept at financial engineering. Unfortunately Intel doesn't sell financial products, maybe it should consider joining that market.
I was going to suggest the very same thing. Sears and GM, both had "Sears Financial" and "GM Financial". Why not "Intel Financial" or "Intel Capital" or something.
Gate thickness has been a bigger concern in the past but Finfets and GAA tech (and new dielectrics before that) have allowed for continued progress without running into gate thickness issues. There are roadmaps with a clear path for at least the next 5 - 8 years or so, after that it becomes a bit sketchy. Not saying that they won't be able to continue to improve on silicon FETs, but the path gets a lot less clear from what I've seen (caveat being that I'm not a process engineer).
Yep, we're always running out of resources 20-30 yrs down the road. The answer is why invest in looking for something with no possible returns until at least 20-30 yrs have passed. To feel secure?The process roadmap has always been blank when you go 10 years out, it isn't like back in 1985 they had a roadmap with everything planned out to 1995, let alone 2020.
The industry doesn't need to know where it is going in 10 years for companies ASML and Canon to build the next generation of equipment. ITRS serves that industry, so they don't need to care beyond that horizon, even though there are obviously some people at places like IBM or academia doing blue sky research further out. Even with the roadmap, we end up seeing detours like how EUV was delayed and delayed and delayed and we ended up doing double and even quadruple patterning on 193nm sources. EUV was out near the end of the roadmap for at least a decade until they were finally able to make it work, but it didn't halt progress in the meantime.
I'm sure there will be similar detours if high NA EUV is delayed, and we'll have to see whether they can go to an even shorter "beyond EUV" wavelength or can make e-beam fast enough for mass production. As long as there are some companies willing to fund that exploration, they'll find a way like they always have in the past.
Even if TSMC ended up the last man standing at the leading edge of logic, there will still be the big DRAM firms who need to keep shrinking (though their "process name" metrics are different, they work in the same process domain - Samsung has just begun making DRAM using EUV) so there will be companies willing to fund this because demand for more RAM has never stopped even during shortages when prices shoot up.