Perhaps that explains why PTL-U had a PL2 of 57W in the leak, whereas LNL has a PL2 of 37W.
Perhaps that explains why PTL-U had a PL2 of 57W in the leak, whereas LNL has a PL2 of 37W.Intel nodes are optimized for high power/high frequency so for Lunarlake like products they'll lose perf/w.
SHAMEFUR DISUPURAY, EEK-kun.Oh I absolutely was not going to respond in this thread, but i have to…Intel has no plans for desktop compute (“cpu”) chips (or mobile) to be manufactured at anything other than an Intel fab.
Speaking of which, TSMC pushed back backside power delivery to a later node.I hear you. 18A brings the backside power delivery. 20A was supposed to be proof of concept type silicon. Remember the days when Intel would never consider outsourcing their processors to fabs from outside companies. In my mind, Intel has already lost this round for outsourcing their silicon to TSMC. We already know TSMC makes real good silicon. Intel is acknowledging this by using TSMC.
I am seriously disappointed.
Speaking of which, TSMC pushed back backside power delivery to a later node.
If Intel wants to catch up, I think a lot hinges on that. If 18A still holds on to it, it might be a stabilisation. Otherwise it's starting to look like AMD-Nvidia software support where AMD is always one gen behind or worse.
Great post.
We may end up with the unthinkable.
Intel, needing money and customers for IFS badly, will accept business from AMD.
Don't give them ideas for codenames!I could envision a pitch where whoever leads Intel then says that they need to accept to build Ryzen volume and cannibalize their sales for a year or two, and survive financially until whatever next Saviour Lake comes.
Why?Don't give them ideas for codenames!
Apparently good news regarding Intel and 18A.
I don't have a subscription, but this is what I think this article says.
I wonder if anything regarding 20A is mentioned.
https://www.semiaccurate.com/2024/09/09/what-is-really-going-on-with-intels-18a-process/
Charlie is usually doom and gloom for Intel so the news can't be all that bad. I won't say "good" because he says he went looking for bad news as far as 18A delays or yield issues, but says he wasn't able to find anything like that. But in this case since Intel is so beaten down I suppose the absence of bad news counts as good news.
He's been shilling Intel for quite a while now.Charlie is usually doom and gloom for Intel so the news can't be all that bad. I won't say "good" because he says he went looking for bad news as far as 18A delays or yield issues, but says he wasn't able to find anything like that. But in this case since Intel is so beaten down I suppose the absence of bad news counts as good news.
Great post.
You just reminded me of something else.
Recent shortcomings have come from AMD lacking volume for OEMs to put their chips in laptops. Not the first time nor the last.
But what's interesting about it is how the forces are slowly playing out. Intel at this point sells laptops not because they're good (might change with LNL ofc but still not out yet), but because they have a reliable, ready-to-go volume of chips that AMD doesn't have. Considering AMD's culture, I don't really picture them pressing TSMC much harder for more wafers. They're penny pinchers not large volume investors, and TSMC is famously busy with many clients.
It goes like this:I think it went somethink like:
Intel OEMs: "Hay AMD, can we have extra 40 million CPUs this quarter, with no notice what-so-ever, so that we don't have to sell Raptor Lake?"
AMD: "No, we don't have 40 million CPUs sitting around"
Intel OEMs: "Can you ask TSMC to fab extra 40 million CPUs this quarter, with no notice what-so-ever, so that we don't have to sell Raptor Lake?"
AMD: "We can ask, but I doubt they have capacity for 40 million CPUs idle, sitting around"
Qualcomm probably subsidizing OEMs to launch designs to enter the market.Still doesn’t explain how Qualcomm got all those design wins from multiple top tier bands and AMD just had Asus. Exclusivity is a dumb idea, particularly if after the exclusionary period ends and multiple OEMs don’t put out their chips.
Fair. But still more variety then with what’s available for the HX 370.Qualcomm probably subsidizing OEMs to launch designs to enter the market.
It's not really that many design wins either, just some high profile ones in absolute numbers, they there really isn't that many X1 models available either.
Now they reiterate that N2 will not have BSPDN, but say that A16 will. It still isn't clear to me if you HAVE to use BSPDN if you use A16, because there may be some cases where it really doesn't make sense to incur the additional cost. We'll have to see.
I think what happened is Intel ran a few test batches of TSMC N4P or similar on Arrow Lake CPU's and up against 20A. The results were TSMC silicon was faster, more efficient and better than 20A and the cost for TSMC silicon was reasonable. There is no clock regression with TSMC on the 5nm node. They probably had some yield issues as well with 20A. 18A has the backside power thing going for it. I have said 20A was kind of a cheater node anyway. I think 18A is closer to real world 3nm.You missed this:
Speculation but it seems good as any.
Whether due to execution sucking or finances not allowing it, they couldn't get extra 10% for 18A over 20A, nevermind extra density.
Therefore, in the spirit of trying to reduce costs, they took 20A, added extra libraries, and renamed it 18A. Hence 18A is only 15% over Intel 3, same as the original claims for 20A. What do you do with the 20A naming? Cancel it!
When morality of a nation falls, the fall of a country is not too far behind.
Nvidia CEO Jensen Huang spoke at the Goldman Sachs Communicopia conference on Wednesday and was asked about the geopolitical risk tied to Taiwan and what he would do if something happened.
“In the event that we have to shift from one fab to another, we have the ability to do it,” Huang said, in conversation with Goldman Sachs CEO David Solomon. “We won’t be able to get the same level of performance or cost, but we will be able to provide the supply.”