John Carmack
Member
- Sep 10, 2016
- 160
- 268
- 136
There is a certain expectation when you go to misinformation merchants with questions.
I provide this not as an answer to your question but because the game looks fun and I want to play too.
I provide this not as an answer to your question but because the game looks fun and I want to play too.
ErichZannMusician
Member
- Jul 13, 2024
- 70
- 75
- 46
Did one year ago -
1. Warren Buffett halve the stake of B-H in Apple?
2. The Philadelphia Semiconductor Index fall by 22% in less than one month?
3. The Sahm rule get triggered after the latest unemployment data?
4. The 2/10 yield curve come just 9 bps close to turning positive?
I await proper answers.
desrever
Member
- Nov 6, 2021
- 167
- 445
- 106
Intel has a beta of 1.05 and AMD is 1.68. If the drop was only due to the market dropping, you'd expect AMD to drop ~1.6x as much as Intel.
The only thing that really matters for a company's survival long term is cashflow and here is Intel's:
Here is AMD's for comparison:
Good luck to them if they can't turn this around but even survival would require government bail out at this rate considering we aren't even in a recession yet and Intel is running with billions in negative cashflow.
ErichZannMusician
Member
- Jul 13, 2024
- 70
- 75
- 46
Great, now you've given us Intel's future path.
ErichZannMusician
Member
- Jul 13, 2024
- 70
- 75
- 46
Selling off their fabs to Abu Dhabi investment groups?
Intel sold half of the Ireland fab to Apollo and half of the Arizona fab to Brookfield investment groups.
ErichZannMusician
Member
- Jul 13, 2024
- 70
- 75
- 46
Retaining majority stakeholding in both while AMD in its case became a minority stakeholder.
51-49 for Intel vs 44-56 for AMD to be precise.
Edit: Not to mention the WSA which was an albatross around AMD's neck till they finally transitioned to TSMC.
Last edited:
So Intel claimed in the earnings release that 18A is 15% enhanced perf/W and 30% improvement in area over Intel 3, not 20A.
This explains why 20A is extremely limited release for certain -S 6+8 SKUs, because it'll barely have any improvement over predecessors, and in reality it'll underperform seriously as Intel 4 did.
In original claims, they said 20A offers 15% improvements and 18A offers further 10% improvement over Intel 3. If 18A still offers 10% improvement over 20A, that means 20A is less than 5% over Intel 3.
Density against competition is going to be very close.
Crestmont N6 - 1.46mm2
Crestmont Intel 4 - 1.01mm2
N6 - 16% over N7
N5 - 1.84x over N7, but according to Angstronomics, it's actually 1.5x
N3 - 1.7x over N5
Hypothetical Crestmont N7 - 1.69mm2
Hypothetical Crestmont N5 - 0.92mm2, 1.12mm2(Angstronomics)
Hypothetical Crestmont N3 - 0.54mm2, 0.66mm2(over Angstronomics)
Intel 3 - 1.1x density over 4: 0.92mm2
Intel 18A - 1.3x density over 3: 0.7mm2
Best case scenario, N3 and 18A might be really close in density, especially if we consider N3B is an underperformer and N3E with 1.6x is the good node, which will make Crestmont N3E be 0.7mm2.
This explains why 20A is extremely limited release for certain -S 6+8 SKUs, because it'll barely have any improvement over predecessors, and in reality it'll underperform seriously as Intel 4 did.
In original claims, they said 20A offers 15% improvements and 18A offers further 10% improvement over Intel 3. If 18A still offers 10% improvement over 20A, that means 20A is less than 5% over Intel 3.
Density against competition is going to be very close.
Crestmont N6 - 1.46mm2
Crestmont Intel 4 - 1.01mm2
N6 - 16% over N7
N5 - 1.84x over N7, but according to Angstronomics, it's actually 1.5x
N3 - 1.7x over N5
Hypothetical Crestmont N7 - 1.69mm2
Hypothetical Crestmont N5 - 0.92mm2, 1.12mm2(Angstronomics)
Hypothetical Crestmont N3 - 0.54mm2, 0.66mm2(over Angstronomics)
Intel 3 - 1.1x density over 4: 0.92mm2
Intel 18A - 1.3x density over 3: 0.7mm2
Best case scenario, N3 and 18A might be really close in density, especially if we consider N3B is an underperformer and N3E with 1.6x is the good node, which will make Crestmont N3E be 0.7mm2.
desrever
Member
- Nov 6, 2021
- 167
- 445
- 106
Maybe you forgot what happened around 2008 that would make companies do poorly.
Also not sure what your even trying to say. AMD turned around barely means that Intel is in a good position?
If PS4 was a flop I doubt we get Zen. The console deals kept AMD afloat enough that their Zen moonshoot could launch and it happened to coincide with Intel serially misfiring.
If Intel executed like NV do then even Zen would not enough to bring AMD back.
Intel are facing AMD that has reliable execution and is now established in many markets. They have a far higher cost basis than AMD ever did and a dwindling cash supply. On top of that mind share is being eroded in client and if that domino falls completely I don't see an easy way back for Intel.
Why do keep claiming this? N3B is superior in all respects to N3E except for cost.
Quality is better with N3E, ie: power consumption is a bit less, performance a bit better.
Have they upgraded it? Originally it was a small reduction in performance for a greater reduction in costs (less fab steps).
AnandThenMan
Diamond Member
- Nov 11, 2004
- 3,979
- 589
- 126
ATI acquisition saved AMD I'm willing to die on that hill.
Intel has incredible momentum on their side I think it will take a couple more product cycles of failure before they are truly at risk of becoming irrelevant.
Agreed.
I don't think Intel has any momentum on their side. And if we are to believe what reliable Intel people have said, Gelsinger and management are cutting everything on design teams to bet at their foundry play.
They're certainly more resilient, but the situation isn't pretty at all. They'll need consistent good to amazing execution from now on to get back on track. Arrow Lake, Lunar Lake, Panther Lake, Wildcat Lake and Nova Lake will need to deliver. And Intel will need Granite Rapids, Diamond Rapids and Future Rapids to execute quickly to regain competitiveness on servers.
The only thing N3B is better at is density; which the better yield of N3E more than makes up for it.
Intel is basically going bankrupt trying in vain to save the foundry.
Pretty much. If foundry doesn't work out, Intel will have wasted away all their future into a failed gamble.
If we look at current Intel competitors in the CPU space, there's a lot of Intel DNA that were fired or layed off in the 2010's. I feel Intel is basically funding their competitors once again with the current layoff and multiple products/designs cancellations they did. There's already a new startup by ex-AADG fellows who were working on the cancelled RYC and TTL.
N3E is superior in both yield, performance and power. N3B only leads in density. The mediocrity is what made them iterate N3 versions so fast. There was some doubt expressed towards their execution because of N3B.
This is comparable to 20nm where almost no one used it and waited for 16nm FinFET.
They got no real alternatives. You think the design team is better? Heck Arrowlake barely on par with Zen 5 should tell you something.
Here's the past 20 years of Intel: Terrible designs propped up(IOW shielded) by phenomenal process technology, which got exposed when they lost the process lead.
It's been known for those keeping close track of Intel since the 90's that they never had a good design team. It was the process team that kept the open secret from being realized by the public.
There's a saying "whatever Intel giveth, Microsoft taketh away". In this area, "whatever process team giveth, design team taketh away".
Multiple times when Intel should have lost due to crappy CPU arch, it moved to a new process and stopped it sucking.
It is not a good sign, not a good sign at all when the Moore's Law gains are really, really slowing down they are spending all they can at gaining leadership. The timing is really bad here. 18A is both RibbonFET and PowerVIA for mere 15% gain in performance and 30% density. 14A is another 15% gain in performance at 20% density.
The changes needed are being more revolutionary every generation for a more pathetic gain every generation. Watch Nanosheet transistors give 10% performance and 10% density, or something.
There's a saying "whatever Intel giveth, Microsoft taketh away". In this area, "whatever process team giveth, design team taketh away".
Multiple times when Intel should have lost due to crappy CPU arch, it moved to a new process and stopped it sucking.
It is not a good sign, not a good sign at all when the Moore's Law gains are really, really slowing down they are spending all they can at gaining leadership. The timing is really bad here. 18A is both RibbonFET and PowerVIA for mere 15% gain in performance and 30% density. 14A is another 15% gain in performance at 20% density.
The changes needed are being more revolutionary every generation for a more pathetic gain every generation. Watch Nanosheet transistors give 10% performance and 10% density, or something.
I don't think the foundry revival by Gelsinger was an error. But I think he went too far with it and made the company worse by. Cutting on Intel (The design and product company) to prop their bet on Intel Foundry is tanking the company.
As an example of what I mean: Intel management, to save on costs and keep the foundry play, basically cancelled RYC and TTL, cancelled Forests after Clearwater and merged (killed) E Core team into P Core team for them to be unified (Basically P core won the resulting political feud) and build up something from it. Part of Xeon staff also divested to work on their HPC AI solutions, etc. And the cancellation of multiple products and downscaling of future products line-ups.
While INTC Design was never good, you don't just cancel everything. That's basically throwing away their future to make the bet on foundry work. And while at that, the brain loss of veterans that are leaving Intel, including AADG people.
If Intel Foundry fails, there are foundries who can manufacture their products (Granted, not at the scale that Intel requires. At least not for cheap). But if your design is flat out uncompetitive, there's no foundry who can save it. Specially when advancements are getting harder and harder to come by.
The thing is, when they started it, the economy was falsely propped up on forced lockdowns, so decline was pretty much guaranteed after that, but up until then things looked very rosy, so you invest.
Hopefully it's the E core that takes the lead, and Gelsinger made sure of that. Because if it's the other way around, then it's pretty much death.
So rather than my prediction that it'll be peak with Arctic Wolf or the successor and downwards, it might be Skymont/Darkmont is the peak.
Cancelling Clearwater Forest successor would also be a mistake and guarantee ARM server dominance.
I have this feeling, a distant possibility that Gelsinger might be some sort of a Trojan Horse, brought in with the guise of "saving Intel" but the real goal is to destroy.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 20K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 12K
-
Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)- Started by Vattila
- Replies: 13K
-
Facebook
Twitter