Couple Intel slides on Arrow Lake-S:

+9% IPC P-Core, +32% IPC E-Core compared to 13th/14th gen.

Single thread Intel Ultra 9 285K (Blue) vs AMD 9950X (Red) vs Intel 14900K (white line):


Games, 14900K vs 285K, arrows denote drop in power consumption vis-a-vis 13/14th gen:


Games, 285K vs 9950X:




Very interesting if true. Looks like Intel heavily focused on power consumption (much like AMD), with ARL-S not always able to beat Intel's own previous gens or AMD's current gen. Earlier multi-threading benchmarks were pretty bad for Intel, might be why there's no MT slides - but we'll have to see.

 

Fek i can't find my own comment about a month ago(?) where i said i 'd heard that the upcoming TSMC made Intel Core Ultra 9 285K, 265K and 245K will support DDR5 6400MT out of the box with no oc. Probably not in this thread. Anyway, it's true. Also support for CUDIMMs is there too. And all on TSMC's latest 3nm process node. I hope they 're good, we need the competition back up and alive!
https://www.techpowerup.com/cpu-specs/core-ultra-9-285k.c3773

EDIT: What is very interesting here is that these new Intel cpus have no hyper threading whatsoever. Now this comes at a very strange time where AMD are heavily invested towards adding one more thread per core, actually going for 3 simultaneous threads per core. Don't be surprised if AMD's next gen is 16cores/48 threads hahah. Also the Intel cpu is no longer a monolithic design. It's chiplets like AMD but they call it "tiles" because they stack 'em. Hmmm... can't wait to try these Intel Ultras...
EDIT 2: Those new Intels have a very interesting hardware feature for the techies. All cpus before it had the VRMs from the motherboard delivering voltage to the cpu. Intel Ultras have a DLVR (Digital Linear Voltage Regulator), which actually serves as a mediator between the VRMs and the cores themselves executing a type of "check and regulated delivery" of voltage to the cores. And there also is a DLVR bypass function. Now why do i think that all this 13th and 14th gen fiasco led to Intel actually having to install such a device on their cpus to keep in check reckless mobo manufacturers (and users for that matter) hehehe. Btw these new cpus can have P cores individually tweaked which may or may not lead to interesting results. We 'll see...
EDIT 3: Correction. The 6400 mt/s ram native support is only for CUDIMMS. Normal UDIMM support is at 5600mt/s. Sorry for this but i just learn as i go along, after other more thorough sources than Intel's own presentation and while being on social media with all techies i know too. Oh and 4 DIMMs on board drops it to 4800MT/s. All this is yet to be confirmed. Stay tuned

 

Last edited: Oct 12, 2024

Per CNBC & Reuters:

The company registered a net loss of $16.99 billion, or $3.88 per share, compared with net earnings of $310 million, or 7 cents per share, in the same quarter a year ago.

Restructuring [...] including lowering head count by 16,500 employees and reducing its real estate footprint.

Intel said the manufacturing unit had $7 billion in operating losses for 2023, steeper than the $5.2 billion in losses the year before. The unit's revenue was down 31% from the year before. CEO Gelsinger said that 2024 would be the year of worst operating losses for the company's chipmaking business and that it expects to break even on an operating basis by about 2027.

As of Thursday’s close, Intel shares were down 57% in 2024, while the S&P 500 had gained 20%.

Intel will be replaced by Nvidia on the blue-chip Dow Jones Industrial Average index after a 25-year run, underscoring the shift in the chipmaking market and marking another setback for the struggling semiconductor firm.

 

Intel cuts thousands of jobs

Intel is in the midst of its transformation. The classic processor business is changing massively. Intel reported a sales increase of around nine percent to 7.4 billion dollars for its Client Computing Group (CCG). This was primarily due to the new AI PCs, it said. Since the end of 2023, around 15 million computers with CPUs specially designed for AI tasks have been delivered. By the end of 2024, this figure is expected to be around 40 million. However, this also entails higher costs for the development and production of these chips. This increases the pressure on profitability.

There are also problems with products for data centers. Sales in the Data Center and AI (DCAI) segment fell by three percent year-on-year to around three billion dollars. Intel is trying to catch up with the competition with its own graphics accelerators for AI use. But competitors like Nvidia are well ahead. Many data center operators are currently upgrading their infrastructures for the emerging AI era, but for the most part they are not doing so with Intel products.

Intel CEO Gelsinger: Costs too high, margins too low

Gelsinger is preparing his workforce for tough times. The manager spoke of painful news and an incredibly tough day for Intel. "These decisions have challenged me to the core, and this is the most difficult thing I have done in my career," wrote the Intel boss in an open letter to his employees . Cost structures must be adjusted and working methods fundamentally changed. "Our costs are too high, our margins are too low."

Source: https://www.computerwoche.de/article/3487931/intel-streicht-tausende-stellen.html

 

I think Intel is on the right track when they focus on heat/power/noise on this reset. I consider the Intel Ultra CPU's as generation 1, and I probably gonna buy generation 2 which I hope has the same 1851 socket.
I look upon these new CPU as great allrounder's not only good at gaming. I think also these first generation CPU's gonna be pretty good after a couple of months with bios, windows and chipset updates. People have to give new generations some time.....seems like this is completely lost on most so called youtubers. They know negativity sells in form of more views on their videos.

To be a first generation I think this release has been pretty good.....seems like I'm in a very small minority... By the way the most important thing to me is stability and usability. I'm most worried about if the new tiles technology in CPU's will create more latency and stuff like that. Also I'm very curious if the built in thunderbolt 4 on the CPU works with thunderbolt 2 devices.

 

Last edited: Nov 2, 2024 at 2:05 PM

No you are not. For what it's worth mate i do agree with you. These are very good cpus and almost re-designed. There's nothing here resembling the traditional monolithic Intel design. Well, as Wendell said in YT the launch seems rushed. So this gen is plagued with bugs and lower performance in gaming than the competition. But this we already knew.

They do. But it's not the tiles fault, at least as being the main cause imho. I think they -sort of- fucked up because the basic tile where all the other "tiles" are laid onto (kinda like the chiplet as AMD calls it), is the only part made by Intel at 22nm. I wasn't aware of this previously. So this "basis" tile, looks to me as the main culprit for the higher latency of the cpu. I may be wrong but it looks like the only outdated part in there. This along with a few other questionable choices, are also responsible for the sub-par performance in gaming. But there is the good side. The cpu is very good at what counts for us, which is multi-threaded workloads. And without hyper-threading too. Thinking that HT can improve performance by 5-20% on different workloads, i can only imagine what will happen if this cpu tech evolves and HT makes a comeback. It won't be the first time that Intel's done this with HT anyway.
Intel Ultra's not too shabby at single core either and the Efficiency cores have been greatly improved. It is also the first Intel cpu that you can properly cool as it is far more efficient than at least the 3 previous Intel cpu gens.
So to sum it, i can understand they rushed the release because they had to -sort of- prove to both their shareholders and buyers they are on their right path. And they seem to be, this cpu is not bad at all, it just suffers from the usual first gen syndromes. We are patient. Let's see what happens.
There is one last issue that has come up.
https://www.tomshardware.com/pc-com...intel-fabs-nova-lake-almost-entirely-in-house
Even if you don't read it, the title says it all. Now this makes me wonder why they think they can pull this off. To me it's more wishful thinking really as they haven't been able to catch up with TSMC or even come close and all their competition is on TSMC nodes. Surely they 're dying to move production in-house, because Intel is not AMD and their fabs are costly, so when Intel outsourced cpu manufacturing to TSMC, they get very little profit from it because they still have huge bills to pay, even after firing 15000 people. But unless they pull a magic trick and suddenly their fabs appear to be able of 4 or 3nm lithography with yields that can provide a margin of profit, i think it could be disastrous.
EDIT: Supposedly coming next Panther Lake will be using Intel's 18A process for the main compute tile, that is the cpu cores. This is supposed to be an 1.8nm (yes you are reading correctly) process. If they can pull it off successfully, sounds like it's gonna be massive, at least for Intel's foundries.
Cheers
PS: And now that AMD's 9xxx X3D cpus (starting November 7 with the 9800X3D) will supposedly be fully unlocked, AMD will have a clearly leader product line for the moment and competing will become increasingly difficult for Intel.

 

Last edited: Nov 2, 2024 at 8:25 PM

Ok, first let's get out of the way the "clickbait" made-up-while-high-on-LSD name of the node (1,8 my virgin ass).

The only important thing is if they can compete with TSMC. I think nobody believe they can pull that off. I wonder why are telling this.
Perhaps damage control acrobatics to avoid a 0.1% further decline in the stock.
A "white lie" since, you know, they have wasted zillions of dollars of government help?

Anyhow the first gen of these chips seem promising and AMD has fallen short with their last gen. Hopefully the gap will not be high so we have competence.

 

IMO the problem with this release is that Intel has no clear path forward. Not because Arrow Lake is a dead-end - but because there are more "obvious" areas they could improve on than they have money or time for. Intel is forced to choose some of the lowest hanging fruits for 2025/2026/2027 and pray they didn't pick the wrong ones. So now Intel beancounters are fighting Intel design teams and Intel design teams are fighting other Intel design teams to get their ideas in.

> On-package memory is one of the factors that made Apple's M-series processors fast, efficient, and compact. With its Core Ultra 200V (Lunar Lake) processors, Intel adopted the same architecture, which severely hurt Intel's profit margins.

> Gelsinger: Panther Lake, Nova Lake and Nova Lake+: We'll build it in a more traditional way with memory off package in the CPU, GPU, NPU and I/O capabilities in the package

We're killing off what we know would help fix ARL's memory latency, but we're out of money and time. AMD 9000 and Intel RPL mem latency is almost 50% lower than ARL. Gelsinger says they'll focus on moving as much as they can into a SoC-tile and optimise that tile.

> Gelsinger: Graphics, and how that is increasingly becoming large integrated graphics. So, less need for discrete graphics in the market going forward.

Ouch, bad signs for discrete Xe3 (whatever was supposed to come after Battlemage).

> In 2021, Intel was moving full speed ahead to attract US government subsidies. As part of the effort, Gelsinger made controversial comments about Taiwan's vulnerable position; likely intended to extract a better deal from TSMC. The comments drew a response from TSMC's executives, leading to Intel losing its 40% discount on 3nm products.

Wowzers, what an absolute donkey. Gelsinger's a relic. One with a colonial attitude. "Gimme better discount, you dirty Coolie."

 

Last edited: Nov 4, 2024 at 11:39 AM

Well hehe. They meant to pull an "Apple" but they pulled an "Intel" instead. Meaning those mobile cpus are not faster than the M3 both in single and multi-core performance. A little better in battery life for the 15W ones vs something like Macbook Air. What Intel does not comprehend as a company is that Apple, doesn't bring out newly designed products before they have been tested -sort of- well. They rarely, if ever make "bets" with products and under Tim Cook they prosper, as he's proved to be an even better businessman than Jobs, minus the kinda "rock star" crowd appeal that set apart the late Jobs from everyone else in the tech industry.

Yep, i agree. Wtf was he thinking and the same goes for his advisors. To me it's because in the back of their minds, they still think they rule and can get away with almost anything and everyone else will "comply" with their demands. Someone has to wake them up from their deep slumber or they will lose whatever shred of credibility they have left.
I had hopes for Gelsinger when he took over as he was the first in a long time to have a tech background, being a successful engineer in the past etc. But as it seems, he doesn't have the stamina, balls and endurance needed for crisis situations and likely the business mind to lead Intel atm, imho always.
I don't know how skipping 20A (2nm) to 18A will do them any good.
At the moment they have a good new product that should set 'em back on track. They should at least try to capitalize on that for a couple of cpu gens, even with a very narrow margin of profit because of TSMC's involvement.
To me, if they managed to pull off even 20A (like in retail release numbers and not just lab samples), it would seem like an astronomical leap for Intel foundries and eventually Intel cpu/gpu products. If i recall correctly they claim that even their 7nm (or is it 8nm) is denser than TSMC's 3nm hehe. I will sit back and enjoy the show mate
Cheers

 

Last edited: Nov 4, 2024 at 9:00 AM

The competition between Intel and AMD dates back to the late 1960s, when AMD was founded as a secondary supplier for Intel's chips. However, the competition intensified in the 1990s when AMD introduced its own x86 processors, challenging Intel's dominance. Since then, the two companies have been in a battle of blows, with each taking the lead at times. Intel had a big lead in the early 2000s.

 

Yeah, same. After reading about him messing up the TSMC deal, I believe it was the same "better than you" attitude and aura that landed him the CEO job in the first place. Believing in Gelsinger's hype is costing Intel dearly.

Yeah, I think Intel isn't being truthful about the state of 20A/18A.

Haha, well, there's a lot of (tax payer) money involved. Starting to feel a bit like the next couple years are gonna be a bit too interesting for my liking:

> Samsung has established a task force with the objective of reducing the work force at its foundry business by a third. According to Korean media, Samsung is also considering spinning off its foundry business within the next 3-5 years.

NYT, SemaFor:

> The Commerce Department and members of Congress have grown worried enough about Intel to begin discussing scenarios should it need further assistance, beyond the billions in government funds the company is slated to receive.

> Bloomberg reported that Intel has been reluctant to share information sought by US officials tasked with ensuring that the company has a viable turnaround plan.

> Credit ratings agencies downgraded the chipmaker, increasing borrowing costs. If Commerce Secretary Raimondo [who has been encouraging major tech companies to source chips from Intel] can outmuscle Federal Trade Commission Chair Khan, it will show how deep Intel’s problems run. Raimondo's efforts to support Intel are complicated by skepticism about the company's ability to deliver [18A, finances], leading to the establishment of stringent milestones for funding.

> Intel’s wobbles may force Washington to choose between promoting national champions and preventing corporate consolidation. On a clear day, there’s little chance antitrust regulators would allow Intel to merge with its rivals, but [...]. There’s little appetite for a 2008-style bailout in Washington.

> One option being discussed is a merger of Intel’s chip-design business with a rival like AMD or Marvell.

AMD Inside, let's go.

 

Absolutely on the last one haha! I don't think AMD would have much need for them but who knows...
And i hear you on the merger talks. Especially after TSMC's CEO when asked, replied TSMC doesn't want and has no use for Intel's foundries.
But hmmm... now that Qualcomm is one step away from losing their ARM license (not a rumor: https://www.tomshardware.com/pc-com...omms-architecture-license-as-feud-intensifies ), they will be more aggressive as to how they will proceed as they have a huge pie of the mobile phone market and are willing to enter the laptop (and why not desktop too) biz as well. Just a thought of course, as a takeover was on the table earlier on.
Also with the latest tech stock market highlight being Nvidia replacing Intel in the Dow Jones, this is a major hit in Intel's credibility and status. A triumph for Nvidia which has shown amazing growth and stability and whether we like the guy in the leather jacket or not, he is a leader to be reckoned with.
Oh and another irrelevant but fresh rumor: Threadripper 9xxx might have an X3D model, as in the latest BIOS for ASUS' WRX90 SAGE there is an option to disable 3D V-cache per stack but there is no TR cpu with 3D V-cache to begin with, for the time being. So what's this for hehe.
I do believe this to be true, as it points towards the right direction and is a great move, improving on what was already a killer high end cpu. Shit that's fkn exciting hahah. I better sell an amp or two and be ready, that thing will be expensive!
Cheers

 

Hello @taskforce, I read that CPUs could be clocked higher, but cooling is a problem. Increasingly smaller production also leads to more heat being generated. I've read a bit about it and some companies are working on it. How do you see the future of cooling? Just a joke, there was a video recently of oil-cooled PCs, they were dipped in vegetable oil, apparently it worked.

QUOTE. "Intel is working with Purdue University in West Lafayette on so-called piezo fans. Tiny springs on piezo elements move at high frequency and thus fan the heated air away - this only makes sense if the heat is generated above the chip surface rather than on it, as is the case with counterheat. In October 2007, Sun patented a liquid cooling system that does not have any moving parts and uses a periodic, micromagnetic field. As with counterheat, electromagnetic fields must be generated in a targeted manner here too."

In the past, AMD was constantly in the red and was kept alive with loans. Now Intel has suffered a loss in sales due to Untermerische misdivorces and other poor market conditions. They had to lay off employees, the planned factory in Germany is on hold, but the most important question is: will Intel disappear from the market? I think no, it's just that the market for Intel is and is getting smaller.

 

should i buy intel stocks right now ?
whats the mimimum i have to buy ?

 

Intel share: forecast and price target for 2025 The experts' assessments of Intel are as follows. A total of 46 have come to the average recommendation of "Hold". The sales recommendations must be given as follows: Sell ( 4 ) and Strong Sell ( 2 ). But how many experts are in favor of buying Intel shares now? Buy ( 1 ) and Strong Buy ( 5 ).

If investors hold the title, this is what the analysts say: 34 experts recommend holding the share. Intel shares are currently trading at €21.40. The experts see the average target price for the next 12 months at around $33.59. Is this an undervaluation and does this create opportunities for a further increase?

It is better to buy gold, that is the only thing that is certain. It has value and the value will continue to rise.
If you love shares, buy Nvidea or Microsoft shares.

 

was hoping for the experts commenting in this thread to give their opinion.
..would they do the bet if getting a 1000 money for free to trade ?

 

I think we're not quite at the bottom yet.

> The German Ministry of Economics has withdrawn plans to subsidize Intel's Magdeburg plant with €10 billion in taxpayers' money after the company decided to “pause” the project for at least two years.

 

Yeah, its bad. While I'm not a big fan of Intel for their business practices and their high prices when there were no competition, I don't cheer for this. We need at last two strong players in the CPU market. Like nVidia is too strong atm and has no competition in ai, so their prices are extremely high just to please the stock market.
I hope Intel gets a grip soon.
And I'm also worried that there is only one company delivering the machines for lowest nanometer lighography, and as it seems with TMSC there is only one that can use it.

 

found an interesting reddit
They are waiting for JD Vance to become VP of US so he can threaten to pull NATO unless Intel gets a better deal and the “freedom” to not pay taxes or respect unions.

I heard this from credible sources in Ohio.