Clearwater looks like a decent product, a good Arm competitor.
BTW, what's up with rumors that Intel cancelled future E-Core based server product past Clearwater Forrest? Some rumors regarding the demise of the E-Cores and that development team.
-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
News Intel 2Q24 Financial Results
Page 20 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Clearwater looks like a decent product, a good Arm competitor.
BTW, what's up with rumors that Intel cancelled future E-Core based server product past Clearwater Forrest? Some rumors regarding the demise of the E-Cores and that development team.
DrMrLordX
Lifer
That's possible, though outside interests haven't got that kind of control over the company yet. They own a large chunk of a few fabs and that's it. If they can form a shareholder's coalition, however . . . or just buy them out while the share prices are low . . .
@controlflow
The biggest concerns with Clearwater Forest are:
1). Sierra Forest has little to no commercial availability. It's basically unobtanium at this point.
2). Word on the street is that Intel will kill mont-based Xeons after Clearwater Forest
It's a rumour, and it has gained some traction. As to what's up with it or why, it's possible that Intel needs to trim the team(s) responsible for said products. It costs money to develop and deploy a Xeon. Sierra Forest likely hasn't paid for itself in terms of revenue. They probably didn't ship many units. Clearwater Forest would have to really turn the situation around, and if Intel doesn't plan on high volume shipping for that product either, then . . . cost reductions, yay.
Last edited:
Josh128
Banned
This is my thoughts exactly. Pat could have been more aggressive with cuts, but Im sure he did what he thought was necessary at the time. If there is an emergency meeting, its possible another huge round of cuts will be announced. I still think Pat is the right guy for technical strategy, but I'd hope he has outside consultants looking for places to trim for him.
igor_kavinski
Lifer
Intel has their own fabs. Even if it's older process tech than TSMC, it's theirs! They could've come up with a better plan to monetize them. Sold larger sized CPUs than AMD until they could catch up in process tech. Then AMD would've found it harder to compete with them, at least in MT throughput. Where AMD wouldn't have wanted to go higher than 20 cores max, Intel could've brought the hammer on AMD with up to 30 REAL cores! That's 60 threads, yo!
This is what I think killed Intel's desirability:
Stupid insistence on heterogeneous core processors, pretending that they were somehow going to take over the world overnight and AMD's larger cores would fall out of favor with the enthusiast crowd.
Cutting out AVX-512 just when it could have taken off with developers with AMD's AVX-512 support.
Maniacal insistence on high frequencies, power be damned.
Not taking AMD APUs seriously. What APU does Intel have? Absolutely NONE. There goes at least one million CPUs sold per quarter, mostly global sales in countries where GPUs are too expensive.
Utter lack of innovation. So much money spent on R&D and they couldn't put one interesting unique feature on their CPUs to differentiate them from AMD's??? Shame on you, Intel fellows!
High speed memory? Check. High speed storage a la Optane? Check. Arguably industry's lowest failure rate in SSDs? Check. Best in class consumer hardware video encoding/decoding engines? Check. eDRAM tech? Check. Unique packaging technologies portfolio? Check. Yet Intel squandered all of these advantages and let themselves rot to the core with complacency and poor execution due to zero sense of urgency.
And there's probably a dozen more reasons on why Intel lost their edge.
This is what I think killed Intel's desirability:
Stupid insistence on heterogeneous core processors, pretending that they were somehow going to take over the world overnight and AMD's larger cores would fall out of favor with the enthusiast crowd.
Cutting out AVX-512 just when it could have taken off with developers with AMD's AVX-512 support.
Maniacal insistence on high frequencies, power be damned.
Not taking AMD APUs seriously. What APU does Intel have? Absolutely NONE. There goes at least one million CPUs sold per quarter, mostly global sales in countries where GPUs are too expensive.
Utter lack of innovation. So much money spent on R&D and they couldn't put one interesting unique feature on their CPUs to differentiate them from AMD's??? Shame on you, Intel fellows!
High speed memory? Check. High speed storage a la Optane? Check. Arguably industry's lowest failure rate in SSDs? Check. Best in class consumer hardware video encoding/decoding engines? Check. eDRAM tech? Check. Unique packaging technologies portfolio? Check. Yet Intel squandered all of these advantages and let themselves rot to the core with complacency and poor execution due to zero sense of urgency.
And there's probably a dozen more reasons on why Intel lost their edge.
igor_kavinski
Lifer
Trust me. Those consultants will do more harm than good. Outsiders can never judge thousands of individuals in a workforce accurately to determine which ones are the most productive and crucial to the company's bottomline. This was supposed to be a critical function of HR but almost every company's HR dept manages to attract complete bozos who care nothing about improving the lives and productivity of the company's employees. They don't even try to understand what tasks each employee is assigned and if they are overworked or overburdened and instead just give complete control to the employees' managers to EFF UP the lives of hardworking employees.
Intel has 4 large fabs IIRC - Oregon, Arizona, Ireland, Israel. It's not exactly low balling to convert them to EUV capacity. Enough capacity so that Intel would not have to go to TSMC. And maybe a little extra to start accepting outside customers.
It's almost as if it wasn't enough of a challenge: to escape death, and live. Gelsinger quadrupled the magnitude of the hurdle and set the goal to emerge as a world leading foundry.
Very disappointing if true. The E core team seems the only part of Intel right now that has their act together. Intel is in a viscous spiral right now of needing some home run product, but because of financial hardship is cutting so many projects that could actually be that home run. I never would have though it possible, but I think Intel is in serious danger as a company right now. At least AMD in the darkest Bulldozer days had the console revenue to keep them going. Intel doesn't even have that.
Markfw
Moderator Emeritus, Elite Member
I read that its not selling. And I can understand. If you wanted a server and you choices were a cut down features CPU that is efficient or another server that is just as efficient, but full featured (bergamo) which would you pick ?
The buzz word is AI, including server CPU based AI Inference. Unlike P-Core based products that have AVX-512 and AMX matrix operations, E-Core based products have neither.
In the meantime, AMD has AVX-512 even on the dense cores...
If E-Cores and E-Core based server products were indeed cancelled, this could be the contributing factor.
It isn't clear from the news coming out of Intel, but my belief is that for a bit there Intel was considering whether to continue fabbing chips at all. That would explain the large capacity buys from TSMC and passing up on delivery of EUV scanners they had previously ordered. When Gelsinger came in he re-committed Intel to fabbing chips, but that required the scale of becoming a real foundry (not the laudhable halfhearted attempt a decade ago) and catching up with TSMC process-wise.
The problem with trying to do five processes in four years when you only have four fabs is that you're going to tie up a lot of that capacity in upgrades and preproduction activities before you're ready for mass production. Those nodes are all short lived - it isn't like there were a lot of products on Intel 4 or will be on Intel 3 and 20A - they are trying to skip through those as quickly as possible but tackling some new "learning" with each one (first EUV, then nanosheet, then BSPDN) That implies a continued need for external capacity until you get back to a more normal process pacing and are able to more effectively utilize your own fab buildings.
If they can keep their head above water long enough to get a fab or two in Ohio online then assuming they are able to deliver on their process roadmap they'll be fine from a foundry standpoint, and the customers will come simply due to the "made in the USA" (or for non US companies the not "all our eggs in one potential geopolitical hot spot") advantage.
Its just going to take a lot of money to get there - they were really lucky with the timing of the CHIPS act but even then have still needed PE money. If they hadn't borrowed against their overseas cash to drive buybacks/dividends so aggressively they wouldn't have needed the PE guys , but that's what you get when finance people are have too long a leash.
Rumors on reddit from what appears to be ex-Intel employees that Intel will only have 1 CPU design team going forward.
Got a link on that?
Seems interesting since the whole point of Royal was to subsume the p and e cores. So they’re going to wind up merging anyway. If so it sounds like Panther (aka Coyote) Cove and Arctic Wolf would be the last microarchitectures for p and e.
Commentary about Panther Cove being renamed Coyote Cove:
Looks like even more cost cutting measures are on the horizon: projected new cuts
Looks like Pat's plan to turn Intel into a foundry business was horribly ill timed. They simply don't have the resources to do it, IMO.
Looks like Pat's plan to turn Intel into a foundry business was horribly ill timed. They simply don't have the resources to do it, IMO.
Lunarlake has no direct successor. I want to know which bunch of dimwits authorized that?
If Intel fails then they should force x86 to be open so there's competitor to AMD in that space. Completely open the floodgates.
No, new products take time to ramp. Also, AMD has lot more reputation than they did few years ago.
Now they are paying for it. Even if he is a good technical guy, I'm afraid that's going to hurt his chances quite badly.
If Intel fails then they should force x86 to be open so there's competitor to AMD in that space. Completely open the floodgates.
Other sources are saying NO, so I wouldn't take that seriously.
Cause Zen 5 is so awesome that they couldn't get something which Tremont got it working 4 years ago.
No, new products take time to ramp. Also, AMD has lot more reputation than they did few years ago.
Pat made one critical mistake. Him and his management team completely flubbed on future revenue projections. They didn't see after the lockdowns were over the computer demand would plummet. So they overinvested, didn't cut sharebuybacks and dividends to pay off their massive debt, and put it into fabs when they were doing good.
Now they are paying for it. Even if he is a good technical guy, I'm afraid that's going to hurt his chances quite badly.
Last edited:
Things that need to change at Intel before they can have a future
-The company Intel is beyond the point of MBAs merely influencing engineers. Their core decisions are driven by them. This needs to change completely. We thought bringing in Pat Gelsinger was enough. It seemed to have made almost no difference.
-The company must be always on the lookout for new opportunities for their main business. What is their main business? Making good CPUs. If they think 1W platform power is low, then try to get it to be 500mW. If they think they have $10 CPUs, look for ways to get it down to $1. Smaller, Lower power, Faster, Better, that is the core of Moore's Law. And out of the four, Smaller + Lower Power benefits disproportionately with Moore's Law and new compute architectures.
-Stop coasting after success. Try to do even better. Lunarlake looks like the most promising chip from them in a while, however it has no direct successor. Remember Lunarlake itself is a follower! It brings nothing new, nothing innovative. Basically copying Apple. So if you were so worried to copy it, why not continue it?
They were FORCED to make Celeron, FORCED to make Pentium M, FORCED to make Core 2, FORCED to make Lunarlake. Why should a commercial company making microprocessors be forced to make microprocessors? The two paradigms are antithetical to each other.
-The company Intel is beyond the point of MBAs merely influencing engineers. Their core decisions are driven by them. This needs to change completely. We thought bringing in Pat Gelsinger was enough. It seemed to have made almost no difference.
-The company must be always on the lookout for new opportunities for their main business. What is their main business? Making good CPUs. If they think 1W platform power is low, then try to get it to be 500mW. If they think they have $10 CPUs, look for ways to get it down to $1. Smaller, Lower power, Faster, Better, that is the core of Moore's Law. And out of the four, Smaller + Lower Power benefits disproportionately with Moore's Law and new compute architectures.
-Stop coasting after success. Try to do even better. Lunarlake looks like the most promising chip from them in a while, however it has no direct successor. Remember Lunarlake itself is a follower! It brings nothing new, nothing innovative. Basically copying Apple. So if you were so worried to copy it, why not continue it?
They were FORCED to make Celeron, FORCED to make Pentium M, FORCED to make Core 2, FORCED to make Lunarlake. Why should a commercial company making microprocessors be forced to make microprocessors? The two paradigms are antithetical to each other.
Last edited:
They had multiple hangovers. In that analogy, Intel is an alcoholic in his 50's with serious liver problems and still drinking like a madman. He needs to stop drinking immediately and even that may not be enough, because the "Liver" is at a critical point.
They are beyond the point of people saying "Maybe you should do something about drinking". They have friends about to leave them, and family members worrying about whether he will survive.
misuspita
Senior member
Yeah, but to continue the analogy, when the drunk Intel came home and used to beat the little kid it was ok for them, they had the power and the money. Now the kid grew, it doesn't have the same weakness, it's got the server market on a continued growth path and Intel is suddenly powerless and the old techniques don't work anymore. 😀
We still have Nvidia vs AMD vs Apple. It could be enough to sustain competition and a rising tide.
Neither of them are competing with AMD in x86 though.
My take on x86 is that the users can choose whether to do their computer work on x86 or an alternative. Every year, there are fewer and fewer spaces where x86 is the only option.
There are options, but x86 is still by far the best platform for PC gaming though, which is my primary interest.
NTMBK
Lifer
Qualcomm and Apple are competing in laptops already, and I'm sure other ARM companies will join soon.
Wasn't mentioned already ?HXL posted this on Xitter. The pattern is obvious:
Team | AheadComputing
Meet the architects of the modern data center. Our executive leadership and board bring decades of CPU design expertise from Intel, AMD, Apple, and more.www.aheadcomputing.com
It's been at least a week that the whole net knows about that.
TY for mentioning it here.
Oh boy.... AMD and ARM will have an edge...
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
-
-
