Intel CEO Confirms SMT To Return to Future CPUs

[Vayra86]

No, they really dont have a major power draw issue. Not even a minor one. I dont get this misinformation stuff, really. Intel has consistently for the past decade or more has the cpus with the lowest amount of power draw in the market, its called the t lineup.

You amongst other people are taking K and KS cpus that are MEANT to run balls to the wall and then somehow conclude that intel has a power draw issue, lol. Which isnt even true for k chips, the 285k is drawing the same amount of power as the 9950x in mt workloads and much less in non mt workloads...

They combat their P core power draw with E core efficiency. That alongside the footprint of said P cores; and the vulnerability issues with HT. Its a wild mix of problems for which the e core is their solution. Right now they are stuck between rock and hard place because of that strategy, because the cost of their latest greatest is too high and margins too low. As I tried to say prior - its an economical problem their E core strategy has landed them into; problems that AMD does not have.

Im not saying these are bad CPUs for you or me. They are bad for Intel though. They are grasping at straws to keep the ship afloat. Their strategy has no future and they need heavy readjustment. Their T CPUs are not scalable, and their K versions are power hogs. Whichever way you twist it, they are juggling a massive number of configurations all based on a flawed strategy and it costs them too much.

 

[Tyler-98-W68]

@evernessince

Clearly it hasn't been an effective approach either, as Intel's better getting it's rear whooped in server and regular customers aren't attracted to the latency penalty of the e-cores and the scheduling issues they bring. It'd be an entirely different conversation if the design were actually successful but it isnt'.

Scheduling issues with thread director? Weird how I've had systems from the 12 13 14 and arrow lake generation and haven't touched a single thing related to scheduling, but all over the place people are circle jerking thread lasso and the like programs to make sure their amd chips are "working right" instead of just working out of the box.

it's ok to be a delusional amd fanboy, there isn't an issue with e-cores there never has been and never will. Have another gulp of copium

 

[evernessince]

Scheduling issues with thread director? Weird how I've had systems from the 12 13 14 and arrow lake generation and haven't touched a single thing related to scheduling, but all over the place people are circle jerking thread lasso and the like programs to make sure their amd chips are "working right" instead of just working out of the box.

it's ok to be a delusional amd fanboy, there isn't an issue with e-cores there never has been and never will. Have another gulp of copium

Not perceiving an issue isn't proof that it doesn't exist.

If you get so angry that someone pointed out well reported scheduling issues so as to call them names, well then this is you:

Both Intel and AMD have improved their scheduling over the last couple of years, Intel in regards to it's big little design and AMD in regards to dual CCD. Intel's issue was more pronounced and thus they gained more. AMD's issue was never that big to begin with and only impacted select use cases for dual CCD processors, which represent a minority of the consumer market.

I could understand someone with the opinion that Intel's issues are mostly fixed but to say they never existed at all it an outright and obvious lie.

 

[JustBenching]

They combat their P core power draw with E core efficiency. That alongside the footprint of said P cores; and the vulnerability issues with HT. Its a wild mix of problems for which the e core is their solution. Right now they are stuck between rock and hard place because of that strategy, because the cost of their latest greatest is too high and margins too low. As I tried to say prior - its an economical problem their E core strategy has landed them into; problems that AMD does not have.

Im not saying these are bad CPUs for you or me. They are bad for Intel though. They are grasping at straws to keep the ship afloat. Their strategy has no future and they need heavy readjustment. Their T CPUs are not scalable, and their K versions are power hogs. Whichever way you twist it, they are juggling a massive number of configurations all based on a flawed strategy and it costs them too much.

I partly agree - these are not good CPUs for Intel, but that's not because of anything relating to power. It's the size of the P cores compared to AMDs that are not financially viable - especially if they have to resort to TSMC on top of that. That's the whole point of ecores, it's not about power.

But what do you mean their K cpus are power hogs? Im really intrigued to understand what you mean by that. This is chart with the power draws. AMD is casually hanging up there at the >250w yet no one calls them power hogs. It's an intel only thing, which makes me think that the criticism isn't based on reality but on the brand that's on the box. On a normal desktop usage even if it involves a lot of MT the Intel chip will just consume less cause it's that much more efficient in idle / semi idle workloads. So i dont really get it.

Spoiler

 

[iameatingjam]

They combat their P core power draw with E core efficiency. That alongside the footprint of said P cores; and the vulnerability issues with HT. Its a wild mix of problems for which the e core is their solution. Right now they are stuck between rock and hard place because of that strategy, because the cost of their latest greatest is too high and margins too low. As I tried to say prior - its an economical problem their E core strategy has landed them into; problems that AMD does not have.

Im not saying these are bad CPUs for you or me. They are bad for Intel though. They are grasping at straws to keep the ship afloat. Their strategy has no future and they need heavy readjustment. Their T CPUs are not scalable, and their K versions are power hogs. Whichever way you twist it, they are juggling a massive number of configurations all based on a flawed strategy and it costs them too much.

I mostly agree.... with intel being stuck between a rock and hard place and ecores being a being a part of that. But you do know ecores are called that because they are efficient area wise and not power wise, right? on RL ecores are total power hogs for the amount of work they get done, at least at the higher boost clocks given in 13th and 14th gen. But you can fit 4 of them in the space of 1 pcore and overall they do more work. Without them, intel would have been pretty screwed when it comes to multithreading.

But maybe its me out of the loop, talking about what I know from lga1700 cause I don't know a whole lot about 1851.

Also remember the lag in response time for cpus. Not too long ago it was all about who could have the most cores. Intel couldn't do it with all pcores with their current tech, so they did this instead. It does make sense in a way, when they made their own chips that is. But by the time x3ds came out, at least amongst gamers, cores didn't matter so much anymore. Though ecores are really useful for other, more boring things.

What about this though ( just spit balling). What if ecores are intel building a new type of pcore ( or just core... if they ever ditch the dual design) from the ground up. I mean its hard to do something like that in one gen, it might not work out and then you've burned a lot of money and have nothing to offer ( though intel is good at that too).

Anyway... All you can really do is... small iterations, node shrinks... frequency increases etc. But to completely redesign your cores you'd' need a lot of time. Maybe that was the idea behind ecores. I mean the biggest improvement in arrowlake was the ecore, not the pcore. Develop a new core slowly while keeping the ol' reliable around at the same time. But even if it was... with the constant changes in leadership and disjointed communication and layoffs.... who knows. If it was the idea at some point they may have forgotten already.

Can't say I have much confidence but I really do hope intel can pick themselves back up. Its really in the interest of everybody in the diy space.... including amd.

 

[JustBenching]

I mostly agree.... with intel being stuck between a rock and hard place

In the DIY space Intel will be forever doomed. Not because of how good or bad their CPUs are, it's just that no matter what they actually do, the narrative shifts. When Intel is leading in gaming performance - "who games at 720p with 800$ gpus, right, it's MT performance that matters". For the last couple of years that Intel is offering workstation levels of performance on a 160$ i5 everyone suddenly games at 720p with 3.000$ GPUs and MT is irrelevant, e cores are just cinebench accelerators

In the server space Intel doesn't have an interconnect solution that can compete with AMDs in terms of scaling - their server CPUs just draw an ungodly amount of power just doing nothing cause the interconnects are just not anywhere near as efficient as AMD's. Which is funny actually, cause it's the exact opposite in the desktop market, the whole reason AMD is (Imo of course) lacking in the desktop segment is exactly that, it offers CPUs designed for the server and so in the desktop, they are just drawing 30+ watts just to boot. They are decently efficient in MT workloads but the moment you hit them with something less demanding they will just casually draw twice the power of an Intel part.

 

[iameatingjam]

In the DIY space Intel will be forever doomed. Not because of how good or bad their CPUs are, it's just that no matter what they actually do, the narrative shifts. When Intel is leading in gaming performance - "who games at 720p with 800$ gpus, right, it's MT performance that matters". For the last couple of years that Intel is offering workstation levels of performance on a 160$ i5 everyone suddenly games at 720p with 3.000$ GPUs and MT is irrelevant, e cores are just cinebench accelerators

In the server space Intel doesn't have an interconnect solution that can compete with AMDs in terms of scaling - their server CPUs just draw an ungodly amount of power just doing nothing cause the interconnects are just not anywhere near as efficient as AMD's. Which is funny actually, cause it's the exact opposite in the desktop market, the whole reason AMD is (Imo of course) lacking in the desktop segment is exactly that, it offers CPUs designed for the server and so in the desktop, they are just drawing 30+ watts just to boot. They are decently efficient in MT workloads but the moment you hit them with something less demanding they will just casually draw twice the power of an Intel part.

The goal posts do keep shifting... that is true. I made an edit above about how everything used to be about core count, and that was probably part of the decision making process for ecores, it was just, things changed. Since x3d, amount of cores doesn't seem to matter anymore, even if they are useful for a lot of things that aren't gaming. Infact I'm pretty sure that argument used to be used in favour of amd.

But I suppose we shouldn't veer too off-topic, especially into x brand vs y brand.

 

[RejZoR]

In the DIY space Intel will be forever doomed. Not because of how good or bad their CPUs are, it's just that no matter what they actually do, the narrative shifts. When Intel is leading in gaming performance - "who games at 720p with 800$ gpus, right, it's MT performance that matters". For the last couple of years that Intel is offering workstation levels of performance on a 160$ i5 everyone suddenly games at 720p with 3.000$ GPUs and MT is irrelevant, e cores are just cinebench accelerators

In the server space Intel doesn't have an interconnect solution that can compete with AMDs in terms of scaling - their server CPUs just draw an ungodly amount of power just doing nothing cause the interconnects are just not anywhere near as efficient as AMD's. Which is funny actually, cause it's the exact opposite in the desktop market, the whole reason AMD is (Imo of course) lacking in the desktop segment is exactly that, it offers CPUs designed for the server and so in the desktop, they are just drawing 30+ watts just to boot. They are decently efficient in MT workloads but the moment you hit them with something less demanding they will just casually draw twice the power of an Intel part.

That's absolutely not true. No one believed AMD would ever rise from the ashes like they have with Ryzen CPUs. And I don't think Intel fumbling around a bit was the main reason or Intel's fabs. They took over desktop and they took over servers. Intel absolutely can make a comeback.

 

[Tyler-98-W68]

In the DIY space Intel will be forever doomed. Not because of how good or bad their CPUs are, it's just that no matter what they actually do, the narrative shifts. When Intel is leading in gaming performance - "who games at 720p with 800$ gpus, right, it's MT performance that matters". For the last couple of years that Intel is offering workstation levels of performance on a 160$ i5 everyone suddenly games at 720p with 3.000$ GPUs and MT is irrelevant, e cores are just cinebench accelerators

This is totally it. So tonight I decided I would take my super crappy Core Ultra 5 225 system with fake E-Cores (4 of them) and load the crap out of it and try to play a really demanding game.

Running CB23 and spawning a whole piss pile of traffic.

Weird how i get almost a playable frame rate.

So then I took my AMD system which is amazing because it's X3D and is the bestest because everyone says so. It has 96Gb of ram vs 32GB of the Ultra 225, Same GPU (5070Ti)

But why is it performing less.......I'm playing a demanding game (amd goooooood fer gamez!!!) and doing some rendering, and it's giving me less performances. How is that happening. 16 core 32 threads, vs 10 core 10 threads?

Then we switch to my 285K and 5090.....and its way better

But remember Arrow Lake sucks for everything gaming right.....

 

[JustBenching]

But why is it performing less.......I'm playing a demanding game (amd goooooood fer gamez!!!) and doing some rendering, and it's giving me less performances. How is that happening. 16 core 32 threads, vs 10 core 10 threads?

Mainly it's because the amd hardware doesn't have a thread scheduler like Intel chips have. Ive noticed similar behavior between my 12900k and the 9800x 3d. Unless you use process lasso manually the 9800x 3d craps the bed when you have demanding stuff running on the background.

 

[Tyler-98-W68]

Mainly it's because the amd hardware doesn't have a thread scheduler like Intel chips have. Ive noticed similar behavior between my 12900k and the 9800x 3d. Unless you use process lasso manually the 9800x 3d craps the bed when you have demanding stuff running on the background.

I can't be bothered to waste time with a program when I can just use hardware that works without any bullshit. But wasn't that one guy just saying how bad Intels scheduler was?

 

[Vayra86]

I partly agree - these are not good CPUs for Intel, but that's not because of anything relating to power. It's the size of the P cores compared to AMDs that are not financially viable - especially if they have to resort to TSMC on top of that. That's the whole point of ecores, it's not about power.

But what do you mean their K cpus are power hogs? Im really intrigued to understand what you mean by that. This is chart with the power draws. AMD is casually hanging up there at the >250w yet no one calls them power hogs. It's an intel only thing, which makes me think that the criticism isn't based on reality but on the brand that's on the box. On a normal desktop usage even if it involves a lot of MT the Intel chip will just consume less cause it's that much more efficient in idle / semi idle workloads. So i dont really get it.

Spoiler

Compare the 9950x with the 285k and there is a gap. At the same time that 285k has lower peak performance. AMDs 7950x should be compared to same gen and not Intel's latest. Going lower, these issues persist unless you start tweaking and limiting the cpus further.

 

[Squared]

Compare the 9950x with the 285k and there is a gap. At the same time that 285k has lower peak performance. AMDs 7950x should be compared to same gen and not Intel's latest. Going lower, these issues persist unless you start tweaking and limiting the cpus further.

A 3.4% gap. Once you consider memory can be different, binning variance, and updates that's not much of a gap. Even in games the 9950X lead is tiny; only X3D chips have a gaming edge.

 

[Konceptz]

Either way less than 1% average on all of them showing E-cores make near zero average difference. Sure they migh make a difference on some games, but overall just buy a 9800X3D

I was more so showing that e-cores aren't horrible like most people think they are.
Are great for productivity.

 

[JustBenching]

W

Compare the 9950x with the 285k and there is a gap. At the same time that 285k has lower peak performance. AMDs 7950x should be compared to same gen and not Intel's latest. Going lower, these issues persist unless you start tweaking and limiting the cpus further.

What gap though? The 2 cpus trade blows in performance while the 285k has much better efficiency scaling (im not going to bother posting the data, computer base has it on their 285k review). In the majority of tasks the intel part is more efficient, and sometimes the margin is absolutely massive. As you go down the stack the difference becomes bigger and bigger due to how core deficient amd chips are.

Funny thing is the 7950x draws 7 times more power than the 13900t, ive never seen anyone claim how amd has a huge power draw issue. There definitely is a brand bias behind these kinds of statements cause the data dont really support them.

A 3.4% gap. Once you consider memory can be different, binning variance, and updates that's not much of a gap. Even in games the 9950X lead is tiny; only X3D chips have a gaming edge.

Yeap, which is why i suggested brand bias. Doesnt make much of a sense for someone to dislike arrowlake while liking zen 5. I mean the 9950x spikes up to 70+ watts just browsing the web but what do you know, intel has a power draw issue...

 

[Vayra86]

W
What gap though? The 2 cpus trade blows in performance while the 285k has much better efficiency scaling (im not going to bother posting the data, computer base has it on their 285k review). In the majority of tasks the intel part is more efficient, and sometimes the margin is absolutely massive. As you go down the stack the difference becomes bigger and bigger due to how core deficient amd chips are.

Funny thing is the 7950x draws 7 times more power than the 13900t, ive never seen anyone claim how amd has a huge power draw issue. There definitely is a brand bias behind these kinds of statements cause the data dont really support them.


Yeap, which is why i suggested brand bias. Doesnt make much of a sense for someone to dislike arrowlake while liking zen 5. I mean the 9950x spikes up to 70+ watts just browsing the web but what do you know, intel has a power draw issue...

You may be right about that and for sure there is always a form of bias. It has plagued AMD for a long time, too, with similar minor gaps between products. At the same time, Intel's recent history is painted by a LOT of bad press and that has its influence too I think. Vulnerabilities hit them far harder than they did AMD; power spiking / thermal and IHS issues going back as far as Coffee Lake... they all play their role, fair or not. It was bad enough to spark a whole delidding business in the DIY space.

Its a trust issue before everything else; consistency has been something I also mention wrt AMDs GPU business and how the lack of that hurts them big time. On CPU, AMD has steadily built on consistency and it is paying off now, while Intel rebranded and changed and changed without really moving the status quo.

Lets also not forget how Intel has itself held the better mindshare position for a few generations not too long ago even despite pretty weak generations of CPU going back all the way from Sandy Bridges' success story - AND despite keeping consumers tied to quadcores for a full decade or more. AMD broke that fortress down and customers rewarded it.

 

[Sol_Badguy]

A 3.4% gap. Once you consider memory can be different, binning variance, and updates that's not much of a gap. Even in games the 9950X lead is tiny; only X3D chips have a gaming edge.

2.5% in latest review:

AMD Ryzen 9 9950X3D Review - Great for Gaming and Productivity

The new AMD Ryzen 9 9950X3D brings Zen 5 with 3D V-Cache to the high-end. This new $700 flagship offers the best application performance, beating even the 9950X, and at the same time you get a fantastic gaming experience that's better than any other non-X3D processor on the market.

www.techpowerup.com

What gap though? The 2 cpus trade blows in performance while the 285k has much better efficiency scaling (im not going to bother posting the data, computer base has it on their 285k review). In the majority of tasks the intel part is more efficient, and sometimes the margin is absolutely massive. As you go down the stack the difference becomes bigger and bigger due to how core deficient amd chips are.

Oh, Arrow Lake can guzzle. It doesn't shy away from it:

Intel Core Ultra 9 285K Review

Finally! Intel's new Arrow Lake architecture is launched. The new CPUs are full of design changes, like removal of Hyper-Threading, new Lion Cove P-Cores, an improved Thread Director and more. In our review we got surprising results that were both impressive and disappointing.

www.techpowerup.com

Intel Core Ultra 7 265K Review

The Intel Core Ultra 7 265K offers eight strong Lion Cove cores, just like the 285K, and merely four E-Cores fewer, for an almost 50% price difference. Despite the "Ultra 7" branding it offers plenty of application performance that can beat even the Ryzen 9 9900X. Things don't look so good in...

www.techpowerup.com

Funny thing is the 7950x draws 7 times more power than the 13900t, ive never seen anyone claim how amd has a huge power draw issue. There definitely is a brand bias behind these kinds of statements cause the data dont really support them.

The biggest problem is how it uses much more power than the 7950X3D and it marginally beats it (in application avg.). Efficiency wise it's horrible.
But this has been expressed on this forum in the past, so I wouldn't say that people are trying to sweep it under the rug.

Zen4 was the best gen until now in my opinion, it had it's own issues (mainly temps and 7950X power draw), not perfect obviously, but when I look at the 7950X3D, the 7800X3D, the 7900 and the 7700 I see that efficiency wise they don't really have correspondents in Zen5 except the 9700X.
9950X3D is basically on par with the 9950X (sometimes marginally better) in apps and better in games but the power draw is also on par.
Better relative to the 9950X than the 9800X3D is relative to the 7800X3D.

 

[JustBenching]

The biggest problem is how it uses much more power than the 7950X3D and it marginally beats it (in application avg.). Efficiency wise it's horrible.
But this has been expressed on this forum in the past, so I wouldn't say that people are trying to sweep it under the rug.

The 7950x is an identical CPU to the 7950x 3d minus the 3d cache. It isn't "horrible" in efficiency, it just has a higher power limit. The efficiency curve of the 7950x is identical to the 7950x 3d. We need to start diferentiating between "this cpu is inefficient" and "this cpu has a high power limit". The 7950x is not inefficient, it's the most efficient zen 4 cpu and more efficient than the entire zen 5 lineup (excluding the 9950x). It just has a high power limit.

 

[Sol_Badguy]

@JustBenching

I meant compared to the 7950X3D.

AMD Ryzen 7 9800X3D Review - The Best Gaming Processor

The Ryzen 7 9800X3D establishes AMD as the leader in gaming performance. This Zen 5-based X3D chip is not only fast, it also comes with full support for overclocking. Besides gaming, application performance is considerably improved over the 7800X3D, but that comes at a price.

www.techpowerup.com

Yes, higher power limit, that does what exactly?
Well apparently guzzle power, because the marginally increased performance is certainly not proportional to the difference in power draw.
This is what I'm questioning, why turn an obviously great CPU into a power guzzler just so you can brag that it won?

AMD Ryzen 7 9800X3D Review - The Best Gaming Processor

The Ryzen 7 9800X3D establishes AMD as the leader in gaming performance. This Zen 5-based X3D chip is not only fast, it also comes with full support for overclocking. Besides gaming, application performance is considerably improved over the 7800X3D, but that comes at a price.

www.techpowerup.com

Also affecting temperatures.

This approach of chasing the performance crown no matter what, selling your soul if you have to, or immolation like Raptor Lake is just stupid.
This isn't innovation, it's lack of balance and common sense. AMD isn't spotless in this regard.

 

[RejZoR]

I was more so showing that e-cores aren't horrible like most people think they are.
Are great for productivity.

Der8auer once ran Remnant 2 on E-cores only as a showcase of how games perform. It was pretty okay.

W
What gap though? The 2 cpus trade blows in performance while the 285k has much better efficiency scaling (im not going to bother posting the data, computer base has it on their 285k review). In the majority of tasks the intel part is more efficient, and sometimes the margin is absolutely massive. As you go down the stack the difference becomes bigger and bigger due to how core deficient amd chips are.

Funny thing is the 7950x draws 7 times more power than the 13900t, ive never seen anyone claim how amd has a huge power draw issue. There definitely is a brand bias behind these kinds of statements cause the data dont really support them.


Yeap, which is why i suggested brand bias. Doesnt make much of a sense for someone to dislike arrowlake while liking zen 5. I mean the 9950x spikes up to 70+ watts just browsing the web but what do you know, intel has a power draw issue...

That's because when you buy a hypercar you're not obsessing over "fuel" consumption. That 7950X is a hypercar. That 13900T is a Toyota Prius with a 450 horsepower V6 engine in it. And 13900T is about as exotic in the wild as such Prius would be...

 

[iameatingjam]

You guys this thread is supposed to be about intel's smt. I'm guilty too for helping it get off topic but I feel like this is going a little too far from where it started.