True nature of E-cores and how effective are they?

[dragontamer5788]

data center revenue... does that include commercial desktops/small business desktop type servers or are these included in the Client Computing Group?

Operating Segments

Intel’s data-centric and PC-centric operating segments include technologies and solutions for processing, data analysis, storage, and data transfer.

www.intc.com

Client Computing Group​

CCG creates platforms designed for end-user form factors, focusing on higher growth segments of 2-in-1, thin-and-light, commercial and gaming, and growing opportunities in areas such as connectivity.

So yeah, CCG includes commercial desktops and commercial laptops. "Data Center" is...

Datacenter and AI Group​

DCAI focuses on developing leadership data center products, including Intel® Xeon® server and field programmable gate array (FPGA) products, as well as driving the company’s overall artificial intelligence (AI) strategy.

e-cores do not help in gaming.

Depends on the game, depends on how the code was written. You can write code to do anything. The question is economics. The managers / technical leads choose what to be good with as they write the code, as they test the code, as they debug and iterate the code.

It seems like Apple's and Android's big.LITTLE cores do fine in various mobile-gaming tasks. Because in these situations, power-efficient gaming is a bigger premium than the desktop or laptop world.

 

[Easy Rhino]

Cool but what does it have to do with ecores?

Ecores are meant to improve the wattage/performance ratio.

 

[Assimilator]

Unfortunately for you, I know how to read a 10k document.

https://www.intc.com/filings-reports/annual-reports/content/0000050863-22-000007/0000050863-22-000007.pdf

Page 86, breakdown of revenue by Intel's sectors. Client computing group is by far the largest. Of the CCG portion of Intel, the laptop group makes far more revenue than the desktop group.

Intel's #1 segment, by volume, is the laptop portion of their "CCG" / Client Computing Group. Desktop and Datacenter (and other portions) are smaller.

EDIT: That's 25 Billion to Laptops, and 23 Billion to Data-center, and only 11 Billion to Desktop.

I stand corrected - I was under the impression that data centre has always been the highest revenue generator.

 

[TheoneandonlyMrK]

We only use Windows in our day-to-day lives because of the large back catalog of supported legacy software and the commercial software pledge to it. It's always been glacial in the pace which it adapts to modern computing and it carries decades of baggage in legacy code it simply cannot get rid of. I mean; really, Windows 11 22H2 still ships with the phone dialer application introduced in NT 4.0.

If mostly everything I've ever used wasn't designed straight against Microsoft Windows, i'd be a long-time Linux user by now, and it is in Linux that you should expect to see proper support for the bleeding edge, for the corner cases and for all sorts of wacky hardware that may appear someday. I don't think this is a discredit towards Intel - but rather, towards Microsoft, and even then, it's not like Microsoft can do much about it - you can already imagine the endless whining and complaints if they ever decide to axe software backcompat and limit it to say, apps designed for Windows 8.1 and later only - damned if they do, damned if they don't.



A properly optimized operating system should be able to discern between these two types of architectures and assign tasks suitable for each type of core, Android phones have been doing it for years, and now there's hardware-assisted thread scheduling in Alder Lake so it really should be a matter of software optimization, which... Windows is just not, you step outside of its comfort zone and all things can happen.

Yes but your rant on windows has completely missed the point I made.

There's no rework of scheduler or bios or anything else required.

The E core's work exactly as Intel intended.

They're not bothered about efficiency(or they're efforts border on offensive) so long as ST is competitive and MT is too.

Then job done, mic dropped, party time.

Believe what you want, as will I.

 

[Dr. Dro]

Yes but your rant on windows has completely missed the point I made.

There's no rework of scheduler or bios or anything else required.

The E core's work exactly as Intel intended.

They're not bothered about efficiency(or they're efforts border on offensive) so long as ST is competitive and MT is too.

Then job done, mic dropped, party time.

Believe what you want, as will I.

I didn't rant, though, and I didn't say that the E-cores don't work as Intel intended? I said that Windows doesn't work as it should.

 

[Zach_01]

In my view, the strategy of small cores is directly linked to intel's slowness in advancing in chip manufacturing. With no extra space from a denser lithograph, they needed effective cores per area to match the massive MT performance of the Ryzen line up

This strategy is limited by TDP and heating etc...

I believe so...

Yeah this is the sort of stuff discussed when I initially added "Sometime ago I met a self-confessed AMD jock suggesting intels E-cores are just a poor attempt to overshadow AMD's "core count" which can't be achieved with performance cores alone". He did mention limitations in the chips make-up due to higher temps or power consumption hence e-cores are somewhat just filling the gap.

Can't complain though, with or without e-cores ADL's done a fantastic job... it would be interesting to see if e-cores play a more refined role in the long run especially if Intel sticks with monolithic designs... although not sure if this is correct, are earlier rumours of Meteor Lake moving to multi chip modules (MCM) now official or is intel playing DIE HARD with mono-bono-4-life?

Personal opinion...

Intel is struggling to get their node(s) straight for more than 5 years now, and TSMC is waving them from afar...
If Intel could, the E-cores concept wouldn't be on the table for at least another 5, maybe more years. They had to be creative and looked into mobile (phones/tablets and such) industry to their solution.

E-cores work... dont work as intended... Who knows what really Intel and AMD intend... we can only assume and we ought to not believe what they are serving us on their advertising.

The thing is at the end that ADL, as benchmarks show, have the performance to compete. At the cost of power... yes as they are on 10nm (they can call it whatever they like BTW, its still 10nm) against the 5/7nm nodes. Most likely the could do a better job on efficiency but thats not what today is about. Competition is good but sometimes not... And you will understand what Im saying further down.

MCM design wont/cant help them to reduce power. They still have the 10nm at their disposal for the moment and so much they can do.
Only to make fabrication a little more cost effective as smaller dies have better yields. But these things cant be done from 1 day to another.

AMD has taken the MCM path clearly for:
1. The unified design across all segments
2. Smaller dies

Both lead to better profit margins and not better efficiency.

If they haven't had the 5/7nm nodes from TSMC they would be on exactly the same position with intel, power wise. But AMD is working hard to catch up... lol

Hence...

Another matter (or not) that I haven't really see on discussions, after presentation of AM5...
That AMD is catching up on Intel on power. Top tier desktop CPU is now in the 200W (I say 200+W) territory.
We can understand that when:
A 105W TDP CPU has 140+W PPT it means that a 170W TDP CPU has.... ???W PPT... 200W?... 220W?

Those slides suggest the following to me...






That 7950X will have a 220~230W PPT
If you do the math combined on those 2 slides and the fact that 12900K is a 240W PPT, you will find it.

Along with Intel, nVidia, Radeon... we can now welcome AMD CPUs to the inefficiency in the name of competition.

I wonder what will stop them...


Edit: typo(s)

 

[MrJiggyDancer]

View attachment 260079
So with E-cores disabled, 8P cores get 20K scores, that 2500points per P-core
8E cores take up the same die space as 2P cores, let say Intel make 10P cores, that would get them 25000 points, meanwhile 8P+8E get them 27700 points.

There are non-K version for people who prefer better effieciency at stock (much cheaper too), and have no clue how to tune their PC. Only idiots pay more for K version and not tune their PC to how they like it.

Apples to oranges. You're comparing overclocked P+E scores to power limited P-only score.

 

[MrJiggyDancer]

So Intel's long term strategy, which consists of parallel core development, which they are specifically betting on with many (if not all) their upcoming architectural designs with, including tile based chiplets, is also to, wait for it, completely discontinue optimizations for said designs?

Interesting opinion.

No my argument is Intel won't put that effort in.
It is not how they are using those core's.

This isn't arms big little, there's work's.

I'd argue even ARM's implementation is flawed. Do we really have any evidence that b.L really adds any efficiency in their architecture?

 

[MrJiggyDancer]

I stand corrected - I was under the impression that data centre has always been the highest revenue generator.

12th gen laptops do not have better battery life than 11th gen laptops...

 

[Gungar]

I'd argue even ARM's implementation is flawed. Do we really have any evidence that b.L really adds any efficiency in their architecture?

Efficiency in price (way less expensive than real 8 core cpu), in energy consumption probably not or very little difference.

 

[Bomby569]

A CPU is a package, for the consumer what matters is the performance not the tech. Intel had to do this because they were running out of space on the die and the thermals problem. It's a win for them. A necessity for sure, but many breaktroughs come out of necessity.

AMD seems to be doing fine without them (trusting their own words), the new Ryzen will have a smaller die.

Two different approaches, none are wrong or right. It's actually better they are trying their own thing and not just copying each other. For the consumer it's not relevant at all.

 

[Panther_Seraphin]

Its back to the days of Athlon XP era with the Ghz war

We have just moved from Raw speed to Core counts

Imagine trying to market a 10 Core part vs a 16 Core part to the average Joe at the same prices. Who do you think people are going to for? Or more Accurately 4-6 core Intel vs 6-8-12 core AMD

I can actually see there being an advantage in a Server environment in certain designs. Think storage or VM hosting where getting 80% of the core performance for 50-60 die size and power draw will be big positives. Or where PCI-e lanes are more beneficial and you arent needing the latest and greatest CPU performance. I believe this is where Bergamo is possibly being aimed at with 128 "E" cores instead of the 96 "P" cores of Genoa.

 

[Vario]

View attachment 260079
So with E-cores disabled, 8P cores get 20K scores, that 2500points per P-core
8E cores take up the same die space as 2P cores, let say Intel make 10P cores, that would get them 25000 points, meanwhile 8P+8E get them 27700 points.

There are non-K version for people who prefer better effieciency at stock (much cheaper too), and have no clue how to tune their PC. Only idiots pay more for K version and not tune their PC to how they like it.

It is nice to have onboard video. The price is close in the real world between the F and K.

 

[Mussels]

So...help me here.

You're responding to someone saying that you are not using your 12700k. Your response is that you disabled a large chunk of the silicon...thus literally disabling the components that this thread is meant to discuss.
You then link to a video that compare a 12900F and 12900k. One of the video's conclusions is that there's a 25% difference in power draw between the F and K SKUs...and the performance difference is 0-4%. So...the result is that you pay more for the k, you have a much higher power draw, and you have a boost that is functionally within the error for the testing methodology to be reasonably chalked-up to regular process variation. You then say you can get the performance of the k to that of the f by disabling the nice shiny new E cores...and experience an uplift by tuning...despite the literal cited video stating more frequency<>better performance???

You have to be trolling...right?

This is the trouble i've been having discussing this on the forums (and keep getting called a fanboy, a shill, etc etc)
If you need to change almost every aspect of the stock behaviour, that's not some elitist awesome superduper thing to brag about - it's a sign the product is bad.


No one argues intel don't have great single threaded performance - the problem is that they're making you pay for more and more terrible E-cores to get it, and then to sustain it you need to disable the E-cores to keep the TDP down...

 

[nguyen]

This is the trouble i've been having discussing this on the forums (and keep getting called a fanboy, a shill, etc etc)
If you need to change almost every aspect of the stock behaviour, that's not some elitist awesome superduper thing to brag about - it's a sign the product is bad.


No one argues intel don't have great single threaded performance - the problem is that they're making you pay for more and more terrible E-cores to get it, and then to sustain it you need to disable the E-cores to keep the TDP down...

LOL yeah let play pretend that Intel non-K series don't exist , as those non-K model require zero tweak to achive great performance and efficiency, and then some people with 5800X come in and say they can tweak their 5800X to achieve better efficiency

 

[Vayra86]

LOL yeah let play pretend that Intel non-K series don't exist , as those non-K model require zero tweak to achive great performance and efficiency, and then some people with 5800X come in and say they can tweak their 5800X to achieve better efficiency

Non K also likes to boost over 5 Ghz and RPL is going further in that direction.

Intel® Core™ i9-12900 Processor (30M Cache, up to 5.10 GHz) - Product Specifications | Intel

Intel® Core™ i9-12900 Processor (30M Cache, up to 5.10 GHz) quick reference with specifications, features, and technologies.

www.intel.com

Nobody is playing pretend here, the max turbo on this non K part is 202 W. Meanwhile this is supposed to be a '65W TDP' CPU like the old days.

So tell me, as a novice that knows the old Intel 65W non K CPUs, how does this power behaviour match my cooling solution scaled to 65W? @Mussels is fully correct here. When you buy Ryzen, you get a CPU that by default has full customization options, much like AMD CPUs had historically. But people buy Intel because 'its good out of the box'...

 

[nguyen]

Non K also likes to boost over 5 Ghz and RPL is going further in that direction.

Intel® Core™ i9-12900 Processor (30M Cache, up to 5.10 GHz) - Product Specifications | Intel

Intel® Core™ i9-12900 Processor (30M Cache, up to 5.10 GHz) quick reference with specifications, features, and technologies.

www.intel.com

Nobody is playing pretend here, the max turbo on this non K part is 202 W. Meanwhile this is supposed to be a '65W TDP' CPU like the old days.

So tell me, as a novice that knows the old Intel 65~77W non K CPUs, how does this power behaviour match my cooling solution scaled to 65W? @Mussels is fully correct here. When you buy Ryzen, you get a CPU that by default has full customization options, much like AMD CPUs had historically. But people buy Intel because 'its good out of the box'...

Show me 1 review where 5800X can achieve its advertised performance with a 65W cooler then, I bet you can't

Meanwhile 12700 with stock 65W cooler review, pretty much matching 5800X (with a better cooler) in performance and came out with better efficiency.

Then the discussion devolve to people can tweak their 5800X, etc....which is just coping

 

[Vayra86]

Show me 1 review where 5800X can achieve its advertised performance with a 65W cooler then, I bet you can't

Meanwhile 12700 with stock 65W cooler review

I don't care about how it compares to AMD, we're talking about what the CPU does and 'how it works' within 65W.

Your very own review link shows up to 28% perf loss when the 65W limit is enforced and running on the RM1 stock cooler. In every benchmark there is a noticeable loss of performance. The CPU throttles when the limit is above 65W. Stock however is not limited to 65W peak.

Just stop the red/blue pissing contest for a minute, god almighty.

 

[nguyen]

When you buy Ryzen, you get a CPU that by default has full customization options, much like AMD CPUs had historically. But people buy Intel because 'its good out of the box'...

I don't care about how it compares to AMD, we're talking about what the CPU does and 'how it works' within 65W.

Your very own review link shows up to 28% perf loss when the 65W limit is enforced and running on the RM1 stock cooler. In every benchmark there is a noticeable loss of performance. The CPU throttles when the limit is above 65W. Stock however is not limited to 65W peak.

Just stop the red/blue pissing contest for a minute, god almighty.

Hm...I sense some sort of hypocrisy here, may be just the wind.

12700 achieve commendable performance with its stock cooler, but it's bad because it lost performance vs maxing it out, and also maxing it out is bad because it lose efficiency, what a stupid argument

 

[Vayra86]

Hm...I sense some sort of hypocrisy here, may be just the wind

You're the one bringing up the argument about non K being nicely usable untweaked? I'm showing you that you'll lose a whole lot of performance that way, which is the point @Mussels was making. So 'stock settings' != 'stock performance' at all.

The fact this also goes for Ryzen, I'm not disputing at all...

 

[nguyen]

You're the one bringing up the argument about non K being nicely usable untweaked? I'm showing you that you'll lose a whole lot of performance that way, which is the point @Mussels was making. So 'stock settings' != 'stock performance' at all.

Try cooling 5800X with a 65W Wrath Stealth cooler then

 

[Vayra86]

Try cooling 5800X with a 65W Wrath Stealth cooler then

That's a 105W part, not 65.

 

[nguyen]

That's a 105W part, not 65.

But the "242W" 12700 run just fine with a 65W cooler? matching 5800X performance?
Are you coping?

 

[Vayra86]

12700 achieve commendable performance with its stock cooler, but it's bad because it lost performance vs maxing it out, and also maxing it out is bad because it lose efficiency, what a stupid argument

See, this is where its all going wrong: NOBODY is saying the 12700 is bad. Or any other Intel ADL part.

The point is the massive gap between advertised TDP of 65W and peak of 202, or higher in the upcoming gen, and how this affects stock settings and user experience. The gap on competition is smaller and the clock behaviour is less bursty as a result, which also affects cooler requirements.

Ryzen is going in a similar direction and I dislike it just as well.

But the "242W" 12700 run just fine with a 65W cooler? matching 5800X performance?
Are you coping?

Can you grow up a little, maybe?

 

[nguyen]

See, this is where its all going wrong: NOBODY is saying the 12700 is bad. Or any other Intel ADL part.

The point is the massive gap between advertised TDP of 65W and peak of 202, or higher in the upcoming gen, and how this affects stock settings and user experience. The gap on competition is smaller and the clock behaviour is less bursty as a result, which also affects cooler requirements.

Ryzen is going in a similar direction and I dislike it just as well.

If you don't know how to make your equipments work as you intended, the problem is on you.