Intel's Entry-level Core i3-12100 "Alder Lake" Beats Ryzen 3 3300X Comfortably

[RandallFlagg]

The real question here is why AMD never released ryzen 3 with their 5xxxx generation. Is it because they wanted to maximise profits with targeting limited number of chips to higher price processors? But a budget oriented company that survived with selling 200 usd processors would not abondon their core base surely ? Or would they ?

I think AMD is focusing on higher margin products as much as possible with the fab allocation it has, without completely abdicating in any market.

It makes total sense from a business standpoint.

 

[Tomorrow]

The real question here is why AMD never released ryzen 3 with their 5xxxx generation. Is it because they wanted to maximise profits with targeting limited number of chips to higher price processors? But a budget oriented company that survived with selling 200 usd processors would not abondon their core base surely ? Or would they ?

Simple economics. The 7nm yields are so good that it made little sense to disable half of a working 5800X silicon to make 5300X for example. Essentially AMD became victim of it's own success in the low end. Im sure given the option they would love to flood the OEM market with cheap Ryzen 3's that are Zen3 based. But it has to make financtial sense.

 

[Patriot]

Simple economics. The 7nm yields are so good that it made little sense to disable half of a working 5800X silicon to make 5300X for example. Essentially AMD became victim of it's own success in the low end. Im sure given the option they would love to flood the OEM market with cheap Ryzen 3's that are Zen3 based. But it has to make financtial sense.

Yeah, they dont have the 2-3B to burn on marketing like Intel, they have to make profits where they can.

 

[Vayra86]

new architecture product beats old architecture product
THIS IS INSANE! HOW IS THIS POSSIBLE?!

Well, Intel did build up some recent history where they couldn't meet that spec, so every question is valid by now

 

[RedBear]

By the time this 12100 actually coming out and, with budget MB actually available, The 3300x would be a 2 year-old processor.
Great victory intel, Comfortably ?

It's a great victory for consumers, if anything else, a new affordable CPU with decent performance in the middle of a CPU shortage is nothing to laugh at. I see this argument about the 3300x being nearly 2 years old everywhere, but the actual issue is that AMD doesn't care about this segment, where there's little money to make, and at best it's going to throw away some old Renoir APUs with defective graphics. If those rumoured Renoir-X are confirmed eventually we will compare the Alder Lake S i3s against those "new" CPUs based on the Zen 2 architecture and everyone will be happy (or not, maybe AMD apologists will find another way to grab at straws).

 

[Chrispy_]

From a recent findings from hardware unboxed, it is the other way around.
There are something seriously lacking / cut from the E-cores.
Depending on workloads,
Sometimes it matches skylake, but sometimes it performs as worse as the bulldozer
E-cores performance just isn't stable.

You do realise that video is testing 4C/8T on the P-cores vs 4C/4T of the E-cores? The games it really fails at are the same games the old 4C/4T i5 CPUs also fail at today because many games now require more than 4 threads.

Intel aren't making any zero P-Core products anyway, so what matters is the "hybrid" mode HWU tested with 2P and 2C - those results are respectable, if not earth-shatteringly fast, but this is a budget i3 we're talking about, not $1200 of i9/Z690/DDR5.

The other point I want to make is that for IGP gaming, even the E-core only configuration would have been fine. Yes, in worst-case scenarios the minimum framerates were less than half of the P-core results but the 7-game average still had over 60fps 1% lows. The IGP will never require that sort of performance, it'll be unlikely to do much better than 1080p30 average in AAA games.

 

[QuietBob]

The 3300X is a great little piece of silicon, I have mine at 4.5 GHz all core and it performs admirably. In typical use (excepting compute-heavy workloads) one would be hard pressed to tell the difference between this lowly quad and any current CPU. In games it performs virtually the same as all Zen 2 chips, thanks to its full complement of L3 cache.

The problem is, the 3300X was nearly impossible to get at launch, and hasn't seen a successor to this day. I was lucky to get mine in the end, but already at a premium. And given its scarcity, the price eventually reached the level of the Ryzen 5 3600, at least where I live.

Intel, on the other hand, launched their i3-10100 around the same time, and hasn't had any problems keeping the supply steady. It is a slower chip overall, but at the current price of EUR 75 / USD 85 for the "F" variant (inc. 23% VAT), it offers exceptional bang for the buck.

As much as I support AMD, at the moment I couldn't honestly recommend any of their CPUs to someone on a tight budget. The i3-10100 and the i5-11400 at EUR 150 / USD 170 (inc. tax) make a lot more sense if you can't afford Zen 3.

I'd love to see a 5300X from AMD, but this is highly unlikely to ever happen. It is good that Intel is launching the i3-12100. We need more CPUs for the budget buyer, and a 4c/8t SKU with a powerful single thread will cover the needs of a vast majority of users.

 

[Minus Infinity]

Not sure why AMD abandoned this space with Zen 3, but smart move from Intel to muscle in on this turf. A Ryzen R3 5400 4c/8t should have been a priority IMO. More people buy at the lower end than top end despite what forums would have you belive. Also, AMD has abandoned HEDT for this gen, they have cancelled Chagall except for possibly one model and Intel is ready to unleash Sapphire Rapids Workstation CPU's to move in on turf Threadripper had to itself.

 

[T_Zel]

As much as I support AMD, at the moment I couldn't honestly recommend any of their CPUs to someone on a tight budget. The i3-10100 and the i5-11400 at EUR 150 / USD 170 (inc. tax) make a lot more sense if you can't afford Zen 3.

It has been tough to recommend AMD for budget builds for quite a while now. Between the AMD price inflation and Intel price reductions, the 10400 and 11400 have been my go-to recommendations in that space pretty much all year.

Not sure why AMD abandoned this space with Zen 3

I'd say it's a combination of factors. TSMC doesn't have the 7nm wafer capacity (and AMD doesn't have enough allocation) to keep up with current demand for consoles, CPUs and GPUs as it is. Then the yields on mature 7nm are almost certainly fantastic, in which case AMD would have to disable cores on chiplets that would otherwise be a 5600X at worst to create a 4 core product for the budget segment. When they're selling through of their more expensive products that use the same chiplets, deliberately gimping some to create lower end products is like pissing money down the drain. This is likely why the 3300X was also so hard to find, there just weren't enough chiplets with defective cores to justify making many of them when the 3600 was already selling like hotcakes.

If they had enough capacity to meet demand of 5600X and up, with chiplets to spare afterwards, then we'd start to see those lower end 4-core models. But they don't, and probably won't until Intel's 12th gen starts sucking up enough of the market demand. Though if 3D v-cache is as much of a gaming improvement as advertised (and the pricing is reasonable), you can probably write off any chance for budget Zen 3 SKUs.

 

[Crackong]

You do realise that video is testing 4C/8T on the P-cores vs 4C/4T of the E-cores? The games it really fails at are the same games the old 4C/4T i5 CPUs also fail at today because many games now require more than 4 threads.

Intel aren't making any zero P-Core products anyway, so what matters is the "hybrid" mode HWU tested with 2P and 2C - those results are respectable, if not earth-shatteringly fast, but this is a budget i3 we're talking about, not $1200 of i9/Z690/DDR5.

The other point I want to make is that for IGP gaming, even the E-core only configuration would have been fine. Yes, in worst-case scenarios the minimum framerates were less than half of the P-core results but the 7-game average still had over 60fps 1% lows. The IGP will never require that sort of performance, it'll be unlikely to do much better than 1080p30 average in AAA games.

Lacking HT is also a feature lost in E-cores.
I would say the comparison is fair enough.

Someone might do 4C8T P-core vs 8C8T E-core comparisons in the future, but is it "fair" ?

I would say the 6+0 12600 would make a clear picture if the E-cores are really worth it or just trouble makers.
Since all the major reviews are showing E-cores are hindering the overall CPU performance in some ways
Performance dip in switching cores is annoying.
Some applications just right up crashing using E-cores only
And the Windows scheduler is out of control making things worse.
You will never know when , or how scheduler decides to throw your "important" things into E-cores and end up slowing / crashing, unless you manually assign them core by core, but who would do that on a daily basis?

It is clearly a major problem when Intel themselves have to present "Parking the E-cores away" as a solution.

 

[RedBear]

Not sure why AMD abandoned this space with Zen 3, but smart move from Intel to muscle in on this turf. A Ryzen R3 5400 4c/8t should have been a priority IMO. More people buy at the lower end than top end despite what forums would have you belive.

It's a consequence of the chip shortage to a large extent, Lisa Su herself said as much some time ago. It's also possibly related to the move from 4 cores per CCX in Zen 2 to 8 cores per CCX in Zen 3, probably it's become even more difficult to get enough defective CPUs that could be repurposed as a 4C/8T and sacrificing perfectly functioning 5600x/5800x for a 4C/8T 5400X (or however they'd call it) makes no sense whatsoever in the current market.

 

[Chrispy_]

Someone might do 4C8T P-core vs 8C8T E-core comparisons in the future, but is it "fair" ?

It's not about "fair" it's about use of die area which directly impacts the cost of the chip. Big chips cost exponentially more than smaller chips.

If you want to consider "fair" in terms of die area, the fair comparison is 1C/2T of P-core vs 4C/4T of E-core. Realistically neither will do well as there are plenty of apps that fail to even run on only 2 threads and performance hurts if you don't have at least 8 threads in enough titles for 8 threads to be a relevant number these days. Scale it up a bit though and you can have a competent 8 E-cores for the same die area as just 2C/4T of P-core.

At any rate, I was proposing (like Intel clearly have in their slides) a minimum config of 2P and 8E cores. It's not like anyone's suggesting Intel make CPUs using only E-cores. The only person who even brought that up is you and that was specifically in an article that artificially looks at just E cores solely for the purposes of gauging their relative performance. The E-cores are not actually supposed to be used in isolation.

 

[Crackong]

If you want to consider "fair" in terms of die area, the fair comparison is 1C/2T of P-core vs 4C/4T of E-core. Realistically neither will do well as there are plenty of apps that fail to even run on only 2 threads and performance hurts if you don't have at least 8 threads in enough titles for 8 threads to be a relevant number these days. Scale it up a bit though and you can have a competent 8 E-cores for the same die area as just 2C/4T of P-core.

if you wanna consider "fair" in terms of die area
Please check the Intel official die shot for actual die size.
A 4 E-core cluster is much "wider" than a P-core

1x 4 E-core cluster is about 1.5 P-core
Since they don't do odd number cores anymore
They "fair" comparison in your mind would be 6 P-core vs 16 E-cores, which is not available in any current configuration and therefore impossible to test right now.
So
3 P-cores vs 8 E-cores maybe?
But who would do that test?

 

[Chrispy_]

Please check the Intel official die shot for actual die size.
A 4 E-core cluster is much "wider" than a P-core

Well, duh, if you include bits of the uncore interconnects, sure - it's 36% wider (146 pixels vs 107 pixels), but you're also conveniently ignoring that it's 10% shorter to suit your argument.

Ignoring the uncore the die area is more like 5:1 E-cores to P-cores but that's just as pointless as measuring the width of the core logic block because there's tons of support logic that isn't counted for both P and E cores so the 4:1 figure is an approximation that has been thrown around a lot because Intel have been saying it themselves since before Alder Lake even launched, including the word "approximately".

I'm not sure why you're getting so hung up on the exact size ratio anyway. Whether it's 3.3x smaller or 5.1x smaller than a P-core is irrelevant when it averages about 70% the performance for somewhere between 20-35% the die area.

Not every application requires maximum performance on all threads. This is Microsoft's first 'big.LITTLE' scheduler and Intel's first draft of the Thread Director, of course there are some teething troubles, just like there were with actual big.LITTLE for ARM-based smartphone SOCs half a decade ago. But once the performance-critical processes are running on the P-cores without fail we'll hopefully see more of the background OS and application bloat running on E-cores with vastly-reduced power consumption and using far less silicon real estate. That power budget can be used for higher-clocks on P-cores where it matters, or better battery life/quieter fans in laptops. As for the real-estate, that will translate to lower production costs and thanks to healthy competition between Intel and AMD that will translate to cheaper CPUs for us. It's a win-win situation if Alder Lake's hybrid architecture becomes the new normal - the only downsides are these teething troubles and they were expected and, quite honestly, not as bad as I thought they were going to be.

 

[Crackong]

Well, duh, if you include bits of the uncore interconnects, sure - it's 36% wider (146 pixels vs 107 pixels), but you're also conveniently ignoring that it's 10% shorter to suit your argument.

Ignoring the uncore the die area is more like 5:1 E-cores to P-cores but that's just as pointless as measuring the width of the core logic block because there's tons of support logic that isn't counted for both P and E cores so the 4:1 figure is an approximation that has been thrown around a lot because Intel have been saying it themselves since before Alder Lake even launched, including the word "approximately".

I'm not sure why you're getting so hung up on the exact size ratio anyway. Whether it's 3.3x smaller or 5.1x smaller than a P-core is irrelevant when it averages about 70% the performance for somewhere between 20-35% the die area.

Not every application requires maximum performance on all threads. This is Microsoft's first 'big.LITTLE' scheduler and Intel's first draft of the Thread Director, of course there are some teething troubles, just like there were with actual big.LITTLE for ARM-based smartphone SOCs half a decade ago. But once the performance-critical processes are running on the P-cores without fail we'll hopefully see more of the background OS and application bloat running on E-cores with vastly-reduced power consumption and using far less silicon real estate. That power budget can be used for higher-clocks on P-cores where it matters, or better battery life/quieter fans in laptops. As for the real-estate, that will translate to lower production costs and thanks to healthy competition between Intel and AMD that will translate to cheaper CPUs for us. It's a win-win situation if Alder Lake's hybrid architecture becomes the new normal - the only downsides are these teething troubles and they were expected and, quite honestly, not as bad as I thought they were going to be.

Well, the cluster design is a key feature and contribute to many of the E-core characteristics.
You are going to throw all the interconnects and the caching structure away? oh good luck with that.

It is you who mentioned die area first and now you went back to "approximately 1:4", oh boy.

Anyway

I 've stated my point very clear at #59
All the major reviews are showing E-cores are hindering the overall CPU performance in some ways.
And Intel themselves have to present "Parking the E-cores away" as a solution to evade these problems.
Big.Little on x86 just isn't ready.
Don't beta test for free.

 

[Why_Me]

Ulike AMD, Intel shows it cares about people on a budget.

 

[watzupken]

I feel it is no surprise that Alder Lake i3 should easily beat the Zen 2 based Ryzen processor, as we have seen it beating the Zen 3 processors as well. I guess AMD is not looking to compete at the lower end now because of limited allocation from TSMC, and also the cost involved. Given the lack of options for quad core processors over the last few years, I think most people buying a gaming system would have moved up to a 6 core processor since it is proven that some games prefer a higher core count than 4.

 

[Valantar]

Ulike AMD, Intel shows it cares about people on a budget.

"Cares" is... strong. It shows that they have more production capacity than AMD and can thus more easily allocate some of that to lower end hardware. It also shows that they know to jump on a wide open market when they see it. I sincerely doubt there is any degree of altruism or caring involved.

 

[ratirt]

I dont calculate CPU upgrades, because frankly CPUs are not something you should be switching out every year or two. A good CPU will last damn near a decade while still providing enough performance. If you play at 60 FPS, like most gamers do, then the i5s of the sandy/ivy era are just now getting to the point where they can no longer maintain 60 FPS for 1% lows even with an OC on them. If you're a more casual gamer or building a PC for mroe then just gaming, then the CPU tends to last even longer. By the time you want to upgrade, the platform has also evolved significantly.

Fans of AMD will point out upgradeability, I'll point out that the 8700k made available shortly after the ryzen 1000 launch is still competitive in games with the 5600x, 4 years later. If you bought into intel, there is no need to upgrade to more cores, you could if you wanted to, but frankly there's no need. By the time you have the money and want to upgrade, there will be a new generation of budget hardware with new features you might want, such as better RAM compatibility, faster PCIe, integrated wifi 6, whatever.

Low demand isnt how I would describe it, seeing as the ryzen 3300x sold out immediately and even the 3100 was hard to come by. People can blame the shortages, but intel during their "chip shortage" a few years back still launched and maintained a library of lower end models. They might not have had as much stock as would have been nice, but they didnt just ignore the market like AMD has been doing lately.

Shortage or no, you always address the higher margin stuff first, then move to lower offerings. What you dont do is completely ignore the market, as that allows your competitor to scoop up all those sales unopposed.

Oh you dont calculate. So you clearly twist and turn to get your way. Why even discussing anything? Fans will point out? You have clearly issues with someone liking a product for what it is and if that's the case he must be a fan (FAN a code for FANboY) which clearly is an attack at my way thinking or an opinion implying I'm a fan nice bro.
You should calculate upgrade path since it is way cheaper to upgrade just a CPU and it doesnt have to be a year long CPU. People are still on Ryzen 1000 series or even 6th gen Intels and they are fine.
Not sure where that 1 year CPU change come from. If you say there is not need to upgrade you are only referring to yourself. Cause the only thing you do is play games. Well news flash buddy, people do more than just play games so there is a need for more cores otherwise nor Intel or AMD would go that route.
By the time you have money and want to upgrade? With that logic you will never buy anything. There is always something new and considering Intel, 2 CPU lines just this year so go figure.
They sold out cause there were is short supply and you already know why I told you about it in my previous post.
What you don't do is criticize a billion dollars companies saying you know better with hundreds of advisors and people working for the company. What they did is probably best for the company anyone would ever do. So I totally disagree with that statement. It's just plain arrogant from your site. What you can tell for sure is, that maybe, AMD had to cut a product from production and they chose this product due to chip shortages.

 

[Chrispy_]

All the major reviews are showing E-cores are hindering the overall CPU performance in some ways.
And Intel themselves have to present "Parking the E-cores away" as a solution to evade these problems.
Big.Little on x86 just isn't ready.
Don't beta test for free.

Yeah, no disagreement there.

Alder Lake's hybrid architecture, DDR5, and Windows 11's scheduler are all beta tests. I won't mock people for wanting to play with the latest toys if they have the disposable income to do so but I do it for a day job so when it comes to my own money and recommendations to others I generally avoid the early-adopter/paying beta tester options.

Hopefully by the time I actually want to use W11 at home rather than dealing with the enterprise deployment headaches at work, hybrid architectures play nice with the scheduler and I'll be able to pick up a CPU from Intel or AMD that has whatever the optimum number of P-cores is with as many E-cores as they can fit on the remaining die area. I would happily pay a grand or so for something like a 6C/12T+24c solution, provided that the teething troubles we're currently seeing are resolved.

 

[Mussels]

Intel 12th gen beats 3 generation old AMD cpu!



I mean yes, we want fast budget chips.
But I'm glad this wasn't some major press release or anything, it's more of an expectation that if it wasn't at least this good, it'd be a massive failure.


I'd be quite curious how it goes against the 5600g, the cheapest zen 3 chip

(Edit: my god my spelling was garbage when i wrote this)

 

[N3M3515]

hahahahah, this is so funny, Intel I3 12100 beats, AMD NON EXISTENT cpu. It was available like for 5 minutes...

 

[Mussels]

hahahahah, this is so funny, Intel I3 12100 beats, AMD NON EXISTENT cpu. It was available like for 5 minutes...

Yeah, it's a pity there was stock issues with them.

That said, the 1600AF is in stock at the same price and the 3600 OEM was and is, barely higher priced
(At least here in Au)

 

[Caring1]

I thought the R3 3300X was meant to be OEM only but they released a few to the public too?

 

[Mussels]

I thought the R3 3300X was meant to be OEM only but they released a few to the public too?

More or less, but here in Au i can get almost every other OEM model
These are great chips, great prices. But there is literally only the 1600AF cheaper ($200 Au)