Is Intel going to deliver a processor/chipset worth waiting for?

[dgianstefani]

Dr. Dro said:
You clearly misunderstood. 12th Gen is a different type of core - Alder Lake, with a different P-core architecture. However, 13th and 14th gen are bit by bit, physically identical processors with zero changes or improvements in between them. The sole exception is the configuration with 3 E-core clusters sold as 14700K - which was possible in 13th but never made commercially available. They have the same capabilities, characteristics, internal model number, revision, etc.

Not true, since there are substantial temperature improvements between 13th and 14th gen, even with same power limits and boost clicks etc. So while some specs on paper are identical, there's hardware differences.

Gica said:
This is a simulation of reading this topic. From the first post to the last, all read carefully. It took 5 minutes 30 seconds.
Statically, the processor consumes 1-1.2 W.
With scrolling, it jumps up to 2.4 W.
Moving to the next page brought a peak of 10W.
If I spend an hour reading (forum, history, news, etc.), the processor consumes less than 2W.
Another hour for youtube 1080p, the same processor has an average of 5.5W.

It should be noted that this consumption also includes the IGP consumption. The consumption for reading a forum and playing youtube in parallel is the same as the power consumption of a dedicated entry level video card in idle.
It is absolutely ridiculous to talk about the consumption of a processor, now, when it offers computing power far beyond the needs of a home user. Yes, everyone needs a Ferrari for the streets of Calcutta.

View attachment 337682
Click to expand...

Yep. Whereas the "more efficient" AMD CPU I have idles at 30-40 W since IO die is always active. You can get it down to 25-30 W but that requires crippling performance with slow RAM, IF etc.

Backside power delivery along with gate all around ribbon transistors are both huge advances. Zen 5 is architecture improvement but no packaging changes, so idle issues will continue.

Zen 6 is supposed to introduce better packaging and maybe a new IO die which will hopefully bring Zen desktops into modern idling numbers with much better low load efficiency. Not everyone is running prime 95 or cinebench all day, in fact normal use, and even most gaming is much closer to idling than full load.

 

[N/A]

The one worth waiting for should be well thought out and future proof. I see no point in dropping 299 on a motherboard that covers only 2 gens let alone everything else. Io die not idling is an engineering problem. You'd think that by now they know how to park idling transistors completely. Do they even care.

 

[nomdeplume]

Let me be clear, personally I don't have a problem with all that. I'm an enthusiast enough to take everything apart and improve what I can, and I'm also able to cope with any risk involved in the process. I'm not afraid to try new things. Couple of weeks ago I used LM for the first time, and on the most expensive piece of a hardware I ever had. And it was easier than applying the damn mx-6, if not counting the 4 layers of nail polish for the SMDs.
The thing is I like my PCs cool & quiet, I like having headroom to OC and tweak the hell out of my hardware, and saying "90°C is fine", just because you failed at making it better at the factory, makes me wanna puke. It's the same shit Intel was doing with their TIM, and everyone was fine with it because it just works, just as the mostly useless E cores do.

Oh, believe that I'm not clueless about suddenly righting path that saw them overcoming scruples that allowed manufacturing practices less than their utter best to come into daily use. I do know who I'm directly talking to though, not an industry body. A peek through my posting history might reveal a few insightful comments on the 90C is fine sentiment that went across the grain. Entirely possible I intruded on the TIM discussions as well, with nods by site staff on that one.

Since I do agree the larger question was answered: Artificial firmware and physical construction limitations are heavily in place. My focus is shifting towards working towards accounting for all that I would like to accomplish in a new system build.

Much appreciate all the helpful thoughts expressed.

 

[SL2]

The thing is I like my PCs cool & quiet, I like having headroom to OC and tweak the hell out of my hardware, and saying "90°C is fine", just because you failed at making it better at the factory, makes me wanna puke.

At the factory? I'm pretty sure that's not what happened, see delidded temps. It's by design, but that doesn't anwer all the questions for me tho.

If AMD made the lid thicker just for the sake of heatsink compatibility and stay at the same height as AM4, why couldn't they just make the socket taller instead?

For a while I thought the extra thickness was to make room for 3D chips (and thus use thinner lids for those), but that was obviously wrong. The V-cache layer seems very thin.
There's still a chance that it was done for som other future stacked layout, but I don't really believe that anymore.

 

[GerKNG]

not in the foreseeable future.

 

[JustBenching]

Not true, since there are substantial temperature improvements between 13th and 14th gen, even with same power limits and boost clicks etc. So while some specs on paper are identical, there's hardware differences.


Yep. Whereas the "more efficient" AMD CPU I have idles at 30-40 W since IO die is always active. You can get it down to 25-30 W but that requires crippling performance with slow RAM, IF etc.

Backside power delivery along with gate all around ribbon transistors are both huge advances. Zen 5 is architecture improvement but no packaging changes, so idle issues will continue.

Zen 6 is supposed to introduce better packaging and maybe a new IO die which will hopefully bring Zen desktops into modern idling numbers with much better low load efficiency. Not everyone is running prime 95 or cinebench all day, in fact normal use, and even most gaming is much closer to idling than full load.

Yeah, major concern for amd is both the idle / light load power draw and the efficiency in heavy MT tasks. The lack of cores hits them pretty hard, to the point that a 3 gen old i7 is on par with amds latest and greatest 7700x / 7800x 3d.

 

[AusWolf]

Reality is in your own answer here. The Zen 2 3300X was just like a 4 core Zen 3. The main thing that change performance in both ADL->RPL and Zen 2- Zen 3 was cache changes. They are the same microarchitecture. Mucking with the cache doesn't change the pipelines or instruction width.

I've seen this marketing hype many times, where a 'new generation' CPU came out but was really same as the old one with more / faster cache and claimed to be a redesign. HP did the same thing with some of their PA-RISC chip releases, and I'd bet IBM did it with Power. Heck car makers do it all the time. New sheetmetal, same frame and powertrain.

Now I'm not saying it doesn't change performance - but it's not a new microarchitecture. Fact is, Zen 2 was starved by its cache, and Alder Lake had latency issues with its cache. These are all just evolutionary moves. for that matter - surprise - Zen 4 is as well.

But hey, don't believe me.

Believe Intel :

Intel Docs List Raptor Lake With Same Microarchitectures as Alder Lake

Details about Intel's next-generation cores revealed.

www.tomshardware.com

Oh, Raptor Lake is definitely the same stuff as Alder Lake, there's no doubt about that. Just like anything between Sandy Bridge and Kaby Lake was also the same. Even Coffee Lake was the same, to be fair, as it's just a couple of extra cores on top. But no, Zen 3 is not like that. Like I said, the cache hierarchy changed, not just the amount. Zen 4 shares way more similarity with Zen 3 than Zen 3 does with Zen 2.

 

[dgianstefani]

Oh, Raptor Lake is definitely the same stuff as Alder Lake, there's no doubt about that. Just like anything between Sandy Bridge and Kaby Lake was also the same. Even Coffee Lake was the same, to be fair, as it's just a couple of extra cores on top. But no, Zen 3 is not like that. Like I said, the cache hierarchy changed, not just the amount. Zen 4 shares way more similarity with Zen 3 than Zen 3 does with Zen 2.

It's literally not. The cache and memory subsystem was reworked in Raptor Lake, along with many other, more minor, hardware changes. Plus process improvements besides the architectural ones.

Sandy bridge to Kaby Lake was same? What are you smoking I want some please, about the only consistent spec is that they were quad cores...

 

[GhostRyder]

Personally I doubt its going to be game changing. Not saying its not going to be an improvement, just not one thats going to be as big a leap as people think/worth waiting for. Most of the time, if we play the waiting game we just will never buy because there is always something 'just around the corner'.

Now I could be wrong, we have seen huge changes before. But I think compared to current from both parties its not going to be revolutionary.

 

[AusWolf]

It's literally not. The cache and memory subsystem was reworked in Raptor Lake, along with many other, more minor, hardware changes. Plus process improvements besides the architectural ones.

Sandy bridge to Kaby Lake was same? What are you smoking I want some please, about the only consistent spec is that they were quad cores...

As an end user, I can't really see or feel any of those changes. Reading about them in a review is surely interesting, but when the CPU is already fired up in your system, who cares anymore?

At least the change in cache hierarchy between Zen 2 and 3 is something you can see or feel in certain applications, even if 95% of users don't care about that, either.

 

[Onasi]

Sandy bridge to Kaby Lake was same? What are you smoking I want some please, about the only consistent spec is that they were quad cores...

I mean, why stop there, let’s say that Intel hasn’t REALLY majorly changed the overall Core architecture since Nehalem. Those all were optimizations, you see.
Technically not even false, strictly speaking, I think internally at Intel everything since Nehalem IS one family. Just, uh, not in the way some people think.

 

[dgianstefani]

As an end user, I can't really see or feel any of those changes. Reading about them in a review is surely interesting, but when the CPU is already fired up in your system, who cares anymore?

At least the change in cache hierarchy between Zen 2 and 3 is something you can see or feel in certain applications, even if 95% of users don't care about that, either.

That's like saying a car made in 1970s is same as a car made in 2010, because they both have four wheels and a four cylinder engine, and when you drive it you feel similar g forces.

The change in cache and memory is something you can feel with ADL to RPL, it directly results in much better FPS, just one of many measurable things.

 

[RandallFlagg]

Oh, Raptor Lake is definitely the same stuff as Alder Lake, there's no doubt about that. Just like anything between Sandy Bridge and Kaby Lake was also the same. Even Coffee Lake was the same, to be fair, as it's just a couple of extra cores on top. But no, Zen 3 is not like that. Like I said, the cache hierarchy changed, not just the amount. Zen 4 shares way more similarity with Zen 3 than Zen 3 does with Zen 2.

But yes, Zen 3 is like that. Even clocked the same, Kaby Lake was significantly faster than Sandy Bridge.

That's +20% *at the same clock*. And yes, same microarchitecture.

Now, lets compare difference in Zen 3 vs Zen 2 on the same game.

Oh look, +6%. And oh my, that can entirely be attributed to - higher clock speeds!

All AMD did was muck with the cache on Zen 3.

 

[AusWolf]

That's like saying a car made in 1970s is same as a car made in 2010, because they both have four wheels and a four cylinder engine, and when you drive it you feel similar g forces.

No - it's like saying that your 1 year newer car has gone through some fundamental changes in its driving characteristics because the company revised the design of the wheel locking nut.

The change in cache and memory is something you can feel with ADL to RPL, it directly results in much better FPS, just one of many measurable things.

Isn't that because it has more cache and supports faster memory? I mean, sure it's an improvement, but I wouldn't put it on the architecture's tab.

But yes, Zen 3 is like that. Even clocked the same, Kaby Lake was significantly faster than Sandy Bridge.

I'm not saying that it wasn't. I'm just saying that the changes each architecture went through in that era were incremental, nothing fundamental. Surely, they added up over time.

Now, lets compare difference in Zen 3 vs Zen 2 on the same game.

Oh look, +6%. And oh my, that can entirely be attributed to - higher clock speeds!

All AMD did was muck with the cache on Zen 3.

Cherry picked example. I never said that the revised cache / CCD hierarchy improved performance in every single application.

 

[dgianstefani]

No - it's like saying that your 1 year newer car has gone through some fundamental changes in its driving characteristics because the company revised the design of the wheel locking nut.


Isn't that because it has more cache and supports faster memory? I mean, sure it's an improvement, but I wouldn't put it on the architecture's tab.


I'm not saying that it wasn't. I'm just saying that the changes each architecture went through in that era were incremental, nothing fundamental. Surely, they added up over time.


Cherry picked example. I never said that the revised cache / CCD hierarchy improved performance in every single application.

So is X3D not a "fundamental" change because it's just "incremental" adding more cache, and the architecture hasn't changed? You're muddying the waters here with "minor" "major" changes etc. The only thing that matters is end performance per clock. I don't care if they only changed a "wheel locking nut", if the car is 20% faster because of it, it's enough of an architectural change that you can't argue it's the "same".

 

[RandallFlagg]

<snip>

Cherry picked example. I never said that the revised cache / CCD hierarchy improved performance in every single application.

The inconsistency is saying, for AMD, a reworked cache is somehow a brand new microarchitecture. But for Intel, it's the same old same old.

Intel had essentially the same architecture from Haswell up to Comet Lake. Those are 14 stage pipeline chips with new instructions vs sandy bridge. They still had huge performance gains, due to everything from cache reworks to new process nodes and additional cores.

Alder Lake introduced 12 stage (vs 14 for Haswell->Comet Lake) fully unified pipeline with a 6-wide decode vs 4 from the previous gen. It's entirely new. Raptor Lake is the same. I don't know the specs but Meteor Lake appears the same, just new fab tech and IO changes.

And Arrow Lake will likely be the same as well, though performance is expected to be notably higher.

Lunar Lake is the next truly new Intel uArch, not just media hype like we normally get about 'all new' when it's really 'refreshed' (AMD and Intel both do this). This one will likely give 'next gen' levels of performance boost.

 

[Dragam1337]

Not within the next year or two anyways. Amd has better offerings no matter whether your goal is gaming, rendering or whatever. And this is coming from someone who had used intel since the athlon days in the early 2000s, prior to getting the 7800x3d.

 

[RandallFlagg]

Not within the next year or two anyways. Amd has better offerings no matter whether your goal is gaming, rendering or whatever. And this is coming from someone who had used intel since the athlon days in the early 2000s, prior to getting the 7800x3d.

Coming from someone who's first PC was a Corona luggable, and first DIY was a 20Mhz AMD 286 clone, you're wrong.

 

[Dragam1337]

Coming from someone who's first PC was a Corona luggable, and first DIY was a 20Mhz AMD 286 clone, you're wrong.

Aha - in which use case is intel the better choice ?

And that brag was honestly a bit pathetic, and shows you clearly didn't get the point of what i wrote.

 

[JustBenching]

Aha - in which use case is intel the better choice ?

All of them?

Usecase 1) Heavy MT workloads. In most price segments Intel offers vastly more MT performance. The difference is insane actually.

Use case 2) ST / latency sensitive tasks, again Intel is the obvious winner

Usecase 3) The average user that does light tasks and some gaming, although both companies offer very similar performance in games, Intel draws substantially lower power draws for the average user that's browsing, consuming youtube etc content, doing some light productivity (excel and the likes).

Usecase 4) Same as 1, heavy MT but efficiency is of utmost important, Intel is again leading by a huge margin on most price segments. Im currently using a 3 generations old Intel CPU and yet the I need at minimum a 7900x to beat my 3 gen old Intel in performance / watt. The 400$ 7800x 3d loses horribly there.

With that said the 7800x 3d is great if you are playing one of the few games that benefit tremendously from the vcache (MSFS for example) but that's a very niche usecase and it doesn't really justify the price tag of the CPU. The 7950x 3d is a GOAT of the cpu, or it would be if it hadn't have those scheduling issues (and the light load power draw which is nuts).

 

[Dragam1337]

All of them?

Usecase 1) Heavy MT workloads. In most price segments Intel offers vastly more MT performance. The difference is insane actually.

Use case 2) ST / latency sensitive tasks, again Intel is the obvious winner

Usecase 3) The average user that does light tasks and some gaming, although both companies offer very similar performance in games, Intel draws substantially lower power draws for the average user that's browsing, consuming youtube etc content, doing some light productivity (excel and the likes).

Usecase 4) Same as 1, heavy MT but efficiency is of utmost important, Intel is again leading by a huge margin on most price segments. Im currently using a 3 generations old Intel CPU and yet the I need at minimum a 7900x to beat my 3 gen old Intel in performance / watt. The 400$ 7800x 3d loses horribly there.

With that said the 7800x 3d is great if you are playing one of the few games that benefit tremendously from the vcache (MSFS for example) but that's a very niche usecase and it doesn't really justify the price tag of the CPU. The 7950x 3d is a GOAT of the cpu, or it would be if it hadn't have those scheduling issues (and the light load power draw which is nuts).

"One of the few games that benefit from vcache" - hard to take anything you say seriously, and honestly wouldn't have expected anything else from the nr1 intel fanboi on this site...

 

[JustBenching]

"One of the few games that benefit from vcache" - hard to take anything you say seriously, and honestly wouldn't have expected anything else from the nr1 intel fanboi on this site...

So you skipped 4 paragraphs and focused on that. Sure, a lot of games benefit from it, but not a lot of games show a big difference compared to an Intel CPU is my point.

An Intel fanboy with 3 amd laptops and an amd desktop is an oxymoron but sure, whatever floats your boat. Laptops and handhelds, amd is great, desktops - power hogs even for simple tasks.

Kinda new review

Spoiler

Test • AMD Ryzen 5 7500F

Après avoir testé des références peu onéreuses sur plateforme LGA 1700, au tour de l'AM5 et d'un Ryzen 5 au tarif très doux. Tout d'un grand sauf le prix ?

hardwareand.co

Every zen 4 part with an io die is drawing 40w more from the wall sitting there idly doing nothing. That's absurd, id feel like ruining the enviroment using CPUs with iodies.

 

[SL2]

That's +20% *at the same clock*. And yes, same microarchitecture.

That's a 26 % difference between 7700K and 9700K.

At the same time, the difference between same CPU's are only 16 % at TPU.

AT's C IV benchmarks clearly shows larger differences than TPU's.

I'm not saying 2000 through 7000 series are the same tho.

 

[Dragam1337]

So you skipped 4 paragraphs and focused on that. Sure, a lot of games benefit from it, but not a lot of games show a big difference compared to an Intel CPU is my point.

An Intel fanboy with 3 amd laptops and an amd desktop is an oxymoron but sure, whatever floats your boat. Laptops and handhelds, amd is great, desktops - power hogs even for simple tasks.

Kinda new review

View attachment 337907

Test • AMD Ryzen 5 7500F

Après avoir testé des références peu onéreuses sur plateforme LGA 1700, au tour de l'AM5 et d'un Ryzen 5 au tarif très doux. Tout d'un grand sauf le prix ?

hardwareand.co

Every zen 4 part with an io die is drawing 40w more from the wall sitting there idly doing nothing. That's absurd, id feel like ruining the enviroment using CPUs with iodies.

Because everything you said is right for the garbage can.

Benefits compared to intel ? Dafuq are you on about - most backwards logics ive ever heard. If we are going to talk about whether or not vcache benefits games, we are looking at how the equivelant amd cpu without vcache does compared to the one with vcache... aka 7700x vs 7800x3d. And it essentially always helps substantially. How the 7800x3d performs in a game compared to 14900k is entirely irrevelant in regards to whether or not vcache helps, which was your claim that it only does in a few games, and which is obviously pulled right out of your arse.

As for your "points" in your previous post - if MT workload is your purpose with a workstation, then you get a 7950x or threadripper, and beat anything intel has to offer. Simple as that.

And lol at that graph you pulled out of your *** - looked for a long while before you found that obscure french site, huh. Funny how non of the reputable tech sites, like the very one we are on, have anywhere near the same findings. Zen 4 uses about 25w at idle, and while that's more than intel, it's an absolutely meaningless difference - the difference in power cost would be a few dollars per year at most.

You're reaching so hard for someone who allegedly aint an intel fanboi (which you clearly are - you can deny it from here and till hell freezes over... every post you ever made screams it to the heavens).

 

[SL2]

@RandallFlagg Very nice cherry picking BTW, as if only one game would tell us everything.

Using the same two reviews as you did for Far Cry 5, it's a whole different story.

18% between 2600K and 7700K, and 15% between 2700X and 3800X.