Discussion Intel current and future Lakes & Rapids thread

[jpiniero]

I mean there's also a good what, 25% improvement in clock speeds going between the first generation of Skylake and Comet Lake.

Some of that is just headroom Intel left for overclocking with Skylake.

 

[Arkaign]

My prime excitement about this release is that it appears we'll actually have really solid budget and midrange CPU options, if everything clicks. Obviously that's not any kind of certainty in today's world.

The higher end models feel like fairly extreme overkill almost across the board now for the majority of buyers (typical office/business use and gaming). For example a 5600X is within a couple of percent of the 5950X even with an RTX 3090 in 1080(!!) Gaming. Exceptions made for people needing gobs of cores of course, rendering, encoding, scientific, yadda.

For someone just lucky enough to be able to grab a 6600XT or 3060, it's already a huge expense, and I doubt a 12900K over a 12600 will amount to any perceivable difference 🤔

I just hope it's good. Zen3 has been of course a fantastic piece of tech, but the $/perf for most has been kind of shot, which is a bit of a letdown for my pragmatic perspective after the just lights out value of Zen2. The 3600 non X (and even more so the super elusive 3100/3300) were flat out awesome value, something basically absent from Zen3. If AMD feels some pressure, imagine some sub $200 6C/12T action coming down the line, incredibly efficient, not tied to DDR5, and representing a go-to for people that don't have baller cash piles to burn.

 

[arandomguy]

Non K CPUs (such as the 12600) and non Z series boards aren't expected until sometime in Q1. I'm tentatively interested in the 12600 and lower 6 cores (depending on price, GPU features). One concern I'd have is that because it is a different die and package if they will aggressively cost optimization and do things such as not using solder.

 

[Mopetar]

What is a Zen3 tax?

Is there a governing body receiving these funds?

Or is this just loaded language?

I don't think you understand taxes and what they are used for.

X-tax is an expression to indicate an added cost for whatever X is that's essentially imposed by the company for what typically amounts to "because we can."

For example, if you buy a PC from most of the major manufacturers it includes a Windows license whether you want Windows or not. Microsoft gets paid even if you don't want to use their OS and install Linux. This has been referred to as the Microsoft Tax historically.

It's not really a tax in the sense of the word, but because no one really likes paying taxes the term is used for the negative connotation. Zen 3 tax refers to AMD's ~$50 price bump on Ryzen 5000 CPUs because they had Intel squarely beat and could charge $50 more since the competition couldn't match them in terms of top performance.

It's the same as when Intel was dominant and HEDT CPUs were considerably more expensive because AMD had nothing to compete with until Zen came out.

 

[dullard]

Non K CPUs (such as the 12600) and non Z series boards aren't expected until sometime in Q1. I'm tentatively interested in the 12600 and lower 6 cores (depending on price, GPU features). One concern I'd have is that because it is a different die and package if they will aggressively cost optimization and do things such as not using solder.

If the preliminary pricing is correct (one leak was $249 for the 12600 and $279 for the 12600K), then I can't see any reason to get the 12600 over the 12600K. Just $30 for 4 more cores when you need it and faster clock speeds all around just seems like too good of a deal. Alternatively, the 12400 isn't much slower than the 12600 for $46 less (again assuming the leaked prices are correct).

 

[HurleyBird]

Raptor Lake has small increases in IPC (I assume due to more L2 cache), but also it is rumored to have: 3.8% faster clocks, higher allowable PL2 power in performance mode, and faster memory support. While the IPC changes might not be much, the combination of all of the above might get close to 20% gains in the right scenarios.

I don't think there's any way that adds up to 20%.

The biggest improvement in RPL will be doubling the small core count.

 

[Schmide]

X-tax is an expression to indicate an added cost for whatever X is that's essentially imposed by the company for what typically amounts to "because we can."

For example, if you buy a PC from most of the major manufacturers it includes a Windows license whether you want Windows or not. Microsoft gets paid even if you don't want to use their OS and install Linux. This has been referred to as the Microsoft Tax historically.

It's not really a tax in the sense of the word, but because no one really likes paying taxes the term is used for the negative connotation. Zen 3 tax refers to AMD's ~$50 price bump on Ryzen 5000 CPUs because they had Intel squarely beat and could charge $50 more since the competition couldn't match them in terms of top performance.

It's the same as when Intel was dominant and HEDT CPUs were considerably more expensive because AMD had nothing to compete with until Zen came out.

I get the above especially with Microsoft, but if it stays in the company that's a really bad analogy.

I've heard the phrase equally bent into an "x86" tax.

Enforcement of patents is another misuse of tax, that while more fitting, is equally using loaded language.

I stand by my I don't think you understand taxes and what they are used for .

 

[Hulk]

I don't think there's any way that adds up to 20%.

The biggest improvement in RPL will be doubling the small core count.

I agree with this. Generally when moving from one memory architecture to another we see at best a 5% performance increase with bandwidth limited applications. And that was back when the move from one technology to the next were BiG, like the move to from SDRAM to DDR.

 

[Zucker2k]

I get the above especially with Microsoft, but if it stays in the company that's a really bad analogy.

I've heard the phrase equally bent into an "x86" tax.

Enforcement of patents is another misuse of tax, that while more fitting, is equally using loaded language.

I stand by my I don't think you understand taxes and what they are used for .

It is loaded language. This "tax" has never been used in the correct sense. It just means some type of added cost for no apparent reason, from the consumer's perspective.

 

[DrMrLordX]

It is amazing that Intel was able to pull this off.

Why?

 

[Hulk]

Why?

Sorry, meant to reply to the OP about how they pulled it off.

They haven't pulled it off yet. We need fully vetted reviews... performance results across a spectrum of applications, thermals/power, etc.. And oh yeah, the ability to actually go out and buy the parts

 

[DrMrLordX]

Sorry, meant to reply to the OP about how they pulled it off.

They haven't pulled it off yet. We need fully vetted reviews... performance results across a spectrum of applications, thermals/power, etc.. And oh yeah, the ability to actually go out and buy the parts

Pretty much. Anything else is just speculation at this point based on a paucity of data. It's tempting to bag on Golden Cove for apparently only having +19% improved performance vs Cypress Cove (Qualcomm claims to gain the same performance increase in just one generation, e.g. Snapdragon 888 -> Snapdragon 898), but Golden Cove may do better or worse than that in real-world testing.

 

[Ajay]

Higher PL2 doesn't affect ST performance, and neither does better memory support unless the testing is done with tuned memory, which isn't a sensible way of testing anyway.

Well, not for well respected outlets doing reviews. For enthusiasts, tuning memory has become important since CPU manufacturers have stopped leaving extra headroom on the table.

 

[Ajay]

My prime excitement about this release is that it appears we'll actually have really solid budget and midrange CPU options, if everything clicks. Obviously that's not any kind of certainty in today's world.

I suppose that will depend on Intel's rollout plans (based on capacity and yield trend). It might be a situation where higher end products are released first. But, eventually there should be a enough dice to go around.

 

[Ajay]

I have to admit, this is getting interesting enough that I wish I were upgrading next year

 

[epsilon84]

My prime excitement about this release is that it appears we'll actually have really solid budget and midrange CPU options, if everything clicks. Obviously that's not any kind of certainty in today's world.

The higher end models feel like fairly extreme overkill almost across the board now for the majority of buyers (typical office/business use and gaming). For example a 5600X is within a couple of percent of the 5950X even with an RTX 3090 in 1080(!!) Gaming. Exceptions made for people needing gobs of cores of course, rendering, encoding, scientific, yadda.

For someone just lucky enough to be able to grab a 6600XT or 3060, it's already a huge expense, and I doubt a 12900K over a 12600 will amount to any perceivable difference 🤔

I just hope it's good. Zen3 has been of course a fantastic piece of tech, but the $/perf for most has been kind of shot, which is a bit of a letdown for my pragmatic perspective after the just lights out value of Zen2. The 3600 non X (and even more so the super elusive 3100/3300) were flat out awesome value, something basically absent from Zen3. If AMD feels some pressure, imagine some sub $200 6C/12T action coming down the line, incredibly efficient, not tied to DDR5, and representing a go-to for people that don't have baller cash piles to burn.

I'm sure there will be *some* difference in gaming between the 12600K and 12900K, mainly due to clockspeed and cache size differences rather than the lower core count. Unlike AMD, Intel nerfs the cache size of lower end SKUs, it's basically artificial segmentation which I'm not a huge fan of, though I understand why it's done.

That being said, if the 12600K is within 10% of the 12900K in gaming, then that is 'good enough' for me, considering the price difference ($270 vs $600 I believe?)

Of course the added bonus is that AMD will probably be forced to lower prices to compete, as you mentioned.

If the 12600K outdoes the 5800X as rumours suggest, then that means the 5800X will be worth no more than $250 if AMD wants to maintain price / performance parity. A 5600X should then be below $200.

Of course this isn't taking into account DDR5 pricing and higher Z series mobo pricing from Intel, so if AMD plays their cards right they can still be the overall value champion even if the outright gaming performance crown changes hands.

 

[DrMrLordX]

Of course this isn't taking into account DDR5 pricing

Won't there be some Z690 boards that utilize DDR4? And if we're talking games, DDR4 may wind up being the performance champ for a little while anyway.

 

[coercitiv]

My prime excitement about this release is that it appears we'll actually have really solid budget and midrange CPU options, if everything clicks. Obviously that's not any kind of certainty in today's world.

The higher end models feel like fairly extreme overkill almost across the board now for the majority of buyers (typical office/business use and gaming). For example a 5600X is within a couple of percent of the 5950X even with an RTX 3090 in 1080(!!) Gaming. Exceptions made for people needing gobs of cores of course, rendering, encoding, scientific, yadda.

I'm sure there will be *some* difference in gaming between the 12600K and 12900K, mainly due to clockspeed and cache size differences rather than the lower core count. Unlike AMD, Intel nerfs the cache size of lower end SKUs, it's basically artificial segmentation which I'm not a huge fan of, though I understand why it's done.

That being said, if the 12600K is within 10% of the 12900K in gaming, then that is 'good enough' for me, considering the price difference ($270 vs $600 I believe?)

There will be a difference in gaming between every tier of ADL-S. As @epsilon84 already mentioned, Intel segments based on L3 cache as well, and this has a direct impact on gaming performance. In fact, unlike previous gens, even the 12600K will have more L3 than the rest of the i5 line. We've already seen just how important is L3 cache size in gaming.

Therefore, for some users a case can be made to spend more on the Intel CPU and get more cache, which will arguably extend the lifetime of their purchase by 1-2 years. This was certainly the best option during the Sandy Bridge -> early Skylake era, when buying i7 over i5 was arguably the better value over time choice, allowing you to skip 1 or even 2 generations. Nowadays though, the market is far more competitive and the value gamer is arguably better off purchasing from the $150-200 value zone and upgrading more often.

If Alder Lake delivers on the gaming front then their i5 lineup will be quite the wake-up call for some of the forumites who thought buying a cheap 10700K/10900K in the past 6 months was the best idea for "future-proof" gaming setups. With more cache and stronger P cores than the 10700K or even the 11900K, the 12600K should be able to deliver quite a valuable lesson here. It's going to be fun watching people who previously stated they wouldn't buy a 6-core for gaming in 2020-2021, some of them may awkwardly start recommending a 6-core in 2022.

 

[arandomguy]

If the preliminary pricing is correct (one leak was $249 for the 12600 and $279 for the 12600K), then I can't see any reason to get the 12600 over the 12600K. Just $30 for 4 more cores when you need it and faster clock speeds all around just seems like too good of a deal. Alternatively, the 12400 isn't much slower than the 12600 for $46 less (again assuming the leaked prices are correct).

It's so how things will compare against the 12500/12400. If we look at CML/RKL it was the xx400/500 SKUs that had better value (well at least for RKL before the supply issues for both).

The other factor is of course B series vs Z series boards. If you combine with the above, it's likely a 12400/500 + B series board combo would be $100+ if not around $150+ differential in terms of price against a 12600k + Z board.

Thirdly another issue this time around is the socket change and cooler computability. I'm not sure what the retail state of value coolers will be on the onset. The non K series will likely come with a heatsink (rumors pointing to possible improvements), this will at least be able stop gap until the value heatsinks saturate. Might not seem like much but could be another $20-$30 extra otherwise here as well, which adds up.

Then of course given the overall market and supply chain situation we might need to factor in what the actual street pricing and availability will be in reality at the onset (not just for CPUs but motherboards as well, 590 series boards at issues at the onset at launch) at least through the end of this year.

Of course the added bonus is that AMD will probably be forced to lower prices to compete, as you mentioned.

AMD might not need to react very fast simply because real world availability and pricing might be problematic for both CPUs and motherboards. Zen 3 CPUs and boards by comparison are ample now and can hold stock even on sale.

Won't there be some Z690 boards that utilize DDR4? And if we're talking games, DDR4 may wind up being the performance champ for a little while anyway.

Generally it seems the lower to mid range boards will be DDR4. The high end boards will be DDR5. Some segments that are both consumer/business targeted will split between the 2.

There will be a difference in gaming between every tier of ADL-S. As @epsilon84 already mentioned, Intel segments based on L3 cache as well, and this has a direct impact on gaming performance. In fact, unlike previous gens, even the 12600K will have more L3 than the rest of the i5 line. We've already seen just how important is L3 cache size in gaming.

If you look at the results for the rest of the games (5 out of 6) it doesn't paint that type of picture. The cache size differences for Intel CPUs might start to matter if you're targeting extremely high FPS competitive games but even 144hz (if not 240hz) non competitive game targets the difference gets close to negligible.

ADL's cache is also not exactly comparable to RKL and earlier that allows for a size to size comparison.

Personally at this point I'd still have some concerns with respect to thread management and gaming. Gaming are very multi-threaded these days, the problem is that loading is very unbalanced, very dynamic and very burst (therefore latency sensitive) in term of resource demands. It'll be interesting to see the frame pacing and consistency in possibly challenging scenarios.

 

[epsilon84]

If Alder Lake delivers on the gaming front then their i5 lineup will be quite the wake-up call for some of the forumites who thought buying a cheap 10700K/10900K in the past 6 months was the best idea for "future-proof" gaming setups. With more cache and stronger P cores than the 10700K or even the 11900K, the 12600K should be able to deliver quite a valuable lesson here. It's going to be fun watching people who previously stated they wouldn't buy a 6-core for gaming in 2020-2021, some of them may awkwardly start recommending a 6-core in 2022.

I can see your point, but it's not like ADL suddenly makes the 10th/11th gen CPUs obsolete for gaming. It will still run games well for many years to come, just not as well as ADL will, which is to be expected. Newer tech.. better performance. Not exactly a surprise there.

Also the 12600K isn't technically a 6 core CPU, it's a 10 core or 6+4... 10 core lite?

I'm curious how it will fare against the 10 core 10900K, I think the IPC gains in the P cores will be enough to offset the E cores lacking SMT. Should be a pretty close run thing in MT throughput, but the extra IPC should give the 12600K the edge in games.

To be honest I'm more impressed with how Intel managed to get Skylake level IPC on those tiny E cores, rather than the ~20% IPC gains on the P cores.

 

[coercitiv]

I can see your point, but it's not like ADL suddenly makes the 10th/11th gen CPUs obsolete for gaming.

Of course not, but the narrative in the last 6 months very core-count oriented. Core count was king, and this perception was only exacerbated by the lackluster RKL gaming performance. With an underwhelming 11900K and ADL-S being anounced as an 8+8 config, meaning potentially just 8 strong cores for gaming, some seriously underestimated the importance of core performance and cache size in favor of core-count (on an already gaming proven design which is Skylake). An adjustment is in order.

Also the 12600K isn't technically a 6 core CPU

Yes, but the rest of the i5 lineup will be 6+0. These will be the "cheap" gaming SKUs for 2022.

If you look at the results for the rest of the games (5 out of 6) it doesn't paint that type of picture. The cache size differences for Intel CPUs might start to matter if you're targeting extremely high FPS competitive games but even 144hz (if not 240hz) non competitive game targets the difference gets close to negligible.

What type of picture are you talking about? The benchmarks show, irrespective of how big or little the gains are (as some games are fully GPU bound even at 1080p), that the bulk of the gaming performance jump between 10600K and 10900K is due to cache size and not core count.

Whether one would be willing to pay for the performance delta between 10600K and 10900K is not my concern, and not a suggestion from me either. You can find even more testing about core scaling and cache size in the follow-up from Hardware Unboxed where they try to fit in 4c/8t scenario as well. (the i3 is clocked lower unfortunately)

 

[Asterox]

Hm, interesting but expected.

https://twitter.com/x/status/1441712675294101506

Intel 12th Gen Core AlderLake (i9-12900K) Review & Benchmarks – big/LITTLE Performance – Preview – SiSoftware

www.sisoftware.co.uk

 

[arandomguy]

What type of picture are you talking about? The benchmarks show, irrespective of how big or little the gains are (as some games are fully GPU bound even at 1080p), that the bulk of the gaming performance jump between 10600K and 10900K is due to cache size and not core count.

Your data from the video shows the following comparing 8 core vs 10 core and 20 mb vs 16mb L3 -

140 fps range - no cache difference, no core difference
130 fps range - no cache difference, <0.01% core difference
160 fps range - <0.05% cache difference, no core difference
290 fps range - <0.05% cache difference, 2.5% core difference
160 fps range - 2.5% cache difference, 2.5% core difference
500 fps range (esports title, this is the one you time stamped) - 5.5% cache difference, 3% core difference

Cache and memory sub systems can become a significant factor if you are trying to drive extremely high fps (essentially only a factor competitive esports titles), in which every little bit starts mattering. You've driven the frame times so low at that point that any tiny bit of lowered latency in the entire chain starts to factor in.

CML (or RKL, or etc.) have memory subsystems that can achieve low enough latency with at least respectable memory latency that is already enough for even high refresh rate gaming. Whether or not ADL's is the same or not however is another matter.

I suppose you might counter and point out the 6 core results have slightly greater differences. However here I'd wonder about the methodology and whether or not you can draw the conclusions as simply attributable to cache. Why? Because the lower core count configurations of the higher cache CPUs are also dropping quite a bit once shifting down to 6 cores. This to me suggests something else is in play here.

 

[Joe NYC]

Hm, interesting but expected.

I don't know if any of this is expected by anyone. Alder Lake launch seems like a complete cluster F.

Microsoft is making wholesale changes to Windows 11, seemingly with no forethought, just weeks prior to general release. Software that is supposed to be for 100s of millions of users is hijacked to satisfy timing an an oddball CPU that has zero users.

It seems that Microsoft is rushing Windows 11 as if this was a vaccine to fight pandemic, as opposed to deliberate planning, gathering feedback, refining etc... Alder Lake is not Covid, lives are not at risk, Microsoft does not have to cut corners to make an arbitrary deadline.

Sandra seems to be "customizing" its testing to new hardware, as opposed to testing how new hardware runs existing software, how adaptable the new hardware is to all kinds of software users will throw at it...

 

[Ajay]

There will be a difference in gaming between every tier of ADL-S. As @epsilon84 already mentioned, Intel segments based on L3 cache as well, and this has a direct impact on gaming performance. In fact, unlike previous gens, even the 12600K will have more L3 than the rest of the i5 line. We've already seen just how important is L3 cache size in gaming.

Therefore, for some users a case can be made to spend more on the Intel CPU and get more cache, which will arguably extend the lifetime of their purchase by 1-2 years. This was certainly the best option during the Sandy Bridge -> early Skylake era, when buying i7 over i5 was arguably the better value over time choice, allowing you to skip 1 or even 2 generations. Nowadays though, the market is far more competitive and the value gamer is arguably better off purchasing from the $150-200 value zone and upgrading more often.

If Alder Lake delivers on the gaming front then their i5 lineup will be quite the wake-up call for some of the forumites who thought buying a cheap 10700K/10900K in the past 6 months was the best idea for "future-proof" gaming setups. With more cache and stronger P cores than the 10700K or even the 11900K, the 12600K should be able to deliver quite a valuable lesson here. It's going to be fun watching people who previously stated they wouldn't buy a 6-core for gaming in 2020-2021, some of them may awkwardly start recommending a 6-core in 2022.

Sadly, 'future proofing' is a somewhat pointless atm, wrt Gaming. Those who are not ATF ballers, need to sell kidney to buy a decent GFX card. Apparently, this problem will persist throughout this year and next. Who knows what will be affordable in 2023, supply is expected to increase due to increased CAPEX spending by the top Fabs. If one has a 5-6 year timeline between completely new builds (as I do on average), then buying up a level or two up on a CPU makes more sense. Those who are on a two year plan need not worry so much.