Intel Core i9-12900K, i7-12700K, i5-12600K Specs, Pricing & New TDP!

Alder Lake looks' pretty good. I like the i5 and i7 versions best. I'm not tempted at the moment. I'll wait and see what AMD's "3d cache" does since I've a x570 waiting in the wings if need be. In the meantime I'll keep rocking my 3600. :)
Winter is coming. If you need home heating, Alderlake will do :) Honestly, since you're on a 7nm chip already, HODL!!!!!💎 🚀 🌙
 
Has someone that did not look at intel in a very long time, is that still true with the different motherboards in mind ? I have the impression that you are right.

That seem true a 5800x + a good B550 is around 525-530, a deluxe x570 will be around $580

The cheapest LGA 1700 + 12600k seem to be around $520


Same, would it have released at the same time has the latest Ryzen it would have been really good and impressive, but with the amount of power needed, that make it look that AMD will retake the lead next release or right now with a price drop, but maybe there is performance on the driver/windows/game side left on the floor at the moment. It is an impressive Intel jump, but specially with the power in mind, not that impressive jump for the x86 Pc platform has a whole, until PCI xpress 5 has some general people purpose and DDR 5 significant at least.
eh? I think the cheapest "good" B550 is the Asus TUF MATX version, which is about $150. So that would put the total with a 5800x around $550.

Alder lake board prices are inflated right now because you can only by Z690. But the B660 boards should be around end of year or early next year. That said....Gigabyte has some solid boards for around $230. So, the current end price is very similar between 5800x and 12600k.
 
HUB has a decent first look at the 12900k up now. I'll be holding out for the last iteration for sckt AM4 at this point.
 
Everyone is discussing how competitively priced this Intel offering is. I just don't see it when you add the cost of the; MB, custom cooling, and DDR5 in the mix.

It's like saying my pizza is a good price at $15, but then add $25 for meat and another $20 if you want cheese.
 
Everyone is discussing how competitively priced this Intel offering is. I just don't see it when you add the cost of the; MB, custom cooling, and DDR5 in the mix.

It's like saying my pizza is a good price at $15, but then add $25 for meat and another $20 if you want cheese.
If you go with DDR4 over DDR5, its not too bad of a difference. Right now DDR5 is expensive and the performance isn't all that much better then DDR4.
 
Tech Notice reports that an Asus bios update shaved about 25 watts off the power usage in Cinibench R23. He does a lot of video productivity tests. Should have a video out later today, which shows video timeline editing performance for Alder Lake.
 
Well, every Intel version wasn’t entirely compatible (if at all) with the prior generation hardware. When you HAVE to make a new motherboard for a chipset, you do. When you don’t…. Why bother? Would you swap one perfectly fine x570 board for another? I don’t know many people that would- no way I’m buying a dark hero when I’ve got another x570 box sitting here already. My Altus Elite isn’t the nicest board (I spent on Threadripper instead), but there’s really no justification for swapping it with another unless it dies.

Conversely, I didn’t really need features beyond what my Z170 had for my consumer Intel box, but I HAD to upgrade it to get a 10700- there wasn’t an option. Say what you will about cpus within platform generations (I don’t do a ton of upgrading there) - but with two truly high end processors on the same platform, why upgrade the motherboard? It’s still x570.

If I needed thunderbolt or USB4 on a system that didn’t have it, I’d buy an add-in card. Not a new motherboard.

Right, but the choice to make ultra expensive high end variants for every little incremental change of Intel CPU/chipset doesnt' seem less justified than for releasing new/updated boards for very popular X570 chipsets. This was nothing new back in the old Intel HEDT days, for instance X99 spanned both the Haswell-E ( Rampage V series up to Rampage V Extreme) and when Broadwell-E releases arrived the revamped boards came as well including the Rampage V Edition 10 model that I have. Going all the way back to the X58 days, there was a Rampage II Extreme, but when the hex-cores came out with the 980X they launched the Rampage III Extreme which was one of the best of X58 and an early PCI-E 3.0 slot supporter ; this sort of thing continued throughout the years where even if it wasn't a whole "numbered" step, it was a something like a suffix - I can remember certain ROG etc.. refreshes were called "Alpha", like the Threadripper "Zenith II Extreme Alpha". It isn't necessarily that people with highest end boards from that same chipset the previous year will buy a new refresh (a minority will who want the newest features), but especially given the long lasting (and due to shortage hard to acquire after a point) elements of Zen 3 , lots of people were building Ryzen 5000 series as "THE" gaming and all around CPU to use. Thus, were it available they'd likely buy the refreshed X570 Zen3 boards over the previous years' editions, so it seemed strange not to make high end releases for those buying to build a new Zen3 powered system. On such a desirable CPU and chipset, this has to be a larger niche than those who will buy highest-end mobo for a minor incremental upgrade necessitating a whole new CPU and Mobo from Intel over the past 3-4 generations, so it feels strange NOT to do it for AMD as well.

Its not so much about buying a new mobo just to get these things as it is that given the prices of existing mobos and advancing technology, it seems strange that they weren't included already. Regarding TB4 / USB4.x it would be fantastic if you could simply buy a universally compatible expansion card, but such things dont' seem to exist. I see that most of the Thunderbolt 4 cards are specific to certain mobo manufacturers like Asus or MSI.. The Asus card seems reasonable, but even it claims to only support certain Intel chipsets (and in theory is supposed to plug into a header on the mobo itself). Now, I see that some users of certain Asus X570/B550 chipsets apparently can get it to work, but lots won't buy something that claims it doesn't support your chipset. Putting TB4 aside, despite the ratification of the USB4 standard quite awhile ago, there isn't a selection of 40gbps compatible, all features/display modes enabled, USB4 cards being sold out there at all the way one can find USB 3.0 and later cards of varying quality all over the place.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
There is something to be said for platform longevity. AMD supports their sockets long-term, while Intel traditionally does not.

If you buy the top end Intel CPU, that is basically it. If you get a mid-range, then maybe you can buy a used high-end CPU (from the same line) down the road, but it may not be a huge upgrade.

With AMD, if you had got Zen 2 (or even some better Zen 1 boards) you might have been able to do 2 significant CPU upgrades over several years without building a whole new system. So the cost savings over time add up.

Also, even with AM4 at the end of the life, AMD is still doing one last refresh with the 3D cache, they didn't have to do that, but they are supporting their customers and not exploiting them.
 
There is something to be said for platform longevity. AMD supports their sockets long-term, while Intel traditionally does not.

If you buy the top end Intel CPU, that is basically it. If you get a mid-range, then maybe you can buy a used high-end CPU (from the same line) down the road, but it may not be a huge upgrade.

With AMD, if you had got Zen 2 (or even some better Zen 1 boards) you might have been able to do 2 significant CPU upgrades over several years without building a whole new system. So the cost savings over time add up.

Also, even with AM4 at the end of the life, AMD is still doing one last refresh with the 3D cache, they didn't have to do that, but they are supporting their customers and not exploiting them.
Yeah I think we can put our money down that Intel is still going to do their tick-tock cycle and raptor lake will be it before we go to LGA 1750 or whatever it will be.
 
I don't see the Amazingness. Yes it's faster but it's not leaps and bounds. Basically trades blows with 5000 series in gaming. I think further improvements to Windows 11 and the 3d vcache will bring AMD right back in line.
 
Everyone is discussing how competitively priced this Intel offering is. I just don't see it when you add the cost of the; MB, custom cooling, and DDR5 in the mix.

It's like saying my pizza is a good price at $15, but then add $25 for meat and another $20 if you want cheese.
If the 12900K or 12700K was $150 more and the Z690 boards $150 less, I reckon the goalposts in this argument would shift to "Nobody is talking about the price of the CPU?!"

I'll take the cheaper CPU as the better option, thanks. Because the Z690 board prices will settle down. It's launch day, the board partners are a little full of themselves.
 
Last edited:
I am not sold on the hybrid design on the desktop.

Experiments shall occur if I can get one.
I am not either. People are falling for the BS marketing. The ecores are more powerful then I was expecting but this exist cause they can't get the power under control. Imagine how good this would be if it was actually 16 pcores and 32 threads.
 
I am not either. People are falling for the BS marketing. The ecores are more powerful then I was expecting but this exist cause they can't get the power under control. Imagine how good this would be if it was actually 16 pcores and 32 threads.
I think it could make a lot of sense
1) There is chips size and cost here and not just power I think, this would be a giant chips with 16 pcore of those size no ?
2) It seem the the performance of a core does not scale well at all with is size if the drawing that show 4 ecore taking the size on a single P-core are true, despite having just half the performance in many thing, you have twice the performance by size taken on the chips and much less of half the power.

In a world where multithread performance matter a lot but that for many application the performance on the first 1 to 8 thread matter much more than the next 8 thread, it seem logical to have that strategy.

From my understanding if you keep the same price and single thread performance you cannot go to 16pcore/32t, you go to 10c/20T, if the statement imagine if they had much better p-core able to be has good but almost twice has small, sure but that feel almost tautologic.
 
If you go with DDR4 over DDR5, its not too bad of a difference. Right now DDR5 is expensive and the performance isn't all that much better then DDR4.
You can do that sure... however DDR5 depending on the application/game being tested is accounting for a pretty big uplift. I don't know considering the price availability and junkiness of the early DDR5 kits. I'm not sure its even fair to say Intel wins some of these benches in todays reviews... the few sites that have tested both 4 and 5 don't really paint a clear picture imo. Some benches DDR5 accounts for almost the entire uplift on this gen... and in others their is zero difference. Its a first gen product I guess.

I think its a hard sell though to say to someone coming from an older DDR3 machine... buy DDR4 now. Yes you can buy DDR5 later and try and sell your 4 sticks I guess. I doubt anyone really wants to buy the latest greatest gen buy a high end MOBO... and slot the older gen ram in it.
 
I think it could make a lot of sense
1) There is chips size and cost here and not just power I think, this would be a giant chips with 16 pcore of those size no ?
2) It seem the the performance of a core does not scale well at all with is size if the drawing that show 4 ecore taking the size on a single P-core are true, despite having just half the performance in many thing, you have twice the performance by size taken on the chips and much less of half the power.

In a world where multithread performance matter a lot but that for many application the performance on the first 1 to 8 thread matter much more than the next 8 thread, it seem logical to have that strategy.
This why I am interested more in what the rumored x699 HEDT will bring.
 
You can do that sure... however DDR5 depending on the application/game being tested is accounting for a pretty big uplift. I don't know considering the price availability and junkiness of the early DDR5 kits. I'm not sure its even fair to say Intel wins some of these benches in todays reviews... the few sites that have tested both 4 and 5 don't really paint a clear picture imo. Some benches DDR5 accounts for almost the entire uplift on this gen... and in others their is zero difference. Its a first gen product I guess.

I think its a hard sell though to say to someone coming from an older DDR3 machine... buy DDR4 now. Yes you can buy DDR5 later and try and sell your 4 sticks I guess. I doubt anyone really wants to buy the latest greatest gen buy a high end MOBO... and slot the older gen ram in it.
DDR4 and 5 are not interchangeable. MB support 1 or the other. So if you go DDR 4 now you are looking for a new MB if you want to use DDR5.
 
I am not either. People are falling for the BS marketing. The ecores are more powerful then I was expecting but this exist cause they can't get the power under control. Imagine how good this would be if it was actually 16 pcores and 32 threads.
Not exactly. The core scaling will get crazy, quickly, with each iteration. Rumors suggest AMD has plans which could see a 16 E cores chipset slotting in with some of their big cores. And I’m sure Intel has similar plans. Technically, they could do a 6 P core with 16 E cores, now. But their caching scheme and the still monolithic design may not compliment that many cores.
 
I am not either. People are falling for the BS marketing. The ecores are more powerful then I was expecting but this exist cause they can't get the power under control. Imagine how good this would be if it was actually 16 pcores and 32 threads.
Intel is at a FAB loggerhead.

If they made that part it would need to be the size of a thread ripper... it would gobble 300 watts, and need to be sold with Intels Peltier cooling kit. lol

The advantage of the e cores.... is they in fact perform around 90% as well as the P cores, but can be clocked much lower ratio wise vs the P cores. They are also MUCH smaller die wise. They have sandwiched 4 e cores into the silicon space of ONE P core. For background ground and low priority threads it is much better performance wise to have 4 slightly slower cores then 1 race car pushing 2x the heat.

e Cores are perfectly logical. AMD will also be going this way at some point... AMD has patents for Zen 5 that confirm this. https://www.freepatentsonline.com/y2021/0173715.html

Anyway Alder Lake is interesting tech... imo Intel pulled it forward knowing their fabs are Fd up for awhile yet. The only way they could make cores like the P cores... would be to strip the number of them on the CPU. Which would be a marketing night mare. Hell its what they did the last gen when i9s actually LOST core count. To keep pushing IPC to keep up with AMD they where going to have to strip more cores or make something more like a thread ripper to compete with AMDs standard chips. Instead they again IMO pulled forward an idea probably intended for the mid 2020s when AMD will also be going there with ZEN5. By turning one of those P cores into 4 smaller ones... they can keep the marketing machine going with 16 core/24 thread talk... while only really paying in terms of heat and power for a smaller number of overly hot power sucking cores.
 
The advantage of the e cores.... is they in fact perform around 90% as well as the P cores, but can be clocked much lower ratio wise vs the P cores. They are also MUCH smaller die wise. They have sandwiched 4 e cores into the silicon space of ONE P core. For background ground and low priority threads it is much better performance wise to have 4 slightly slower cores then 1 race car pushing 2x the heat.

I dont disagree with you that e cores may still make sense (especially if they only take 1/4 of die space, which Im still not sure about), but I think the performance is less than 90% of P cores.

https://www.anandtech.com/show/1704...hybrid-performance-brings-hybrid-complexity/7
https://www.anandtech.com/show/1704...hybrid-performance-brings-hybrid-complexity/9

Seems like 50% of the performance?

Power seems quite a bit a bit better on Ecores. 8 ecores use like 50 watt full load? 8 pcores use like 250 watt?
https://www.anandtech.com/show/1704...hybrid-performance-brings-hybrid-complexity/4

If you look at purely multithread independant workloads, and compare 4 ecores vs 1 pcore (again assuming same die space), then about 25watt full load for 4 ecores vs 31 watt full load for 1 pcore? 4 ecores will offer about twice the performance of 1 pcore.
I think many server solutions might prefer all ecores and no pcores.
Please correct me if i'm misinterpreting anything, and I do realize this is all very early data and only from anandtech.
 
You can't just compare P/E cores alone, the point is that they work together (giving the P cores the heavy tasks and the E cores lower priority things),

For example, if you look at the overall results, especially in non-game workloads, you see that the Intel chip is about matched or better than the 5950X, which you wouldn't assume just looking at core count and clock speed.

https://www.techspot.com/review/2351-intel-core-i9-12900k/
 
I am not sold on the hybrid design on the desktop.
I am not either. People are falling for the BS marketing.

One thing to bear in mind is that AMD has the same thing in the works. It seems like AMD, Intel, and Microsoft are putting serious time and money into the concept.

But I'm no beta tester. In my 20s, sure, but now I think I'm inclined to treat my home equipment like I would a server environment; better to get an older, proven system than the new hotness.

Steam Deck notwithstanding. Gimme! Now!
 
Good showing by Intel, although I am blown away by how much more power their 10nm is sucking compared to AMD's built on 7nm. I wonder how much of that is architecture vs process, because it's not even close.
 
Good showing by Intel, although I am blown away by how much more power their 10nm is sucking compared to AMD's built on 7nm. I wonder how much of that is architecture vs process, because it's not even close.
While in some scenario it is vastly more (blender, stress test) In some other like gaming, they do not seem to be much different:

Alder-Lake-LTT-20.jpg


Or is it some strangeness of some test setup :
9974_47_intel-core-i9-12900k-alder-lake-cpu-review.png
 
What is recommended air flow cooler and PSU for 12900?
psu depends on gpu. air cooler dont seem to be up to snuff except for the d15 which gets overwhelmed during sustained full loads. even 360 aios are being worked hard.
 
Right, but the choice to make ultra expensive high end variants for every little incremental change of Intel CPU/chipset doesnt' seem less justified than for releasing new/updated boards for very popular X570 chipsets.
While I see your point, I still disagree. You MUST do this for Intel - Z390 worked for 9000 series and some (ish) 8 series. Z490 worked for 10/11(ish). Z590 worked for 11. Z690 works for 12 (and maybe 13). Intel effectively requires you to sell a new motherboard with a new chipset for each generation of processor; you don't see many people with Z490 boards running 11th Gen (aside from the fact that there wasn't really a point) - they bought Z590 boards if they needed that. Your market is literally anyone buying a new CPU - for almost every CPU sold, you sell a motherboard.

On X570, you could buy a brand new X570 Meg Creation on day 1, and use it with a 2700X, 3950X, 5950X, and (supposedly) the 6950X. All on the same board. There's no requirement for upgrades there - so releasing a X570 Creation 2 has a VERY small market of people that want x570, don't own it already, and are buying one of those CPUs. Your market in X570 land is folks that need a CPU and do NOT already have x570 OR want a new x570 board - a much smaller market, and you're also competing with x370 boards, x470 boards, B550, etc that are already in the field.

Intel is almost a 1:1 relationship of CPU:Motherboard sales. AMD is a 1:1 for the first release of the platform, and then VERY much Many:Few because of the longevity of the platform afterwards. We should take a poll here actually, I'm curious what the ratio would be.

On top of that, folks will buy the boards that are available - regardless of how new it is. Doesn't matter if it was one of the first X570 boards out there - a sale is a sale, and the platform doesn't let you make many changes, so there's not much there to tweak. A 2 year old design will still sell today - and all your soft costs are paid for already. Take the larger margin! More on this later.

This was nothing new back in the old Intel HEDT days, for instance X99 spanned both the Haswell-E ( Rampage V series up to Rampage V Extreme) and when Broadwell-E releases arrived the revamped boards came as well including the Rampage V Edition 10 model that I have. Going all the way back to the X58 days, there was a Rampage II Extreme, but when the hex-cores came out with the 980X they launched the Rampage III Extreme which was one of the best of X58 and an early PCI-E 3.0 slot supporter ; this sort of thing continued throughout the years where even if it wasn't a whole "numbered" step, it was a something like a suffix - I can remember certain ROG etc.. refreshes were called "Alpha", like the Threadripper "Zenith II Extreme Alpha".
HEDT is a different market, x399 failure aside. 1, it's prosumer - sales is much more steady state for the lifespan of the platform than for others (both longevity wise and because businesses buy them rather than individuals). 2. Because it's steady state for the life of the platform, you can still compete for a (steadily shrinking mind you) market of folks buying the platform at different stages, while consumer tends to be very front-loaded in comparison. 3. The Motherboard is a much smaller portion of the overall system cost - while upgrades are rare, spending more for a more "featured" version is something folks do as they buy them over time (see the LTT rendering farms for instance, or their editing systems - as they add people, they're steadily building boxes for it, and steadily buy the newer versions of the platform), and thus continuing to compete for that market may (or may not) make sense at times.

The additional layer here is technical - the HEDT chipsets and platforms have a lot more flexibility in them. Taking away USB controllers to allocate PCIE lanes to a thunderbolt controller is allowed (or even built in). X570 does not allow for that - 4 lanes for NVMe, 16 lanes for PCIE, and 4 lanes for chipset. Locked. The rest of the onboard accessories are pretty limited in how you can carve them up - adding in a thunderbolt controller or other things would generally not work given how the chipset is designed. Consumer vs prosumer - there's a lot you can do on TRX40 and X299 you can't on the consumer parts. There's also the age of the platform - you can make minor changes on HEDT, and since the contribution R&D/etc is all paid back, eek out larger margins AND justify the new expenses. My x299 Designare, for instance, is the new 10G model - adding 2 10G ports is not nearly as expensive as the markup they added to the board compared to the prior generation, but folks buying them don't even blink - corporates don't care, creators don't care, and us enthusiasts just shrug if we need the feature (and if not, the old version is still around). That's not as true on consumer kit that is much more price competitive (hence why so many of those high-end boards were discontinued - no one buying them, no reason to keep it up), and without the easy platform flexibility, making it work ~may not~ be possible.

Heck, I ~think~ that some of hte boards with built-in 10G ports turned off a PCIE slot or NVMe slot if you enabled that port. I'll double check that - but I did read it somewhere (they were pulling the lanes from the chipset and not the CPU). It's effectively an add-in card, just built into the board. Many of them included those things AS add-in cards!

It isn't necessarily that people with highest end boards from that same chipset the previous year will buy a new refresh (a minority will who want the newest features), but especially given the long lasting (and due to shortage hard to acquire after a point) elements of Zen 3 , lots of people were building Ryzen 5000 series as "THE" gaming and all around CPU to use. Thus, were it available they'd likely buy the refreshed X570 Zen3 boards over the previous years' editions, so it seemed strange not to make high end releases for those buying to build a new Zen3 powered system. On such a desirable CPU and chipset, this has to be a larger niche than those who will buy highest-end mobo for a minor incremental upgrade necessitating a whole new CPU and Mobo from Intel over the past 3-4 generations, so it feels strange NOT to do it for AMD as well.
But they'll also buy the previous generation X570 board - why make a new one? Sale is a sale is a sale. Tooling, support, etc - contribution margin on year-old platforms is MUCH higher than on a brand new one - you've already amortized out all the R&D costs, so your profits are higher. Dark hero had a pile of BIOS updates and the like to fix bugs that came up - just like any new board - that all cost R&D time and support time, which is all expenses and reduces margin. I suspect they believed that the market of folks who'd pay for a passively-cooled x570 board was large enough to justify the soft costs of building it - only ASUS knows if that was true or not.

If you don't need to release something, don't bother. HEDT again, I suspect, is different because that's a more steady state business that is more flexible on costs and rising prices.

Intel outsells AMD still, and every CPU is a motherboard. AMD doesn't sell a board with every CPU - folks do in-place upgrades.

Its not so much about buying a new mobo just to get these things as it is that given the prices of existing mobos and advancing technology, it seems strange that they weren't included already. Regarding TB4 / USB4.x it would be fantastic if you could simply buy a universally compatible expansion card, but such things dont' seem to exist. I see that most of the Thunderbolt 4 cards are specific to certain mobo manufacturers like Asus or MSI.. The Asus card seems reasonable, but even it claims to only support certain Intel chipsets (and in theory is supposed to plug into a header on the mobo itself). Now, I see that some users of certain Asus X570/B550 chipsets apparently can get it to work, but lots won't buy something that claims it doesn't support your chipset. Putting TB4 aside, despite the ratification of the USB4 standard quite awhile ago, there isn't a selection of 40gbps compatible, all features/display modes enabled, USB4 cards being sold out there at all the way one can find USB 3.0 and later cards of varying quality all over the place.
You're right - I forgot but TB is tied to not only the license, but to the board supporting it (regardless of it being present or not). Takes a header on the board AND the PCIE slots. There are stories of folks getting them to work even still (https://www.reddit.com/r/eGPU/comments/lw0pn1/thunderbolt_4usb_4_pcie_card/), but YMMV. Blame intel? Not sure who's at fault there. Asus and Gigabyte both make boards for them - generally prime/designaire/etc (I've got one on my X299 board; I don't use it currently).

You're right that Zen3 bled the lines between prosumer and consumer - but it's still a consumer platform, even if the CPU is capable of more. Consumer platforms are, sadly, limited. If you want more than that - HEDT is your home :(

Standards take time to work their way through. Support for some things is required in CPU too (not everything supports PCIE flash, even if it has a PCIE slot, and Optane is very limited in what it works on).
 
I dont disagree with you that e cores may still make sense (especially if they only take 1/4 of die space, which Im still not sure about), but I think the performance is less than 90% of P cores.

https://www.anandtech.com/show/1704...hybrid-performance-brings-hybrid-complexity/7
https://www.anandtech.com/show/1704...hybrid-performance-brings-hybrid-complexity/9

Seems like 50% of the performance?

Power seems quite a bit a bit better on Ecores. 8 ecores use like 50 watt full load? 8 pcores use like 250 watt?
https://www.anandtech.com/show/1704...hybrid-performance-brings-hybrid-complexity/4

If you look at purely multithread independant workloads, and compare 4 ecores vs 1 pcore (again assuming same die space), then about 25watt full load for 4 ecores vs 31 watt full load for 1 pcore? 4 ecores will offer about twice the performance of 1 pcore.
I think many server solutions might prefer all ecores and no pcores.
Please correct me if i'm misinterpreting anything, and I do realize this is all very early data and only from anandtech.
I agree... and your right we don't know exact die savings on the e core complexes.

I agree on your server thoughts as well. I have read some people suggesting next gen version of x86 big little may have more P cores... I believe its more likely they will actually have more e cores. I could see Xeon alder (or next gen alder) Intel x86 being more 8 / 16 P/e and 12 / 24... and Intel advertising them as 24/36 core 32/48 thread chips. Especially with all the traction ARM server solutions have been getting. If Intel can show Xeons with somewhat close power usage... with more burst performance with a handful of P cores on tap, I think that might be a very reasonable solution for those markets.
 
If the 12900K or 12700K was $150 more and the Z690 boards $150 less, I reckon the goalposts in this argument would shift to "Nobody is talking about the price of the CPU?!"

I'll take the cheaper CPU as the better option, thanks. Because the Z690 board prices will settle down. It's launch day, the board partners are a little full of themselves.
I am sure people said that about video card prices as well... Board partner prices are not going to settle down on motherboards, they will continue to rise as well.
 
I am sure people said that about video card prices as well... Board partner prices are not going to settle down on motherboards, they will continue to rise as well.

Nah, it's been the opposite all year with video cards. Buy FE or from amd.com if you want to try your luck on MSRP. Partner pricing's gonna be through the roof.

Motherboards, though, aren't video cards, and while they are hot commodities among builders, you gotta factor in all the motherboards people are buying as complete PCs just for the video card. Those mobos are getting flipped on the aftermarket, usually at a loss. Premium motherboard makers know what they got, and part of what they got is enormous amounts of grey market competition.
 
DDR4 and 5 are not interchangeable. MB support 1 or the other. So if you go DDR 4 now you are looking for a new MB if you want to use DDR5.

Just like with pretty much every memory standards change, there will be budget boards that have slots for both standards, you just can't use them at the same time. I remember seeing one such board with both DDR4 and DDR5 slots when looking around at LGA1700 boards, but didn't take note of who made it. Might have been ASRock? Biostar, ECS and PCChips used to make such boards all the time that supported dual memory standards.

I remember having boards with 30 and 72 pin SIMMs, 72 pin SIMMs and 168 pin DIMMs, 168 pin DIMMs and 184 pin DDR. DDR and DDR2 was rare, but it existed. DDR2 and DDR3 was also rare, but existed in lower budget boards. DDR3 and DDR4 wasn't really a thing, but there were weird low end boards that had DDR4 CPUs using DDR3L memory because the IMC supported it. If it's possible, someone will do it and make a product.

What is recommended air flow cooler and PSU for 12900?

Liquid cooling, at minimum a 360mm radiator if you plan to run the CPU hard. You may get away with a 240/280mm CLC for more moderate loads.
 
I am sure people said that about video card prices as well... Board partner prices are not going to settle down on motherboards, they will continue to rise as well.
Yes the supply/demand curve is the one true god, however GPU's and MB's aren't really comparable with cryptomining having skewed things, and a Z690 won't mine any better than a $50 motherboard. The MB's will definitely settle down at least 20% within a month or two, or your money back.
 
Last edited:
Yes the supply/demand curve is the one true god, however GPU's and MB's aren't really comparable with cryptomining having skewed things, and a Z690 won't mine any better than a $50 motherboard. The MB's will definitely settle down at least 20% within a month or two, or your money back.

People keep saying that, but you still can't buy PS5s and XBXs and they don't cryptomine at all. There's more going on here than just cryptomining. I don't know if I'd bet on a 20% lower MB price anytime soon.
 
TLDR:
DDR5 is useless
Gaming the CPUs are close enough that it doesn't matter
Productivity - Intel 20% faster
Pricing - Intel is ~$150 more when factoring in the MB/CPU and reusing DDR4
If buying today, you could go either way.

Yeah DDR5 is for future proofing. But then again Intel is not known for releasing multiple CPU generations on one socket so it is not much of a future proof... 😅
 
Back
Top