Intel's IPO Program Supercharges Underperforming "Arrow Lake" Chips, but Only in China for Now

[AleksandarK]

The long-promised gaming performance uplift for "Arrow Lake" processors is here, but for now, it is only available in China. Called Intel Performance Optimizations (IPO), this feature aims at system integrators rather than end users and offers a balanced approach between stock settings and full manual overclocking, all while preserving warranty coverage. IPO works by applying optimized profiles that adjust a range of CPU and memory parameters. On the CPU side, it fine‑tunes P‑core and E‑core frequencies, ring‑bus speeds, the UPI interconnect, D2D links between tiles, and both PL1 and PL2 power limits. For RAM, IPO raises transfer rates and tightens timings, pushing modules beyond their factory XMP or EXPO profiles. Early results from Chinese OEM Maxsun show that IPO can boost core clocks by about 200 MHz and elevate DDR5‑8000 kits to DDR5‑8400. In gaming tests provided by Maxsun, this translated to roughly a 10 percent uplift in frame rates, an encouraging figure given Arrow Lake's mild launch performance in gaming.

At debut, some Arrow Lake SKUs trailed their "Raptor Lake" predecessors because the chiplet‑based memory controller introduced extra latency and ring‑bus clocks ran slower. Supply‑chain constraints and tariffs have limited IPO's rollout to China so far, where pre‑built systems from vendors like Maxsun are shipping with the feature enabled. Warranty handling stays with the OEM, so users can enjoy the extra headroom without risking hardware support. Behind the scenes, Intel has been issuing firmware and microcode patches since December, most recently microcode 0x114 paired with CSME firmware 19.0.0.1854v2.2, to improve Arrow Lake's efficiency. IPO represents the next step in that effort and could become a model for global "opt‑in" BIOS presets if the pilot proves successful. For now, China is serving as the testing ground for Intel's latest attempt to squeeze more performance out of Arrow Lake. We have to wait and see if Western markets follow soon.



View at TechPowerUp Main Site | Source

 

[Dragokar]

What if this stuff burns the CPU/mainboard VRM? Is it still covered under some kind of warranty from Intel after the OEM/ODM prebuild seller warranty is done?

 

[Denver]

Desperate times call for desperate measures, and the i9 KSSS Plus Ultra 300W delivered brilliantly. The formula for success...

 

[_roman_]

Intel will have to deliver big that I will ever consider any Intel product again. I talk not about small performance gain, i talk about 400% or higher for any scenario with less power. Those intel wlan chips, two different types, were the final nail to refuse any future intel product.

I hardly can remember any intel product that worked. Maybe optane

 

[Why_Me]

I wonder if there's any benchmarks.

 

[ixi]

Intel will have to deliver big that I will ever consider any Intel product again. I talk not about small performance gain, i talk about 400% or higher for any scenario with less power. Those intel wlan chips, two different types, were the final nail to refuse any future intel product.

I hardly can remember any intel product that worked. Maybe optane

Last time I used intel was 6000k cpu . Since that moment intel was going in my view only in the ground, like, brutally.

 

[Scattergrunt]

Desperate times call for desperate measures, and the i9 KSSS Plus Ultra 300W delivered brilliantly. The formula for success...

Make it 600W and make the platform cost insanely high. Definitely is the formula for success then.

 

[Pizdarenkowitch]

Intel will have to deliver big that I will ever consider any Intel product again. I talk not about small performance gain, i talk about 400% or higher for any scenario with less power. Those intel wlan chips, two different types, were the final nail to refuse any future intel product.

I hardly can remember any intel product that worked. Maybe optane

Intel is coming back strong with awesome CPU's based on Ribbon-Fet & Backside power delivery.
So for now skip the trash to buy later when all this comes out.

 

[Scattergrunt]

Personally I just hope they keep going the Intel 200 series direction with efficiency in mind. If they can get the platform cost down and catch up with 14th gen gaming performance while not being too far behind AMD, they can earn back their market share. Intel can still compete. We'll just see if they actually do or not..

 

[alwayssts]

Personally I just hope they keep going the Intel 200 series direction with efficiency in mind. If they can get the platform cost down and catch up with 14th gen gaming performance while not being too far behind AMD, they can earn back their market share. Intel can still compete. We'll just see if they actually do or not..

I think they tried that (efficiency), and it failed (vs the AMD 7000 series). This is just not how Intel typically operates (nor what their processes are typically catered towards); maybe it could be in the (far) future.
I think it's very unlikely they get platform costs down, as that's been a main source of their income since forever (and also something they've taken pride in advancing in-house, which is expensive).

I think by doubling the design they are/were certainly going in the right direction with Nova, and would've (comparatively) done what you wanted if AMD are/were to use 3nm.

But if AMD is using a version of 2nm with high clock potential, they are not messing around and not going to be giving Intel any room to breathe.
I think in either direction; efficiency or performance, zen 6 will win if AMD uses 2nm. If AMD use 3nm, it could/would go the other direction, and this is perhaps why it could be happening.

At the end of the day I don't think Intel is going to claw anything back given their attempt to do so may have just forced AMD to use a better process, but that in itself is a good thing for consumers.

I'm actually happy to hear it would appear AMD is being so ruthless, as in the past I feel they may have used the more cost-efficient option and relied on their arch efficiency/cost to be the deciding factor for some.

I feel now that they've obtained both the market and mindshare, they're not going to let it go, which I feel is/was the right decision (if perhaps surprising given their tendency to stick to their formula in the past).

 

[Scattergrunt]

At the end of the day I don't think Intel is going to claw anything back given their attempt to do so may have just forced AMD to use a better process, but that in itself is a good thing for consumers.

AMD is only gonna be nice for so long before they start doing whatever cause they have dominant marketshare.. they could end up like Intel back in the mid 2000's to mid 2010's. Don't forget that

Competition is good. We ideally want both companies to be very equal in market-share as it demands the best out of both to get a leg over the other..

I think they tried that (efficiency), and it failed (vs the AMD 7000 series). This is just not how Intel typically operates (nor what their processes are typically catered towards); maybe it could be in the (far) future.

... What? 12th, 13th, and 14th gen were not that efficient. Especially not 13th and 14th gen. Intel has gradually gone more and more heat intensive since 9th gen, pushing their wattage higher and higher. And just because it isn't how they 'typically' operate doesn't necessarily mean they shouldn't do it

12th gen WAS an improvement but they immediately went back to running stuff hot again with 13th and 14th gen. I actually quite like 12th gen personally, and I think 12th gen / Intel 200 should be what they follow tbh for efficiency.

I think it's very unlikely they get platform costs down, as that's been a main source of their income since forever (and also something they've taken pride in advancing in-house, which is expensive).

AMD has shown that its possible to keep the same platform. There's a reason Intel doesn't do it obviously, but its pretty clear that its not really that much better to change platform. I can understand that reason but its clearly not working out tbh.. If the platform costs were down atleast maybe it'd make sense.

 

[alwayssts]

AMD is only gonna be nice for so long before they start doing whatever cause they have dominant marketshare.. they could end up like Intel back in the mid 2000's to mid 2010's. Don't forget that

Early/Mid 2000's, 2010's...now...I've always built with AMD (ok, I did use Sandy Bridge because I didn't like Bulldozer, which was complacent imho). I don't feel something like BD will happen again to that extent.
I think the potential move to 2nm (and 12/16 core CCDs; the use of both large cores at higher clock or more cores at efficient lower clock) proves that. I don't think they are, nor will, rest on their laurels.
Rather actually expand their position for different use-cases by introducing those designs to different/larger markets, perhaps even including the ARM ecosystem with something like Sound Wave.
Absolutely no reason to believe what you're saying is true. I get that at some point this general core (zen) architecture will become stale, but I think they're working on it for the move to a new platform (AM6).

You're correct ideally there would be excellent competition to spur better products and pricing (and I do think Nova is doing that; hence AMD perhaps using 2nm instead of 3nm), but also I don't think AMD taking the lead has ever truly left the ecosystem where their pricing was out of control; I think they're generally pretty realistic about it as gaining/maintaining marketshare has always been a concern/priority.

... What? 12th, 13th, and 14th gen were not that efficient. Especially not 13th and 14th gen. Intel has gradually gone more and more heat intensive since 9th gen, pushing their wattage higher and higher. And just because it isn't how they 'typically' operate doesn't necessarily mean they shouldn't do it

12th gen WAS an improvement but they immediately went back to running stuff hot again with 13th and 14th gen. I actually quite like 12th gen personally, and I think 12th gen / Intel 200 should be what they follow tbh for efficiency.

With everything after that; both on mobile (first; Core Ultra) and now with Arrow Lake on desktop. I agree 12th gen was a good one, but it also was the beginning of a ramp of an arch still reliant on high clocks.
It is in-fact why they shouldn't focus on efficiency, as their processes are built toward clocks rather than low-leak/density, which is widely-known and often discussed (as why odd as a 3rd-party foundry business).

I do think Nova will have high(er) clocks, and also why AMD is moving to 2nm; they have to compete in that realm (as they did with w/ 9000), rather than just cost/efficiency/cores, at least as an option.
It may have taken 30 years, but AMD has finally realized that the spirit of the 10ghz Pentium 4 never truly died; that's what Intel does best, and if they will compete again that is how.
AMD needs to keep parity with clocks for high performance, not just equalized performance through efficiency, especially since Intel may even have MORE parallelism than they will. And it would appear they are.

AMD has shown that its possible to keep the same platform. There's a reason Intel doesn't do it obviously, but its pretty clear that its not really that much better to change platform. I can understand that reason but its clearly not working out tbh.. If the platform costs were down atleast maybe it'd make sense.

I don't think we're truly arguing here; it's just how it is; Intel sells a new platform because they can and AMD doesn't because most know it's generally unneeded (and want to foster upgrades as a feature; lock-in).
Intel is clearly aiming towards new builders buying into a newer platform, while AMD often targets people that will solely upgrade their CPUs. There are pros/cons to both (for consumers and the businesses).
But it's also just how it is; both for financial reasons for each company and general strengths to their associated customers (People on Intel wanting those newer features; AMD customers longevity).

By all accounts this will continue: Nova will move to a new socket and use DDR6; a new and expensive platform. To some people this will be a performance/new tech advantage they see as worth the expense.
Others will buy a zen 6 and slot it into a then-ancient AM5 board with features they are content with, which also makes sense, especially if AMD improve their on-package IOD so mem speeds improve.

 

[N3utro]

Intel will have to deliver big that I will ever consider any Intel product again. I talk not about small performance gain, i talk about 400% or higher for any scenario with less power. Those intel wlan chips, two different types, were the final nail to refuse any future intel product.

I hardly can remember any intel product that worked. Maybe optane

Either you are young or you have a short memory. AMD / intel dominance has been a cycle for the last 35 years, and for all of these cycles intel is the one who had the lead for the longest time.

There were many times when AMD CPUs couldn't compare to intel. Biggest difference that comes in mind is when intel first introduced the "core i xxx" product line, with the popular i7 920. Amd had phenom X3 cpus at that time, and they were just not on the same level. Same way intel cpus are not on the same level than amds right now.

The current situation is in part due to intel loosing their edge on the foundry side to TSMC while AMD sided with them. But this is beginning to change. There are talks of a joint venture between TSMC and intel. Intel CEO has been replaced.

Intel is going to make a comeback at some point, no doubt in my mind. And i'm not an intel fanboy, got an amd cpu right now. But it's bound to happen like it has before.

 

[dyonoctis]

With everything after that; both on mobile (first; Core Ultra) and now with Arrow Lake on desktop. I agree 12th gen was a good one, but it also was the beginning of a ramp of an arch still reliant on high clocks.
It is in-fact why they shouldn't focus on efficiency, as their processes are built toward clocks rather than low-leak/density, which is widely-known and often discussed (as why odd as a 3rd-party foundry business).

I do think Nova will have high(er) clocks, and also why AMD is moving to 2nm; they have to compete in that realm (as they did with w/ 9000), rather than just cost/efficiency/cores, at least as an option.
It may have taken 30 years, but AMD has finally realized that the spirit of the 10ghz Pentium 4 never truly died; that's what Intel does best, and if they will compete again that is how.
AMD needs to keep parity with clocks for high performance, not just equalized performance through efficiency, especially since Intel may even have MORE parallelism than they will. And it would appear they are.

It's more that right now Intel needs high clocks to compete. The Pentium 4 wasn't Intel's best, it was a marketing arch designed to look nice on paper with those big clocks but had a ton of issues down the line because of that, and even lost to the Pentium 3 at launch. The Athlon was just a better CPU overall, and Intel had to play dirty to keep AMD in check.

And guess what, Intel's decades of domination started with a design that was inspired by the Pentium 3, and had low clocks. Whenever Intel starts to flex the clocks too much, things ultimately take a bad turn. Pentium 4, Rocket Lake, Raptor Lake.

 

[efikkan]

Personally I just hope they keep going the Intel 200 series direction with efficiency in mind. If they can get the platform cost down and catch up with 14th gen gaming performance while not being too far behind AMD, they can earn back their market share. Intel can still compete. We'll just see if they actually do or not..

So they fall a little short mainly in high-FPS gaming, and the rest are usually within margin of error, and they are faster in quite a few heavy workloads, so we are not talking about a "Bulldozer" type of situation here. Arrow Lake is undoubtedly more powerful than Raptor Lake overall (~15% shown in Linux tests), and considering how inconsistent Raptor Lake is performing, I don't think anyone well informed think Arrow Lake is somehow inferior in reality. That being said, those wanting to game at high FPS and low details will probably go for the other team as of right now.

I think it's very unlikely they get platform costs down, as that's been a main source of their income since forever (and also something they've taken pride in advancing in-house, which is expensive).

"Platform costs"? You mean like motherboards? (because CPUs are very competitive)
With AM4 AMD used to have great motherboard prices, but with AM5 there has been an edge in favor for Intel (Alder Lake/Raptor Lake/Arrow Lake).

But if AMD is using a version of 2nm with high clock potential, they are not messing around and not going to be giving Intel any room to breathe.

Hard to tell for us, as each node will be hit-and-miss. Overall, clock speeds are projected to decrease, not increase, so if they happen to strike a good one and go >6 GHz, then this will be an exception. Future gains will come from architecture, and I don't think the public have enough information to judge Nova Lake vs. Zen 6 yet.
And don't forget, the crazy clock speeds of Raptor Lake was mainly short burts, unless you didn't run it stock, which is technically overclocking. So if AMD achieves higher clocks, it will mainly matter for real usage if they manage to do it sustained and with mixed loads (some threads with high load, some medium, some low). Most benchmarks are unfortunately too short to really showcase this.

AMD has shown that its possible to keep the same platform. There's a reason Intel doesn't do it obviously, but its pretty clear that its not really that much better to change platform. I can understand that reason but its clearly not working out tbh.. If the platform costs were down atleast maybe it'd make sense.

I don't kow what you imply is the reason for Intel, but the real reason reliability. (Motherboard sales don't affect them that much)
AM4 has shown us that long-term support is not that easy, and requires a large matrix to show which chipsets supports which CPU family and features, and even then it's still dependent on the motherboard vendor. AM5 seems to be doing fine so far, but these aren't very different architectures.

I think what most people want is forward-compatibility, but that's going to be tough, and will in best-case have huge compromises. And who is realistically going to buy a motherboard and (presumably) a cheap CPU wanting to upgrade only the CPU 4-5 years later? (especially considering the expected lifetime of a motherboard is rather short compared to a CPU)
What's actually useful is backwards-compatibility, so if a board dies after warranty is expired, you'll have a chance to get a great board to keep the old CPU running, or to troubleshoot a dead system. Backwards-compatibility (with limited features obviously) if by far more achievable. But regardless, having a system that sort-of works is kind of pointless, and AM4 has shown us this is harder the more generations you try to support. And this puts a lot responsibility on the motherboard vendors who usually push out 15-20 variants of each board with very minimal testing, imagine them trying to verify 3-4 CPU generations (more if you count the APUs which are quite different) on all those boards… I think adding such support for workstation platforms would be more realistic.

And guess what, Intel's decades of domination started with a design that was inspired by the Pentium 3, and had low clocks. Whenever Intel starts to flex the clocks too much, things ultimately take a bad turn. Pentium 4, Rocket Lake, Raptor Lake.

It's kind of the other ways around; when someone have an insufficient architecture, they resort to desperate measures, which often have been pushing the clocks to the extreme. Pushing high core counts and lots of L3 cache fits in the same category, as even a minor IPC gain would be much preferred.

 

[Scattergrunt]

So they fall a little short mainly in high-FPS gaming, and the rest are usually within margin of error, and they are faster in quite a few heavy workloads, so we are not talking about a "Bulldozer" type of situation here. Arrow Lake is undoubtedly more powerful than Raptor Lake overall (~15% shown in Linux tests), and considering how inconsistent Raptor Lake is performing, I don't think anyone well informed think Arrow Lake is somehow inferior in reality. That being said, those wanting to game at high FPS and low details will probably go for the other team as of right now.

They're definitely worse in windows environments for gaming. Obviously Linux & non gaming environments they're undoubtedly better but its always their gaming performance that held them back. i wasn't trying to compare them to bulldozer. And we can already take a good guess why they're only a little bit worse anyway.. and you know what? Thats fine.

Its pretty clear that its a issue that can be rolled over with a new gen..

Realistically even if the 13th and 14th gen didn't have the terrible issues they did, it was good that they happened in the end imo, despite the amount of people F'd over it gave intel a rightful slap in the face and proved that making their CPU's wattage monsters isn't winning them any trophies.

"Platform costs"? You mean like motherboards? (because CPUs are very competitive)
With AM4 AMD used to have great motherboard prices, but with AM5 there has been an edge in favor for Intel (Alder Lake/Raptor Lake/Arrow Lake).

Overall cost to get ram, cpu, and mobo. Core Ultra 200 is pretty high compared to even other intel stuff. 12th gen and onward before core ultra 200 had the benefit of all being on the same socket (LGA 1700) so you could reuse the mobos, and that was nice.

Here in the states any decent core ultra 200 CPU is overpriced to all hell, and the mobos arent much better. LGA 1700 seems to be the sweet spot right now as I can find them for cheaper than

I don't know what you imply is the reason for Intel, but the real reason reliability. (Motherboard sales don't affect them that much)

There was certainly something more than that. I haven't really seen a reliability difference between AM5 and LGA 1700 mobo's personally

If I remember correctly it was something about them being able to give more new features to CPU's quicker & letting the mobo manufacterers do more as a result or something. Dont have a source so take that with a mountain of salt.

AM4 has shown us that long-term support is not that easy, and requires a large matrix to show which chipsets supports which CPU family and features, and even then it's still dependent on the motherboard vendor. AM5 seems to be doing fine so far, but these aren't very different architectures.

I mean I didnt say it was easy. But the fact they stretched AM4 out as far as they did and managed to keep them compeitive and arguably comparable products to Intel shows that its certainly something which IS possible. Do I think it should of gone on as AM4? Not really, 2 / 3 generations max would be nice though (AM4 was arguably 4 if you count Ryzen 2000 even though it wasnt that different from the first ryzen gen). AM5 & LGA 1700 had the sweet spot imo.

 

[efikkan]

They're definitely worse in windows environments for gaming.

Yeah, but the margins are smaller with higher details and resolutions, no-one really buys a top CPU and RTX 5090 to game on low, so these differences are always somewhat exaggerated. Those margins will be slightly less when gaming with a xx60/xx70 class GPU at 1440p/4K.

And we can already take a good guess why they're only a little bit worse anyway.. and you know what? Thats fine.

Yes, mostly because they don't do those absurdly high bursts of boosting.
And while it may be somewhat useful for competitive gaming, consistent performance and overall faster application performance is probably desirable for most users.

And it's not like generations with large changes usually don't have performance "regressions", they certainly do, but probably since Arrow Lake overall is just a "modest" advancement, these regressions may "feel" too significant. Perhaps if Raptor Lake wasn't so absurdly boosted, nobody would have noticed.

Its pretty clear that its a issue that can be rolled over with a new gen..

If you mean a good IPC uplift, then yes. Another ~15% should make it a decent performer over overall vs. Raptor Lake.
If it's just throwing cores at the problem, then no.

Realistically even if the 13th and 14th gen didn't have the terrible issues they did, it was good that they happened in the end imo, despite the amount of people F'd over it gave intel a rightful slap in the face and proved that making their CPU's wattage monsters isn't winning them any trophies.

I don't see how it was a good thing.
Their next gen would have had the same architectural priorities regardless whether Raptor Lake was clocked absurdly or kept the clocks comparable to Alder Lake.

Overall cost to get ram, cpu, and mobo. Core Ultra 200 is pretty high compared to even other intel stuff. 12th gen and onward before core ultra 200 had the benefit of all being on the same socket (LGA 1700) so you could reuse the mobos, and that was nice.

I know many like to focus on this kind of hypothetical scenario, but how many upgraded from Alder Lake to Raptor Lake? Like 1-2%? The CPU performance wasn't that much different, but the memory speeds advanced quite a bit. Keeping the motherboard for this kind of upgrade is really an edge-case scenario.

I mean I didnt say it was easy. But the fact they stretched AM4 out as far as they did and managed to keep them competitive and arguably comparable products to Intel shows that its certainly something which IS possible.

Well, let's not pretend like a motherboard from the first gen supported the last CPUs, it was more like cascading support, with most boards supporting a subset of that range. Even upgrading Zen 1 => Zen 3 wasn't always a given.

There is also the longevity of the CPU's usefulness; at the time Zen 1 was mostly competing with Coffee Lake(and even Kaby Lake initially), and I remember lots people arguing the Zen 1 with its glorious 8 cores would be more "future proof" were the same people who rapidly upgraded to Zen 2 or 3, while buyers of Intel's counterparts were satisfied with their purchase for much longer. (And I own both Zen 1+ and Zen 3 BTW.)

Do I think it should of gone on as AM4? Not really, 2 / 3 generations max would be nice though (AM4 was arguably 4 if you count Ryzen 2000 even though it wasnt that different from the first ryzen gen). AM5 & LGA 1700 had the sweet spot imo.

Zen (1)+ wasn't an architectural change, it was a stepping to fix bug(s), so I don't count that as a generation. As AM4 started with a few Excavator chips, it makes 4 generations in total. It's debatable whether Raptor Lake counts as a generation above Alder Lake, but regardless it's certainly not worth the upgrade. A worthwhile CPU upgrade should at the very least be in the 30% range, if not closer to 50% to justify an upgrade. I think would will find few examples where you get close to that without jumping between three generations.

 

[N/A]

Intel needs another Core 2 Duo like IPC boost, not the overclock that comes with an insane power cost. And make the memory controller integrated in the correct tile again. There's at least a 50% performance loss there for no good reason. The memory controller is just a small strip on the side of the tile. the GPU on one side. PCIE lanes on the other and job well done.

 

[_roman_]

Either you are young or you have a short memory. AMD / intel dominance has been a cycle for the last 35 years, and for all of these cycles intel is the one who had the lead for the longest time.

I had for a very long time only intel notebooks. I think most intel processors were 4 core + fake cores at that time. The 900 / 1000 nvidia graphic cards were also not that interesting to switch from my dated second hand notebook.

I started to pay attention with ryzen 1000. I bought quite late in the desktop platform at the end of ryzen 5000 when 7000 did not exists.

What I know for sure in past years, those security issues with intel processors kill a lot of performance. The stability issues of their wlan chips, especially AX200 and AX210, does not make it easier to buy intel. I read Intel has only defective Ethernet chips. Not one revision but nearly every revision. For the CPUs we all know the 430 or was it up to 480 Watt desktop processors. We all know about the stability issues which were not fixed in time for some 13 and 14 core processors. the intel ssd in my first refurbish notebook from lenovo had according to webpages some malware in that drive (source searching for that parrticular drive for that particular lenovo notebook).

--

the days of 35€ low end AMD4 mainboards are gone. Usual mainboard had cost 100-150€ for B550. 210€ for X570

Am5 mainboards still costs 90€. When you compare low end than compare it to low end am5 mainboards.

I doubt I ever saw any intel mainboard in that region.


old intel chipset - maybe compareable to am4.
60€ for H610 // the other mainboards also starts at 90€ without shipment

Mainboards Intel Sockel 1700 Preisvergleich Geizhals Österreich

Preisvergleich für Mainboards Intel Sockel 1700 ✓ Bewertungen ✓ Produktinfos ⇒ Auswahl und Filtern der Produkte nach den besten Eigenschaften und dem billigsten Preis

geizhals.at

This looks like the current intel Socket
Prices start at 96€ - more expensive as anyhting.

Mainboards Intel Sockel 1851 Preisvergleich Geizhals Österreich

Preisvergleich für Mainboards Intel Sockel 1851 ✓ Bewertungen ✓ Produktinfos ⇒ Auswahl und Filtern der Produkte nach den besten Eigenschaften und dem billigsten Preis

geizhals.at

Entry trash mainboards for am5 are 80€ // i know for sure some sellers had proper am5 mainboard for 90€ // this paid list starts at 97€ for the B650 mainboards

Mainboards AMD Sockel AM5 Preisvergleich Geizhals Österreich

Preisvergleich für Mainboards AMD Sockel AM5 ✓ Bewertungen ✓ Produktinfos ⇒ Auswahl und Filtern der Produkte nach den besten Eigenschaften und dem billigsten Preis

geizhals.at

edit: These prices are for those who argue the platform cost is high. Of course you get entry level peripherals and chipsets on entry level mainboards. Nothing wrong with that when you want to use only one or two dram modules, one storage with only one processor with a cheap second hand cooler.

--

And who is realistically going to buy a motherboard and (presumably) a cheap CPU wanting to upgrade only the CPU 4-5 years later? (especially considering the expected lifetime of a motherboard is rather short compared to a CPU)

I did that several times with notebooks. I plan to buy somehting from the second hand market in the future.

I bought my 7600X / X670 mainboard as a placeholder to change later the processor. In may 2023 there was nothing else available as the 7600x. I also went over the board with 2x32Gib 5600MT/s DRAM.

Cheap HP mainboards from hp microtowers (1 piece @ 4 or 5 years) and small formfactor hp computer (1 piece before 3 year end of warranty) die quite fast. Experience from work.

I also think with AMD you will most likely get easier mainbaords as AM5 will most likely have several different kind of chipsets and manibaords. With intel you are limited to certain small range of mainboards.

 

[Caring1]

Intel is coming back strong with awesome CPU's based on Ribbon-Fet & Backside power delivery.

Intels been giving it's consumers the backside power delivery treatment for years, expecting us to just bend over and take it too.

 

[Tek-Check]

Intel is coming back strong with awesome CPU's based on Ribbon-Fet & Backside power delivery.
So for now skip the trash to buy later when all this comes out.

Indeed... We have heard this narrative several times in recent years. Not once it turned out to the true.

Personally I just hope they keep going the Intel 200 series direction with efficiency in mind. If they can get the platform cost down and catch up with 14th gen gaming performance while not being too far behind AMD, they can earn back their market share. Intel can still compete. We'll just see if they actually do or not..

Are you sure you know what this means? "Keep going" the same direction means keep losing market share.

 

[Prima.Vera]

Intel will have to deliver big that I will ever consider any Intel product again. I talk not about small performance gain, i talk about 400% or higher for any scenario with less power.

This are not the '90s anymore. We are already at the limit of the physical transistor size, gateway size, number of transistors, etc. For 400% performance uplift, one needs to go atomic size into picometer sizes, which right now, it is technological impossible. The CPU, as we know it,already reached its limits. Right now, they are just doing minor tunings and optimizations, but that is it mostly. Just look how abysmal are the 3nm (more like 48nm true size) yields. The GPUs on the market are at least 10x times less than they were 15-20 years ago because of those yields. 20 years ago I could by a GPU or a CPU on the launch date from the local retail store. Good luck now finding it after a couple of months at scalped prices.

 

[efikkan]

There's at least a 50% performance loss there for no good reason.

You mean like in a crazy edge-case scenario? (potentially something synthtetic?)
If not, it seems like you might be off by a couple orders of magnitude.

I also think with AMD you will most likely get easier mainbaords as AM5 will most likely have several different kind of chipsets and manibaords. With intel you are limited to certain small range of mainboards.

It depends on where the platform is in its lifecycle. AM5 might have a wide selection now, but it took well over a year before we saw a decent selection of "mid-range" boards. Arrow Lake's LGA1851 is by comparison in its infancy, but still is starting to look good, with all the big vendors (Asus, Asrock, MSI, Gigabyte) having multiple tiers of each chipset (which I think is unnecessary BTW). The predecessor, LGA1700, had a much greater lineup and better pricing than AM5 throughout its run.

 

[Power_VANO]

@AleksandarK there is 0x117 microcode update widely available for Arrow Lake already from multiple motherboard manufacturers. Would be great to do a quick test and comparison, if possible.

 

[Dr_b_]

legit question, with these chips, will the glue that holds them together under the IHS melt and disambiguate the CPU?

This are not the '90s anymore. We are already at the limit of the physical transistor size, gateway size, number of transistors, etc. For 400% performance uplift, one needs to go atomic size into picometer sizes, which right now, it is technological impossible. The CPU, as we know it,already reached its limits. Right now, they are just doing minor tunings and optimizations, but that is it mostly. Just look how abysmal are the 3nm (more like 48nm true size) yields. The GPUs on the market are at least 10x times less than they were 15-20 years ago because of those yields. 20 years ago I could by a GPU or a CPU on the launch date from the local retail store. Good luck now finding it after a couple of months at scalped prices.

I have to agree with this sentiment, largely, but have hope that clever designs like backside power, leveraging quantum effects for advantage, further optimizations and yes even more process improvements including density will make this happen, humanity should not give up. But whether this comes from Mr Funkytown is a coin toss