Discussion Anyone else bored out of their mind due to mainstream CPU market stagnation?

[Wolverine2349]

But games do not care much about more cores.
6c/12t full cores can run all games out there without bottlenecking a 4090.
8c/16t give a decent reserve of CPU resources.
By the time games start using more than 8 cores to any significant degree, you will be better buying whatever is state of the art then than running some 8+ cores CPU from 2024/2025.

Most do not but some are starting to like certain areas of Cyberpunk, Starfield and Dragons Dogma 2. They benefit from more than 8 cores if you look at CPU usage metrics in game benchmarks.

Though for now its marginal but that may be starting to change.

 

[Wolverine2349]

That's still very much a rumor. I see no benefit in Intel making a new die of a soon to be outdated, problematic core. We also don't know what clocks might be before they run into a power limit. I expect ARL and Zen 5 to perform better in just about every scenario. Consoles are limited to 8 Zen 2 cores so I don't know what you think you are missing out on.

Well you may be on to something saying problematic cores. My hope was they would fix this and I could get 12 P core Bartlett Lake.

But can you trust it:

Uhm no:

Raptor Lake design appears to have serious flaws degradation is really severely bad and even lowering clock speeds while it will help it only delays inevitable probably or who knows.

So not going to buy Bartlett Lake if based on Raptor Lake.

If Intel was smart they make an all P core die of Alder Lake based Golden Cove instead of Raptor Lake Raptor Cove based.

Though in that case a 10 + 0 die. Unless 4 e-core clusters take up a little more space than single P core in which case maybe a 12 + 0 die still possible to take same space as 8 + 8 die? Or do 4 e-cores take exact same space as a 1 P core, so it would need to be 10 + 0. Either way Intel smart to release it as they have a budget platform that would kick AMD's tail on AM4 on LGA 1700 with a choice of at least 10 P cores and AVX512 and reliable unlike the disaster of Raptor Lake.

 

[igor_kavinski]

Either way Intel smart to release it as they have a budget platform that would kick AMD's tail on AM4 on LGA 1700 with a choice of at least 10 P cores and AVX512 and reliable unlike the disaster of Raptor Lake.

I forgot all about that! If Barty Lakey has AVX-512 enabled and it has 12 P-cores, Imma buyin' it, degradation or no degradation!

Compare Products

Compare Products

www.intel.com

I think they may release the 2455X in an LGA1700 package with two memory channels/AMX disabled. But if they deliver AVX-512 AND AMX, WOW WOW WOW!!!!

I'll restrain myself now because I'm totally not optimistic about Intel giving AMX or even AVX-512 to peasants but one can dream and maybe manifest?

 

[Wolverine2349]

I forgot all about that! If Barty Lakey has AVX-512 enabled and it has 12 P-cores, Imma buyin' it, degradation or no degradation!

Compare Products

Compare Products

www.intel.com

I think they may release the 2455X in an LGA1700 package with two memory channels/AMX disabled. But if they deliver AVX-512 AND AMX, WOW WOW WOW!!!!

I'll restrain myself now because I'm totally not optimistic about Intel giving AMX or even AVX-512 to peasants but one can dream and maybe manifest?

If its based on Alder Lake Golden Cove 10 + 0 or 12 + 0 and on a rinmg bus I will buy it. Not if based on Raptor Lake Raptor Cove. Cannot trust it.

No way buying a 2455X rebranded on LGA 1700. That is on a mesh or tile and has crippled latency and IPC compared to ring bus client Golden Cove.

I want a 10-12 P core preferably 12 on a single die and in Intel's case a ring bus. But it has to be something I can trust Raptor based of which I cannot. So only if Alder Lake.

Now if they release Lion Cove more than 12 P cores on a single tile I am all in. Its not a ring bus but new tile arch unlike their Saphire Rapids and a single tile is more like single AMD CCX unlike Intel mesh archs of prior gens like Sapphire rapids and such.

That is whether AVX512 or no AVX512. AVX512 not important to me.

Stability and the option for more than 8 cores on a single die are what's important to me. Though the later may not be an option anytime soon if ever unfortunately. And no I am not going to give a Raptor Lake 12 + 0 a try and risk degradation and stability problems even wit lower clocks as badly as I want more than 8 P cores on a single die.

Its either an Alder Lake based 10 +0 or 12 +0 CPU, or a hypothetical Arrow Lake 12 +0 CPU for me or just going to accept I will not be able to get what I want in trusted stability anytime in the next year if ever. In which case stick with 8 core single CCX X3D chips and upgrade them on AM5 and accept some games may be a little core starved but not too bad anytime soon even if more cores would be better for a few games now and upcoming.

 

[poke01]

What is this obsession with Xeon and AVX512? Both are useless right now.

What Intel needs is stability and a roadmap which is far more important to Intel and it’s customers.

 

[igor_kavinski]

What is this obsession with Xeon and AVX512? Both are useless right now.

What Intel needs is stability and a roadmap which is far more important to Intel and it’s customers.

Xeon IS stable because it undergoes far more validation testing.

AVX-512 for when you want to smirk seeing your PC score high on some obscure benchmark and for future-proofing

 

[Wolverine2349]

What is this obsession with Xeon and AVX512? Both are useless right now.

What Intel needs is stability and a roadmap which is far more important to Intel and it’s customers.

Agree. AVX512 does not matter for any gaming or consumer workloads unless you run Linux and RSPS3 emulator.

AVX2 is an instruction games will use.

More than 8 Big cores on a single CCX-CCD/ring bus/tile is what the focus should ne on. And oh yeah stability first. I am talking to you Intel. Intel scrap Raptor Lake dies right now (they were flawed and rushed in a record time) and make Bartlett Lake hybrid and P core only variants ALder Lake based even if it means 10 + 0 P core only die and 8+8 hybrid instead of 12 + 0 and 8+16. You cannot continue to risk the rushed unstable RPL smaller design of 8+16 and maybe same sized 12 +0 which fit in the 10nm same package maybe contributing to the stability issues??

Or maybe if 1 P core is equal to 3.5 E cores in space and not 4 e-core cluster, someone do the math than can make Bartlett Lake Alder Lake based 12 +0 P only die and the 8+8 die refresh. Stability matters and Alder Lake has it and had it unlike Raptor Lake.

 

[poke01]

Xeon IS stable because it undergoes far more validation testing.

AVX-512 for when you want to smirk seeing your PC score high on some obscure benchmark and for future-proofing

Xeon sapphire rapids beg to differ, it was the buggiest Xeon that Intel shipped. Intel’s validation team isn’t that good since they fired most of the good employees.

Bug Forces Intel to Halt Some Xeon Sapphire Rapids Shipments

Sapphire Rapids hits another roadblock, the resumption date is unknown.

www.tomshardware.com

There used to be a time when Intel Xeons had the best QA checks, not anymore.

 

[igor_kavinski]

There used to be a time when Intel Xeons had the best QA checks, not anymore.

Yeah but there haven't been any reports of SR Xeons crapping out in the wild, right? It took them a lot of steppings (a dozen or so?) but they finally got it working without errors

With RPL-S/RPL-S-R, they seemed to have run CP2077 and GB6.3 and declared it worthy of release!

 

[poke01]

Yeah but there haven't been any reports of SR Xeons crapping out in the wild, right? It took them a lot of steppings (a dozen or so?) but they finally got it working without errors

With RPL-S/RPL-S-R, they seemed to have run CP2077 and GB6.3 and declared it worthy of release!

Yeah true.

 

[igor_kavinski]

More than 5 months later, the CPU scene stagnation continues. I'm bored the heck OUTTA my mind!

Strix Halo is too expensive and that not even with 128GB RAM. That one may be priced like $5000.

9950X badly needs a newer, better IOD.

9950X3D, I don't expect any miracles from it since it's the same old 7950X3D formula with some new paint.

Fire Range is also going to be too expensive.

Tried my best to acquire a 9184X on the cheap but no dice. Lowest I saw it for was $999 on eBay. That's JUST the CPU!

Snapdragon desktop is still cooking.

Don't even mention Intel and Apple! They are BOTH dumb, just in opposite ways. Intel can't find their way out of nonsensical decisions made by foolish executives and Apple, well, they have always been too arrogant and stubborn to care about DIY enthusiasts.

AND NO BARTLETT LAKE STILL!

I challenge you to cheer me up.

 

[zir_blazer]

AND NO BARTLETT LAKE STILL!

I challenge you to cheer me up.

Didn't Intel recently said that they were actually going to release it?
And it is not only Raptor Lake Refresh Refresh. It is Alder Lake Refresh Refresh Refresh, since the 6P + 0E and 8P + 8E dies are still present in the lineup, heh.

On the positive side, the only other socket Intel side that had this longetivity is LGA 1151 if you don't include that they artificially segmented Skylake and Kaby Lake from Coffee Lake and Coffee Lake Refresh, but require some modding to get working on the other platform.

 

[igor_kavinski]

It is Alder Lake Refresh Refresh Refresh

As long as it is 12P+0E, boots with minimum DDR5-8000 CL40, hits 5.5 GHz ST and at least 5.1 GHz all core even with HT+AVX-512 without throttling on a decent air cooler, I'll be satisfied.

I need to FEEL the speed. My 12700K feels good at DDR5-7000 CL34 5.0 GHz all core but it's held back by the lame E-cores and disabling them gives me insufficient 16 threads. That's too low. Not that I need it. I WANT IT. AMD refuses to give us more than 8 P-cores on a single CCD so a 12P Bartlett Lake is pretty desirable.

 

[Hulk]

The "problem" with buying the top of the stack CPU as soon as it is available is that you have a long wait until the next generation for an upgrade. On the other hand if you buy lower down the stack you can trade up every 6 months or so when you get bored. Now I'm not considering economics here. Generally bang for the buck is to buy one or two steps below the top and keep if for years. I'm referring to enthusiasts who can't get off the upgrade train because it's fun. I have a 9950X. It's great but I'm "done" until Zen 6. There is no lusting after something faster.

 

[CakeMonster]

9950X badly needs a newer, better IOD.

9950X3D, I don't expect any miracles from it since it's the same old 7950X3D formula with some new paint.

Yeah, I can't cheer you up either. I had pretty much decided on moving to the 9950X from the 7950X because I expected a higher Zen5 uplift. I recognize the 9800X3D and 9950X3D are quite decent jumps, but the former is only 8c and the latter still has all the issues with scheduling. And no need to comment further about Intel WRT to its latest generation. I have no idea what to do actually, and the annoying part is CPU does matter increasingly now in games, so if we had better choices people would be eager to upgrade.

 

[misuspita]

I was expecting to go from my trusty 5700G to Strix Halo, but I still have every perf I need from it so it's kind of hard to justify 1500€ investment in a MiniPC when it still does everything I need. So will have to wait for Medusa Halo instead, darn it.

 

[DrMrLordX]

So will have to wait for Medusa Halo instead, darn it.

By that point, a used Strix Halo might be pretty cheap.

 

[DaaQ]

I am going from FX8350 to 9800X3D!! Haha ppl.

Talk about a jump. Really am hoping tariffs don't go into play because I would like a gen5 M.2 8TB if possible. So far its 2x 4TB g4

 

[DrMrLordX]

I am going from FX8350 to 9800X3D!! Haha ppl.

Try not to get whiplash.

 

[MadRat]

If some former intern junior associate was responsible for AMD x3D, can you imagine how many junior engineers are ignored at Intel? Intel used to always have the next drool-inducing product ready for the competition. Now its stuck in senior design lock, the same stale looks at tried and true. No sense of adventure. Intel once had the best memory management. The lowest latency. The best system responsiveness. The highest frame rates. It wrote industrial standards. And now its tired.

 

[mikeymikec]

If some former intern junior associate was responsible for AMD x3D, can you imagine how many junior engineers are ignored at Intel? Intel used to always have the next drool-inducing product ready for the competition. Now its stuck in senior design lock, the same stale looks at tried and true. No sense of adventure. Intel once had the best memory management. The lowest latency. The best system responsiveness. The highest frame rates. It wrote industrial standards. And now its tired.

I would be curious to see how Intel's R&D budget evolved between say 2014 and 2024, because surely when Ryzen first gen hit in 2017, that should have been a wake-up call and a pointer to their first mistake of resting on their laurels, then along came Apple M1 in approx 2020. I also wouldn't be surprised if a bunch of Intel's R&D people left because the company simply wasn't interested.

With Intel's level of resources, I would have expected a good counter to Ryzen by say 2020 instead of what appeared to be a long last gasp. Considering Intel's Goliath vs AMD's David in the era of nearly ten years ago, I fully expected AMD to produce a last gasp in response to Intel rather than the prize fighter that Ryzen has been. I wonder if Intel had expected the same and for AMD to run out of steam, allowing Intel to overtake them again, for things to be "back to normal".

Sure enough:


source: https://www.statista.com/statistics/263562/intel-expenditure-on-research-and-development-since-2004/

 

[coercitiv]

I would be curious to see how Intel's R&D budget evolved between say 2014 and 2024, because surely when Ryzen first gen hit in 2017, that should have been a wake-up call and a pointer to their first mistake of resting on their laurels, then along came Apple M1 in approx 2020.

The first years after Ryzen came to life were relatively easy for Intel, scaling grandpa Skylake was enough to stay competitive. Then came the Covid years... and Intel was in a very unique position at the time, their deprecated tech was in such high demand that they could not ship enough silicon to OEMs. They were swimming in money and therefore it was so easy to keep ignoring the writing on the wall.

I think it wasn't until Zen 3 came out that Intel could no longer avoid staring in the abyss. In theory they had been trying to steer back the ship for years, but in practice the comfortable position of printing money out of pure inertia never really allowed proper reform: always half-measures, never enough.

 

[Hulk]

I wouldn't be surprised if Intel went chapter 11 within the next few years. Can't believe I wrote that but I don't see them becoming relevant again. Too many optimistic plans and forecasts and no delivery on the promises time-after-time. I fear the end is nigh for Intel. I hate feeling this way because I have been such a big fan for most of my life but the writing has been on the wall for quite a few years I just didn't want to read it.

It's getting painful to watch them wiggle and squirm as they fight being washed down the drain. Raptor Lake degradation issues. Then ARL performance issues and no supply for the 285K. 20A gone with the wind and an over reliance on TMSC follows. Finally a breath of air with Battlemage but basically a paper launch with no real supply. They just can't get it done the way the old Intel could. I have a feeling the upper level management knows it's over but are just going to ride it out, grab all the money they can on their way to the exit. We're probably in the middle of the death spiral about now.

I was so hopeful for Lunar Lake. So much hype but in the end nothing really amazing from a performance point-of-view. Same with ARL except for some disturbingly low scores in a bunch of applications, indicating some type of architectural flaw to me. Then some hope with Battlemage but no supply. Intel is defeated. Every person they put in front of a camera is a corporate shill, just towing the company line, the real passion and passioniate people are gone.

I have a feeling 18A will fail as well. They will put out tiny quantity of something just to pump stock prices but it will be meaningless to the overall health of the company. It's sad. The competition, AMD, TSMC, nVidia, Apple, to name a few have surpassed them and there is no chance of catching up. Intel is the Titanic that hit the iceberg and is in the process of sinking while the band plays on.

 

[Thunder 57]

I would be curious to see how Intel's R&D budget evolved between say 2014 and 2024, because surely when Ryzen first gen hit in 2017, that should have been a wake-up call and a pointer to their first mistake of resting on their laurels, then along came Apple M1 in approx 2020. I also wouldn't be surprised if a bunch of Intel's R&D people left because the company simply wasn't interested.

With Intel's level of resources, I would have expected a good counter to Ryzen by say 2020 instead of what appeared to be a long last gasp. Considering Intel's Goliath vs AMD's David in the era of nearly ten years ago, I fully expected AMD to produce a last gasp in response to Intel rather than the prize fighter that Ryzen has been. I wonder if Intel had expected the same and for AMD to run out of steam, allowing Intel to overtake them again, for things to be "back to normal".

Sure enough:
View attachment 115049

source: https://www.statista.com/statistics/263562/intel-expenditure-on-research-and-development-since-2004/

Now compare that to the amount they spent on stock buybacks:

Intel invested $108 billion in share buybacks between 2005 and 2020.

Funny thing is their stock has still been crap. Too bad there is no breakdown by year.

How Intel's $108 billion buyback gambit backfired—a cautionary tale for tech giants | CTech

Intel’s focus on stock price over innovation left it playing catch-up in the AI race—and tech leaders should take note.

www.calcalistech.com

 

[Thunder 57]

I wouldn't be surprised if Intel went chapter 11 within the next few years. Can't believe I wrote that but I don't see them becoming relevant again. Too many optimistic plans and forecasts and no delivery on the promises time-after-time. I fear the end is nigh for Intel. I hate feeling this way because I have been such a big fan for most of my life but the writing has been on the wall for quite a few years I just didn't want to read it.

It's getting painful to watch them wiggle and squirm as they fight being washed down the drain. Raptor Lake degradation issues. Then ARL performance issues and no supply for the 285K. 20A gone with the wind and an over reliance on TMSC follows. Finally a breath of air with Battlemage but basically a paper launch with no real supply. They just can't get it done the way the old Intel could. I have a feeling the upper level management knows it's over but are just going to ride it out, grab all the money they can on their way to the exit. We're probably in the middle of the death spiral about now.

I was so hopeful for Lunar Lake. So much hype but in the end nothing really amazing from a performance point-of-view. Same with ARL except for some disturbingly low scores in a bunch of applications, indicating some type of architectural flaw to me. Then some hope with Battlemage but no supply. Intel is defeated. Every person they put in front of a camera is a corporate shill, just towing the company line, the real passion and passioniate people are gone.

I have a feeling 18A will fail as well. They will put out tiny quantity of something just to pump stock prices but it will be meaningless to the overall health of the company. It's sad. The competition, AMD, TSMC, nVidia, Apple, to name a few have surpassed them and there is no chance of catching up. Intel is the Titanic that hit the iceberg and is in the process of sinking while the band plays on.

I think there is still a window where they can turn things around. Not confident with their current leadership pulling that off though.