Question Why Doesn't Intel and AMD Release CPUs That Can Run At Their Usual OC Speeds?

[geokilla]

Intel's 10th gen and 11th gen CPUs are pretty decent overclockers. 10th gen K series can hit 4.8GHz all core with little trouble and is 100% stable provided you have good VRMs and sufficient cooling. Same can be said for AMD CPUs. So instead of Intel releasing the 10600K at up to 4.8GHz boosted, why don't they release a 10600K that runs at 4.8GHz all core instead of 4.1GHz and then have it boost to 4.8GHz under the right conditions? They would be able to charge more, take away overclocking, and run all their K CPUs at their max performance. They wouldn't have to rely on consumers and reviewers to do all that work. I mean, who doesn't want to get the most out of their PC? I know overclocking is all about binning but when pretty much ever single K processor can run 4.8GHz, does the question of binning and silicon quality matter anymore?

 

[naukkis]

They release specs for motherboards and their cpu's has to work with them. If they spec motherboards to be able to absolutely maximum overclock all motherboards needs to be build like the most expensive ones now.

Not speaking that overclocking all out from cpu is pure stupidity, stability will be compromised and efficiency is absolutely redacted-if they start to build cpu like that at least EU will ban them from market because their inefficiency.




Profanity is not allowed in the technical forums.


esquared
Anandtech Forum Director

 

[DrMrLordX]

Many reasons why not:

1). CPUs sort-of overclock themselves already.
2). It takes a long time to dial in an overclock like an enthusiast. It would increase binning time/costs.
3). Warranties could no longer be issued, or would have to be lowered.
4). TDPs would go up

 

[scineram]

Because they cannot. I bet the vast majority of conzooomer overclocks would never pass the validation and testing that AMD and Intel require.

 

[VirtualLarry]

Because they cannot. I bet the vast majority of conzooomer overclocks would never pass the validation and testing that AMD and Intel require.

Eggzactly. PCs are used for MORE than just games, some people forget that.

Remember the Pentium FDIV bug? Yeah, Pepperidge Farm Intel remembers.

Now imagine if CPUs were exhibiting intermittent similar bugs in calculations, not because of silicon defects or design, but because of "excessively-high binning".

I recall someone with a female avatar on here a long time ago, mentioning that the 45nm Core2Quad CPUs, when overclocked to a "typical" enthusiast "guaranteed" overclock, would encounter obscure errors in like SSE3 / SSE 4.1 or whatever the bleeding-edge opcodes were at the time, used in some esoteric video-encoder to speed it up.

 

[zir_blazer]

Now imagine if CPUs were exhibiting intermittent similar bugs in calculations, not because of silicon defects or design, but because of "excessively-high binning".

Why imagine when the Pentium 3 Coppermine 1.13 GHz works as an example of that? And before that, the K6-II 550 MHz was also considered a factory overclock.

 

[geokilla]

But isn't Intel already doing this with their non K processors? The 10400 literally runs at 4GHz with a simple click in the BIOS that is apparently defaulted on. Now I don't know if the same applies to OEM PCs or CPUs and PCs that are sold to big organizations where they value stability over speed due to the type of programs that they run, but that's where Intel and AMD would sell them a slightly slower processor that has guaranteed stability right? Like how AMD has their consumer lineup and then the Ryzen Pro lineup.

 

[zir_blazer]

But isn't Intel already doing this with their non K processors? The 10400 literally runs at 4GHz with a simple click in the BIOS that is apparently defaulted on. Now I don't know if the same applies to OEM PCs or CPUs and PCs that are sold to big organizations where they value stability over speed due to the type of programs that they run, but that's where Intel and AMD would sell them a slightly slower processor that has guaranteed stability right? Like how AMD has their consumer lineup and then the Ryzen Pro lineup.

What you're describing is MCE (Multi-Core Enchantment), and that is not done by Intel but by Motherboard manufacturers. They toy with some of the Processor Turbo parameters to make it clock higher than it would by default in all-core load scenarios. This isn't within the Processor specifications, it is outright overclock, and usually done without the end user knowledge.
Also, just because an individual Core is stable at the highest nominal clock speed, doesn't means that ALL cores would be able to do so simultaneously. For example, the super-duper factory overclocked AMD FX-9590 had a base 4.7 GHz and a single core turbo of 5 GHz. I remember that it was already quite hard for overclockers to get to 4.9 GHz sustained all core base, much less the actual 5 GHz. Sure, that specific model was at the absolute limit, and maybe in low end, low core models you can easily sustain the max all core Turbo clocks for all cores all the time, but in the highest end models that already requires already a lot of binning it may not be possible without hitting a power or thermal limit even if the silicon was capable.

 

[Tup3x]

AMD CPUs are more or less maxed out and there's hardly anything to gain with overclocking (and it usually just hurts single threaded performance).

 

[TheELF]

Because they don't need to...
These are desktop CPUs and getting a few hundred Mhz more won't change anything for the user experience in multithreaded workloads, they are still going to take hours, or for games you will get like 10FPS more at most in games that already run at way more than 100FPS, but they are going to get much hotter and increase the chances for it to break down much sooner.
Getting a super high single thread though does increase the user experience, from badly coded games to just waiting shorter times on mundane everyday things.

 

[aigomorla]

Oh and Intel has done what you asked before.
It was called the 8086.

Intel® Core™ i7-8086K Processor (12M Cache, up to 5.00 GHz) Product Specifications

Intel® Core™ i7-8086K Processor (12M Cache, up to 5.00 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

This guy ran at 4ghz all 6 cores all the time back when it took overclocking to get 4ghz.
It also peaked an insane 5ghz at stock, when that always required overclocking.
However.... it required a very large heat sink to keep in check. It even STATED it didn't come with a heatsink, because it required a large heat sink.

Included ItemsPlease note: The boxed product does not include a fan or heat sink

This concept is completely lost or does not register to 70% of the average novice PC builder, like how also power supplies are majorly under evaluated.
This also lead the product to be somewhat under valued, as they were in essence cherry picked 8700k, and people wanting that special hand picked cherry was not as many as they thought.
And they required Delidding the CPU as intel refused to go back to soldering the IHS at that time.

This is why most people also said no we don't want this.
This is why intel said ok, we wont do this anymore.

 

[jpiniero]

^^^^

9900KS

 

[Mopetar]

It doesn't even make a lot of sense below the high end because you'd wind up with half a dozen or more models to account for all the different bins you could create. Look at the server space where you can have a load of chips with the same core count separated by only a few 100 MHz on the clocks.