Stop Neutering Your Products, Intel

[Plouffe]

MSI's MAG B660M Mortar Max motherboard has a secret chip up its sleeve that people all over the world won't get access to until after Raptor Lake launches. Is Base Clock Overclocking worth the hassle in modern games and productivity? Why are we locked out of this on non-K CPUs if we want to try it out?

 

Buy an Intel Core i5 12500: https://geni.us/kv6yR

Buy an MSI MAG B600M Mortar: https://geni.us/eT7fa

Buy a kit of Crucial Ballistix 3200 DDR4 RAM: https://geni.us/kuXYf

Purchases made through some store links may provide some compensation to Linus Media Group.
 

 

 

[Senzelian]

This thing just rekt my 5800X. That only gets 14k points in Cinebench R23. Don't ask me why the score is so low, I have no idea.

I might have to re-install Windows.

[Nystemy]

To be fair, overclocking seems to largely be on the way out. Or rather, already comes stock thanks to improving boosting as the generations tick by.

 

Yes, CPU boosting isn't close to perfect, but if we look at it over time. Things have gotten a lot better, and it likely won't be that many more years before a CPU is clocking itself adequately enough for everyone but the most well versed overclockers to struggle getting something better out of it. _{(It isn't like these companies are looking at every opportunity to edge themselves past the competitor as long as it doesn't create too large direct costs/consequences.)}

 

Though, it is debatable how far out on the knife edge one wants to run it. Yes, one can reduce core voltage a bit more and push the clocks a little higher, but at the expense of system stability. At some point it just won't be stable enough to work from a practical sense. _{(I for one don't like encountering system/program crashes on a monthly/weekly basis at least. But I don't know about you.)}

 

But to a degree, both Intel and AMD can't ship a product that is unstable in its stock configuration. Since it really hurts in the long run. _{(There is reasons why it took a good couple of years for AMD to properly get into the laptop/prebuilt space with their Ryzen products, their earlier CPU architectures had down right abhorrent driver/firmware issues plaguing them. (HP, Dell, Lenovo, etc don't want even more support tickets than they already get) But enthusiasts don't mind the annual system crash if it means a cheaper system with similar performance (not that they actually tended to crash that often, nor that it were a guarantee. But larger vendors looks at the single percentils with a very critical eye). And AMD still to this day haven't really noticeably broken into the server market, despite Epyc being a very competitive product. Here stability is paramount. (even with redundant servers and workload spreading over multiple nodes still don't remove the base desire for hardware stability. A server should only be offline if it has a good reason for such, hardware/system crashes isn't a good reason. Since then one changes to a more stable platform/OS.))}

 

However, stability is debatable. Some wants more and others rather want more performance.

And to a degree it is debatable if modern CPUs aren't already throwing away some stability when boosting, and what stops Intel and AMD from throwing away a bit more stability in the future?

 

As long as the decrease in stability don't start to reflect noticeably in the statistics compared to the prior generation or two, then no one would care to be fair. Since who currently tests long term stability of CPUs?_{(perhaps LTT labs in the future)}  Everyone seems fixated at performance, and power efficiency comes second on the list _{(or first for datacenters)}. Both of these can trade away stability in exchange for themselves. But yes, manually one can push away even more of that stability in exchange for a bit more performance.

 

[Middcore]

As overclocking loses relevance, the window where Intel will be able to get people to pay more for this feature is closing. 

[Nystemy]

8 minutes ago, NastyFlytrap said:

Okay, but that probably just means that the feature is going to disappear, and after a while we'll be wishing we could overclock something, and we wont be able to because it wont exist by then

Intel wouldn't kill a cash cow.

 

And enthusiasts won't stop wanting overclocking support, regardless of how good clock boosting gets.

 

Not to mention that a lot of people buy unlocked CPUs thinking that they run better than their locked variant does. _{(Ie, a fair few people buy overclockable chips and don't even touch any of those settings they paid good money to be able to use.)}

 

The overclocking crowd likewise tends to flock around the highest end CPU, so Intel can rather trivially just notch down the "stock" numbers a bit, and let the clock boosting system run at slightly higher voltage than "optimal", ensuring that the overclocking crowd still has a tiny amount to wave their banner for.

 

However, as I already outlined, Intel still has to ship a stable product, so overclocking will still give users the ability to venture out into the land of instability.

 

So no. The K series of CPUs is not going away. It just doesn't make any economical nor marketing sense to discontinue it.

[Nystemy]

4 minutes ago, NastyFlytrap said:

Okay, but the unlocked CPU's do run a bit better out of the box than the locked parts, or atleast used to!

I have an 8700 and its not as fast as an 8700k, overclocking or not.

According to Intel's naming sheet it just means "unlocked".
https://www.intel.com/content/www/us/en/processors/processor-numbers.html

However, chip to chip variance can still be fairly big. _{(whole percentages isn't unexpected to be fair. And with boosting that makes a big difference from one chip to the next.)}

 

There is though KS versions, as in "unlocked" and "special", where "special" usually means cherry picked for performance. _{(X and XE means the same as KS. As Gamer's Nexus has stated, "X" is an expensive letter to market and why companies loves using it.)}

But K alone should by Intel's own rules be no different than a non K. _{(other than the clock multipliers being unlocked for the user to set to whatever they desire.)}

[Nystemy]

3 minutes ago, NastyFlytrap said:

Yes, but i meant that by default, the 8700 is locked to 4.3Ghz max boost, meanwhile the 8700k is 4.7Ghz max boost

Thats just the stock limits, hence why i said its slightly faster

Well, have to say that Intel does wrong there.

Since by their own naming convention they should both have the same spec. The 8700K shouldn't run 500 MHz faster at base and 100 MHz faster at boost. _{(Ie, the 8700 should likely run at these speeds as stock too, or something in-between in regards to both.)}

 

If it were a 8700KS it would be fair for it to have these higher specs.

 

Intel is stupid for not following its own naming convention, since why else have a naming convention...

[MenacingExiler]

Linus tried on B660 but have anyone tried on Z690 if it's cheaper?

[GodAtum]

What system monitoring tool is he using that shows power draw?

[Psittac]

YESSSS!!!! My first DIY PC was an AMD Barton Core running at 14x222 cas2 on water.  I was a generation late and I went through at least 5 motherboards with the fastest total clock being the Abit AN-7 14x222 and the dfi (non lan-party  *edit: Infinity!!!! it was the infinity) being the highest fsb on the same non-samsung B-die samsung mushkin 256mb sticks at 10x235 cas2? or 2.5

 

Sorry for not having the correct names but it was a looooooong time ago and THIS is what overclocking is about.  All in I could have purchased an Athlon64 and skeezed the bejesus out of the fsb but I couldn't afford it at the time and got everything a little at a time.  The only reason I went through so many motherboards is that I took advantage of newegg and kept returning my broken motherboards.  I had an iwaki industrial pump running on a cpu block with push in soft tube fittings..... as in the fitting's were built into the cpu block and the soft tube inserted into it   It rarely leaked but considering I was suffering from some major mental illness I was awake for 24 hours at a time and the ONLY thing I did was clock test reclock and retest then break my machine and have an existential crisis fearing that it was the last time newegg would exchange my motherboard.  Gotta say that ati AIW-9600pro AGP was built like a tank because every time the water would drip on that and in turn kill the mb, that card lived until it was useless.  But in the decade and a half or more I have brought as much of my own business and friends and family to newegg as a "thank you" for being so chill to someone who was going through some significant turmoil.  (maybe paul was behind my RMA's?)

 

At any rate I blew EVERYONE out of the water on that platform in every benchmark at the time because back then, if you put litteral days weeks and years into adjusting your voltages and timings there really was almost no end to the OC you could achieve?  I think that eventually I either slightly degraded the silicone or found the end's of it's capability (litterally one subtiming at a time). (that ATI card OCed to high heaven with it's little 4 pixel pipelines too).

 

Ahhhhhhhh the day's of literally seeing your OC at work during boot and in Half Life 2, stutter and lag was always there but you could chew away at it one fsb mhz at a time.  You could visually see the difference between cas 2 2.5 and 3 even when using windows

[Pere_U]

In the vidéo, at 6:00 minutes, you said 5.25ghz with 1.2v.

I have the same setup and it seems impossible...

@Plouffe what settings are used please?