News Intel Comet Lake-S Overclocking: i9 CPUs Are Your Best Bet, MSI Data Shows

[Admin]

MSI analysis shows i9 chips are your best bet for overclocking Intel's 10th Gen Comet Lake CPUs.

Intel Comet Lake-S Overclocking: i9 CPUs Are Your Best Bet, MSI Data Shows : Read more

 

[jimmysmitty]

The silicon lottery can be fun. My Q6600 G0 was one that I could OC to 3GHz AND lower the stock voltage (based on its VID) with a still extremly stable CPU. Ran cool as ice as well. But my 4570K is not as happy of an overclocking chip.

 

[watzupken]

Why am I not surprised that i9 should have the better chips? MSI did their homework to quantify the odds of getting a grade A, B and C chip, but I am not sure how useful this information will be for a person looking to buy a new processor.

 

[hotaru251]

but I am not sure how useful this information will be for a person looking to buy a new processor.

if your in the tech circle you know its all silicon lottery unless you literally buy a pre-binned cpu from places like silicon lottery site.

news just gives u the odds but your still at mercy of luck.

 

[watzupken]

news just gives u the odds but your still at mercy of luck.

This is exactly what I mean. I mean it should not come as a surprise that you need to get the highest end chips to increase your odds. Knowing the % of A grade chips is all good to know information, but does not make it a sure shot way of getting it. Even if it says 27% chance of getting an A grade chip does not automatically mean if I buy 4, 1 should be grade A. So in short, it is still highly random.

 

[hotaru251]

So in short, it is still highly random.

hence "silicon lottery"

its gambling like real lottery.

(and places that sell binned cpu mark em up becasue they know ppl WILL buy them even at premium)

its a static %.
like flipping a coin.
50% chance of tails, but even if you flip it a billion times you could never land on tails becasue its a static % that doesnt change.

 

[TerryLaze]

but I am not sure how useful this information will be for a person looking to buy a new processor.

All tiers have a ~30% chance of only giving you default settings and maybe a little O/C.
All tiers have a ~50-60% to do at least a medium O/C.
Only the i9 have a decent ( ~30% ) chance of a high O/C

For a normal person looking to buy a new processor this info is interesting at most but not very useful other than seeing that they will get at least the advertised performance out of them,which already is an achievement in our day of age.

 

[jgraham11]

Tom's is losing credibility again...

This comment is not right and is completely contradictory to what MSI's research says:
"A Level C chip that made it through Intel's Q&A is almost certainly still good enough to perform as advertised -- you just won't be able to hit high overclocks with it. "

What are you basing that comment on? Why are you standing up for Intel without siting evidence for this? Blind faith...

This is going to be the same as the 9900k, overclocked right out of the box to operate at the highest frequency possible, power use be damned!

 

[TerryLaze]

This is going to be the same as the 9900k, overclocked right out of the box to operate at the highest frequency possible, power use be damned!

You are confusing motherboards with CPUs here,what you describe is what happens if you use a motherboard that doesn't comply to intel's guidelines and just pumps as much power into the CPU as possible.
That's the same for ryzen for FX and for earlier intel chips,a lot of mobos use way too much power.
At normal loads even when overclocked to 5Ghz the power draw is pretty low.

 

[PCWarrior]

So according to MSI:

1. For the i9 10900K/KF the required voltage for a 5.0GHz all-core overclock is on average 1.26V; for 5.1GHz is on average 1.30V; and for 5.2GHz is on average 1.35V.
2. The distribution of these chips is 27% great overclockers, 35% average overclockers, 27% limited overclockers.

From the above it follows that:
27% of the chips will do all-core 5.0GHz with a voltage lower than 1.26V
35% of the chips will do all-core 5.0GHz with a voltage equal to 1.26V,
27% of the chips will do all-core 5.0GHz with a voltage higher than 1.26V.

27% of the chips will do all-core 5.1GHz with a voltage lower than 1.30V
35% of the chips will do all-core 5.1GHz with a voltage equal to 1.30V,
27% of the chips will do all-core 5.1GHz with a voltage higher than 1.30V.

Also, following the same logic for the 5.2GHz data, it follows that:
27% of the chips will do all-core 5.2GHz with a voltage lower than 1.35V
35% of the chips will do all-core 5.2GHz with a voltage equal to 1.35V,
27% of the chips will do all-core 5.2GHz with a voltage higher than 1.35V

So it is safe to say that nearly all 10900K/KF chips will be able to hit an all-core overclock of 5.0GHz with a voltage that does not exceed 1.3V and an all-core overclock of 5.2GHz with a voltage that does not exceed 1.4V. Also 62% of them will be achieving 5.0GHz with a voltage that does not exceed 1.26V and 5.2GHz with a voltage that does not exceed 1.35V. And probably the top 27% will do 5.1GHz with around 1.2V and 5.4GHz with around 1.4V.

 

[cryoburner]

All tiers have a ~30% chance of only giving you default settings and maybe a little O/C.
All tiers have a ~50-60% to do at least a medium O/C.
Only the i9 have a decent ( ~30% ) chance of a high O/C

For a normal person looking to buy a new processor this info is interesting at most but not very useful other than seeing that they will get at least the advertised performance out of them,which already is an achievement in our day of age.

Not necessarily. As the article points out, "Intriguingly, the percentages don't add up to 100%."

If you look at the numbers in the graph, 15% of 10600K results are not accounted for, 5% of 10700K results are not accounted for, and 11% of 10900K results are not accounted for. Since these clearly don't fit in above the "A class", it's likely that they belong to an even lower "D class" that's not included in the graphs.

So, why aren't they included in the graphs? And why is MSI releasing overclocking details for processors that should still be under NDA restrictions? The answer is obvious, that these graphs are sanctioned by Intel, and they didn't want "D class" processors mentioned in what is essentially an advertisement to promote their new lineup.

In any case, I don't think there's any mention of what the sample size is for this data, and there's no way of telling whether the CPUs sent to MSI by Intel are even representative of how the processors available at retail will overclock, so not a whole lot can be gathered from this data. Results from other sources like Silicon Lottery will probably be at least a bit more meaningful.

Also, while their data may have shown the i9s overclocking slightly easier than the i7s, that may not be true in most real-world scenarios, where the extra heat from those additional cores is likely to be more of a concern, in a typical case, with a typical AIO cooler, and a power-hungry graphics card installed alongside them.

 

[spongiemaster]

This is going to be the same as the 9900k, overclocked right out of the box to operate at the highest frequency possible, power use be damned!

Only one company had to release multiple "fixes," with varying degrees of success, just so their CPU's could hit advertised boost clocks. It wasn't Intel.

 

[TerryLaze]

FTFY

Only one company had to release multiple "fixes," with varying degrees of success, just so their CPU's could hit advertised boost clocks between loads just barely enough so it shows up on benchmarks. It wasn't Intel.

View: https://youtu.be/3LesYlfhv3o?t=673

 

[cryoburner]

Only one company had to release multiple "fixes," with varying degrees of success, just so their CPU's could hit advertised boost clocks. It wasn't Intel.

No, Intel just had to release multiple fixes reducing their performance to fix security flaws. : P

 

[spongiemaster]

No, Intel just had to release multiple fixes reducing their performance to fix security flaws. : P

The patches didn't lower clock speeds. There is no implied guaranteed level of performance when buying a CPU. CPU's are expected to hit clock speeds on the packaging and marketing materials.

 

[PCWarrior]

No, Intel just had to release multiple fixes reducing their performance to fix security flaws. : P

Well at least Intel is trying to find and patch their vulnerabilities, unlike what AMD does. Intel’s vulnerabilities are well researched and as a result found by the good guys. Intel financially supports that research showing the importance they place on security. Who knows how many undiscovered vulnerabilities AMD cpus currently have and we are completed in the dark about them. Who knows if hackers haven’t already discovered AMD’s unknown vulnerabilities and aren’t currently exploiting them unbeknown to us all. It is like early February 2020 in the USA where you were falsely under the impression that the spread of coronavirus was limited and contained but that false impression was merely due to the lack of testing. AMD is not researching for vulnerabilities and those that were found came from Intel’s funded research and AMD has no plans to patch even those in order to not lose performance. Pathetic.

 

[cryoburner]

Intel’s vulnerabilities are well researched and as a result found by the good guys.

At least that's what you are assuming. The "bad guys" could have known about them along with other as-yet unknown vulnerabilities many years prior. There's no way to tell for sure. And sure, that goes for AMD's hardware too, but Intel's architecture is a more desirable target due to its much larger install-base, especially among servers, so it stands to reason that they would be getting targeted more. AMD also patches exploits, though there are limits to what makes sense to patch. If an exploit requires full physical access to the target system along with a bunch of others caveats that make it incredibly unlikely to perform, there's less need for it to be patched.

In any case, my point was that Intel's processors also see a lot of post-release patches due to flaws with their hardware, and they have resulted in the processors providing less performance than what people expected when they bought them. It was a half-joking but accurate response to a post suggesting that "only one company had to release fixes".

It is like early February 2020 in the USA where you were falsely under the impression that the spread of coronavirus was limited and contained but that false impression was merely due to the lack of testing.

Who was falsely under that impression? My stance from the start has been that its transmission is similar to the flu, and has the potential to spread to a similar degree if left unchecked, but that mitigations put in place should prevent its lethality from becoming anywhere close to prior large-scale outbreaks like the Spanish Flu, which many were comparing it to. And that's more or less been the case so far.