Is The 9900k Being Over-Rated Or Am I Missing Something?

[Daharen]

So for 'stock' performance, hands down, I agree, the 9900k is the best gaming CPU on the market, and 'beats' the 8700k... But here's the thing... Who the hell is buying CPU's like this for 'gaming' and running them at stock? I mean generally speaking, we're definitely into the enthusiast tier once you're looking at 9900k's or 8700k's and I have to figure it goes without saying that you were 'probably' going to try to overclock, so I'm 'inclined' to believe that they are testing these chips in the wrong way, but I'll explain further...

For me what 'really' matters in an enthusiast chip, that everyone 'knows' most consumers are going to overclock, is the break down of the chips overclockability. Now I've asked a few of the reviewers who got these chips whether they were good overclockers, and they told me they were, and that they got up to "5 Ghz"... But here's the thing, if you're an enthusiast, 5 Ghz isn't the margin to beat anymore, and the benchmarks are close enough in gaming that a 1 Ghz difference in overclocking will ACTUALLY cause the 8700k to beat the 9900k in most games, which means it's relevant if it's a better overclocker.

Now Silicon Lottery hasn't released it's specifications yet, but it is revealing that if you go over to Overclocker UK's market, they aren't binning any 9900k's above 5.1 Ghz as stable, and they binned 8700k's as high as 5.3 Stable (At about 6%), and 8086k just as high (I know they're the same chip, but they were binned at 15% for 5.3). 

Given how close the margin of improvement is between these chips, it 'seems to me' that if you take the top 10% of 9900k's and compare them to the top 10% of 8700k's at maximum overclock, the 8700k's win... And... here's the thing... We have no reason to 'believe' that this doesn't hold true for all other percentiles of the two chips... So it seems to me that 'grading' the chip based on what intel sets their stock frequencies to, is kind of a bad way of measuring the actual quality of the chip. After all, stock speeds are really fairly arbitrary, and if at a given percentile, all 8700k's overclock better than all 9900k's then the 8700k is the 'better gaming GPU' pretty objectively for the realistic case use of these chips.

Is there anyone who sees a flaw in my logic?

[Electronics Wizardy]

Give them some time for the 9900k binning, often the batches get better over time, and sillicon lottery just started binning these.

 

And the performance difference here is tiny, you would never notice this difference.

[dizmo]

The 9900k is more for someone who wants it all. The 9900k beats the 8700k when it comes to editing type workloads, no?

[Streetguru]

6 minutes ago, Daharen said:

 

The 9th gen is a joke

 

there's a reason they didn't release an 8700K successor.

[Daharen]

Also, to be clear, I've seen 'review samples' where the reviewers claim it can get up to 5.3 Ghz stable, but weirdly these parts don't seem to be making it to the businesses that make a living binning CPU's and reselling them. Also there are CPU's that are TRUE Silicon Lottery winners, and it is possible that a reviewer got one (Purely by chance even if we want to give intel the benefit of the doubt), and that overclock 'much' higher than average, and aren't marketed by the CPU Binning companies at their potencial overclock, because they represent a sampling so low that they can't ever ensure stock no matter how many chips they purchase. 

If a reviewer claims the 9900k can reach stable 5.3 Ghz, but users aren't experiencing this, then that's no different then the lucky people who win the 'super lottery' on their 8700k's and can keep them cooled and running stable at 5.5 - 5.6 Ghz at 1.45 v. These chips exist, but represent less than 1% of all chips, and so aren't marketed even by secondary sellers. 

[KarathKasun]

As clock speeds rise, you get less OC headroom.  But you also need to OC more to get a tangible benefit.

This is why Intel primarily builds their product stack around core/thread/IO counts.

[Daharen]

1 minute ago, dizmo said:

The 9900k is more for someone who wants it all. The 9900k beats the 8700k when it comes to editing type workloads, no?

Oh absolutely, it beats the hell out of the 8700k as a workstation card, but it's on a chipset that almost never sees workstation use, and is mostly for gamers, and to boot, it is 'explicitly' advertised as the best gaming chip, with reviewers agreeing reluctantly. My point is I don't even know if they should be giving their reluctant agreement. 

[Electronics Wizardy]

5 minutes ago, Streetguru said:

there's a reason they didn't release an 8700K successor.

huh? what is the 9700k? Its normally a bit faster, and not having ht solves some issues with security and performance.

[Daharen]

9 minutes ago, Electronics Wizardy said:

Give them some time for the 9900k binning, often the batches get better over time, and sillicon lottery just started binning these.

 

And the performance difference here is tiny, you would never notice this difference.

I think you're totally right, I think revision 2 of the 9900k will feature slightly better solder, and a smaller silicon die. For me I'm sitting on a 9900k, and 'contemplating' returning it, since it wasn't pre-binned and I would have to get the die lapped in order to have a good shot at competing with the 8700k on revision 1.0 and I already have a 5.3 Ghz 8700k, so I'm reluctant to go through the trouble just to get identicle performance, or get unlucky, and have a dud, that gets less, and of course, not be able to RMA or return it because of modifications. 

Honestly, my 8700k was also 'binned' at 5.3 and I plan on lapping and direct die cooling it anyway, so it will likely clock above that, even if only 0.1 Ghz, and consequently, I am pretty sure 9900k revision 1 can't compete even with modifications, but I don't know that unless I try, so... I 'kind' of wish the reviewers had been a little more thorough in their equalizing. I get that they're not really targeting people doing the silly shit I'm doing, but still.

Edited November 13, 2018 by Daharen
Number Error 1 Ghz = 0.1 Ghz

[Electronics Wizardy]

Just now, Daharen said:

, I think revision 2 of the 9900k will feature slightly better solder, and a smaller silicon die

Very unlikely they will make different version, and the die size will be the same, but there are steppings that are slightly different, but you won't notice the change, performance is the same.

 

 

But really dude, your at the point of never noticing, just enjoy your 8700k, the fps difference is tiny

 

[Streetguru]

1 minute ago, Electronics Wizardy said:

huh? what is the 9700k? Its normally a bit faster, and not having ht solves some issues with security and performance.

No consumer cares about any security issues, or any small performance hits, single threaded performance is basically the same, especially in gaming, and you lose multi-threaded performance for very little gain.

They didn't release a 6c/12t 9th gen chip so they could keep clearing stock of 8700Ks and 8086Ks

[dizmo]

49 minutes ago, Electronics Wizardy said:

huh? what is the 9700k? Its normally a bit faster, and not having ht solves some issues with security and performance.

Your avatar is probably my favorite of any I've ever seen.

[xg32]

1 hour ago, Daharen said:

.

I don't think people are upgrading from 8700k to 9900k

 

Everyone's logic is different, i had a passed down 8600k (it bottlenecked the hell out of mhw) and had to choose an upgrade with the 2080 ti, can't go amd for gaming, silicon lottery no longer sells delidded 8086k, might as well get the 9900k (consoles and games are probably gonna use 8 cores). I probably won't sweat too much if an 8700k at 5.2 is like 2% faster in some games. For games that uses more cores, the 9900k won't be slower, even if someone gets unlucky and gets a 4.7 chip.

[Rune]

I have my 9900k at 5.3 and that was only with one day of overclocking. Giving it some time before I push it any further.