How rare are silicon lottery winners and losers?

[DELTAprime]

My build last January was my first build since the days before overclocking both on the CPU and GPU was dead simple (no K parts existed). I ended up with a 6600k that is stable at 4.6ghz with only a .05 overvolt on air (I can hit 4.7 stable but it requires more overvolting and heat than I like). Most 6600k's I can find online are only stable up to 4.4ghz and with a bigger overvolt. I'm guessing it was a silicon lottery winner. So just how common is it to get a good chip like this vs one that's average vs one that can't overclock for nuts?

[Oshino Shinobu]

It's really impossible to quantify that without actually testing a huge amount of chips. It could change based on the production batch or it could be entirely random. 

 

EDIT: The best samples will often be kept aside by the chip manufacturer to be used in their higher end CPUs, so the "winners" are often just the ones that didn't make the cut and the "losers" are the ones that can meet the requirements of the chip, but not much more.

[samcool55]

I guess the big losers run into stability issues after a few weeks/months when running at stock speeds so they just get RMA'd and fixed.

 

And with big winners it depends what the goal is with the chip.

 

It's possible a chip overclocks like a dream on air or water but completely farts around on ln2.

Of course the other way round is also possible, crap on air/water but great on ln2

 

Even tho in general the contrast between the 2 isn't that big, there are of course exceptions.

[Kinda Bottlenecked]

Ya got a 4670k loser right here. 

[done12many2]

9 minutes ago, DELTAprime said:

My build last January was my first build since the days before overclocking both on the CPU and GPU was dead simple (no K parts existed). I ended up with a 6600k that is stable at 4.6ghz with only a .05 overvolt on air (I can hit 4.7 stable but it requires more overvolting and heat than I like). Most 6600k's I can find online are only stable up to 4.4ghz and with a bigger overvolt. I'm guessing it was a silicon lottery winner. So just how common is it to get a good chip like this vs one that's average vs one that can't overclock for nuts?

 

The word "stable" literally means something completely different to every person on the internet.  It's so subjective that there's no way to accurately compare your overclock to anyone else's.

 

With the amount of these "look, my chip will do 4.8 GHz at 1.2v fully, unbelievably, super duper stable" threads popping up all the time now, it's obvious that a lot of people are full of shit.  

 

By the way, a tighter low leakage chip does not mean that it will overclock high.  Often times it means just the opposite.  Low voltage also doesn't mean cooler.

[Zyndo]

6 minutes ago, DELTAprime said:

Snip

the stats can a lot vary from one family of CPU's to the next, but with the sheer volume of CPU's sold, there are plenty of people out there with "good" ones and "bad" ones. But like @Oshino Shinobu said, without having the exact numbers there is no way to give you a %. You certainly have an outlier though. congrats.

[Bananasplit_00]

i lost badly with my 970, and i got decently lucky with my I5 4690K, its imposible to say without testing an enormus amount of chips but i think Userbenchmark has an avrage overclock of all tested chips and some other statistics, thats probably as close as you will get

[cakevreter2000]

i got a pretty averige (i think) 4790k here with 4.7 at 80 celcius (yes its hot but i just wantet to get there. i run it on stock now becouse its mooooore then anough and i will overclock it once necesery)

[DELTAprime]

Ok, thanks guys. Hopefully Intel get better at making winning chips cause I'd hate to have my next one OC worse.

[SpaceGhostC2C]

I would bet there are more users who don't overclock than those who do, even among K users. And among overclockers, not all of them find the exact limits, or the voltage-frequency curve if you want, of their CPUs. Many achieve what feels like a sweet spot to them and call it a day. So, if you would build a database of all overclocks for a given chip, it would a small fraction of all chips, and an incomplete sample at that. But if you really had them all, at least it would be a random and large sample (after all, overclockers can't choose to buy winners or losers, unless they buy previously tested chips, but those testing them couldn't know before buying).

But if you search on the internet, you are not going to get a random sample, but a selected one: people sharing their experiences may be more often winners than losers, with a share of extreme losers who seek for help on their lack of success at OC. Probably a lot of "meh"s and "good enough for me"s are going to go under the radar.

In a perfectly competitive market, the price difference between K and non K versions, and other locked chips of the same line would be informative of the probability distribution, or at least of the information Intel has on said distribution. Sadly, we don't have such market   

[Guest]

OC skill/knowledge disparity is more prevalent than silicon.. 

 

You'd have to do sciendtific testing of hundreds of chips using the same variables/parameters to come to such a solution. 

 

I wouldn't consider you a winner either, that's just barely above average? Winners are pushing 5GHz with reasonable voltage. 

[Carclis]

4.4 on a 6600k doesn't seem that high and going by Silicon Lottery stats it isn't with 65% being able to hit 4.7Ghz.

 

Don't feel bad though. I seem to be a frequent loser when it comes to the lottery as my 6700k is in top 96% and my R9 fury doesn't overclock at all. Next time will be different though 

[iRileyx]

Its too hit and miss to measure to be honest.

My a8-7650k which is stock 3.3, turbo 3.8, can run at 4.5Ghz at only 1.35 volts(quite low for an amd chip) and it runs fine.
My 480 on the other hand, can run at 1400mhz but will crash after 2 or so hours, i have to turn it down to 1350 if i want it to run properly for a long gaming session (about 8+ hours with no crashing, haven't tested for longer)

[DELTAprime]

Just ran my system through Userbench and at 4.6ghz my 6600k is in the 98th percentile. So that means out of people that have downloaded and ran their tool there are only 647 out of 32,385 6600k's that good or better. Well I'm lowering my expectations for my 7700k build's chances of getting a winner. lol.

 

http://www.userbenchmark.com/UserRun/2234265

[done12many2]

1 minute ago, DELTAprime said:

Just ran my system through Userbench and at 4.6ghz my 6600k is in the 98th percentile. So that means out of people that have downloaded and ran their tool there are only 647 out of 32,385 6600k's that good or better. Well I'm lowering my expectations for my 7700k build's chances of getting a winner. lol.

 

What's does it score in Cinebench R15?

[DELTAprime]

4 minutes ago, done12many2 said:

What's does it score in Cinebench R15?

746. Good?

[done12many2]

6 minutes ago, DELTAprime said:

746. Good?

 

That's pretty decent.  How do those chips do in the single-threaded R15 test?

[DELTAprime]

6 minutes ago, done12many2 said:

That's pretty decent.  How do those chips do in the single-threaded R15 test?

193 in single thread.

 

Anyway I'm off to bed now, it's after midnight. Thanks guys for the responses.