PSA: Don't buy x99 if you are unwilling to overclock, DUH

[Curufinwe_wins]

***Disclaimer*** This thread is for gaming/low thread utilization applications, although I would argue the same logic applies for workstation builders.

 

So this thread is really coming up in response to the new 5820k vs 6700k video placed on vessel recently. In it I was extremely disappointed to see them compare stock speeds only. 

 

DO NOT BUY X99 IF YOU ARE UNWILLING TO OVERCLOCK IT. It is idiocy at best and sheer wastefulness/downgrade at worst.

 

X99 cpu's are ridiculously under-clocked compared to their z170 and z97 counterparts BECAUSE INTEL KNOWS PEOPLE ON X99 SHOULD BE (and want headroom for) OVERCLOCKING.

 

Comparing expected overclocking results between the 5820k and the 4790k at a mild overvoltage of 1.3V shows (5820k 4.4-4.6, 4790k 4.7-4.8) ought to remind you just how much overclocking head room there is when the stock 5820k only turbo's to 3.6/3.4 compared to the 4.4/4.2 (depending on motherboard) of the 4790k. This is also true with the z170 chipset who admittedly runs with and handles a slightly higher voltage but still only sees overclocks in the realm of 4.6-4.8 from a base 4.2.

 

Sure a 5820k at 3.6 trades blows with a 6700k at 4.2 (largely dependent on cpu scalabilty of specific benchmark), but even a super light 5820k overclock to 4.0 Ghz will basically bring the 5820k up from trading blows to better or tied in every test (pure single or double threaded loads excepted), and comparing a 5820k at say 4.5 and a 6700k at 4.7 is a win all day for the 5820k (again pure single or double threaded loads excepted).

 

TL:DR If you buy x99 but are unwilling to actually utilize its potential then you are better served not using the Enthusiast chipset at all.

 

 

 

PS: "low-end" x99 motherboard is basically always higher quality than a mid-range z170/z97 motherboard, with comparisons between the x99 sli plus and the z170 gaming m5 being a complete joke, and I personally think the two motherboards being used to compare the x99 purchase price were idiotic at best.

 

PSS: I don't know if anyone still has the links to that german site that does the fantastic cpu (with oc and non oc) comparisons  (witcher 3 is a good example), but if you do please link below and I'll throw them up.

[manikyath]

well... all i can say is..

 

"no shit, sherlock?"

[Chasem121]

Literally everyone already knows this, anyone will is willing to invest in that sort of platform would also be willing to invest in overclocking-ready cooling for it. You are preaching to the choir.

[manikyath]

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-6700K-vs-Intel-Core-i7-5820K/3502vs2579

 

i'll drop this in here for some actual valuable information.

[jurian123]

"Enthusiast chipset"

 

^ Says it all. 

[JuztBe]

Yeah, it's disappointing that they made a video on stock clocks. Basically it's irrelevant. Guys who are going with this kind of CPUs are OCing for sure. 

[Curufinwe_wins]

well... all i can say is..

 

"no shit, sherlock?"

I did include DUH, but if you have vessel the most recent showdown is extremely misleading in this fashion, and doesn't even bother to MENTION overclocking differences.

Literally everyone already knows this, anyone will is willing to invest in that sort of platform would also be willing to invest in overclocking-ready cooling for it. You are preaching to the choir.

See above, and overclocking-ready cooling doesn't mean they will actually overclock. You would be surprised the number of people that just buy x99 thinking its automatically better out of the box (and vise versa for z170 after overclocking).

"Enthusiast chipset"

 

^ Says it all. 

Indeed.

[Lays]

4.7 @ 1.3v-1.35v is more along the lines of average for 4790k and 4770k, nowhere near 4.9 @ 1.3v is average. Not sure where you saw that.

[Curufinwe_wins]

Yeah, it's disappointing that they made a video on stock clocks. Basically it's irrelevant. Guys who are going with this kind of CPUs are OCing for sure. 

I just couldn't believe they didn't even MENTION overclocking let alone using it. Hell they could have done extremely mild oc's on both just to give an idea what everyone and their brother could reach (which probably would have been around 4.0 for 5820k and 4.4 for 6700k).

[manikyath]

I did include DUH, but if you have vessel the most recent showdown is extremely misleading in this fashion, and doesn't even bother to MENTION overclocking differences.

See above, and overclocking-ready cooling doesn't mean they will actually overclock. You would be surprised the number of people that just buy x99 thinking its automatically better out of the box (and vise versa for z170 after overclocking).

Indeed.

refer to the comparison i sent, they seem to have some pretty valuable information.

 

you seem to forget one thing about overclocking: skylake overclocks as well, and isnt bad at doing so. so in the end the difference between them doesnt change much.

[Curufinwe_wins]

 

4.7 @ 1.3v is more along the lines of average for 4790k and 4770k, nowhere near 4.9 @ 1.3v is average. Not sure where you saw that.

I've not really seen a 4790k that can't hit 4.7 at 1.3 (yes the 4770k is slightly different, just as newer batches of 5820k's are basically a lock for 4.5 at 1.3V), I know 4.9 is a very lucky chip, but I was being generous to the 4790k (up until fairly recently reaching 4.6 at that voltage on a 5820k was also pretty golden).

 

 

refer to the comparison i sent, they seem to have some pretty valuable information.

 

you seem to forget one thing about overclocking: skylake overclocks as well, and isnt bad at doing so. so in the end the difference between them doesnt change much.

I mentioned it throughout my topic, but I would argue it changes quite a bit in the end (mainly because the 5820k is looking at nearly a 40% overclock compared to a 18% OC on the 6700k).

 

Yea I'll throw it up.

[Lethal Seraph]

"no shit, sherlock?"

 

We'll... that depends. Just like Apple user fanbase, some people just have more money than sense.  :lol:

[Lays]

I've not really seen a 4790k that can't hit 4.7 at 1.3 (yes the 4770k is slightly different, just as newer batches of 5820k's are basically a lock for 4.5 at 1.3V), I know 4.9 is a very lucky chip, but I was being generous to the 4790k (up until fairly recently reaching 4.6 at that voltage on a 5820k was also pretty golden).

 

All the chips I see capable of 4.7 are usually doing it at 1.28-1.35v from what I've seen, then occasionally I see one that hits it at a bit lower volts, but very rarely do I see 4.8 or 4.9 at less than 1.35-1.45v.

I have a buddy with a 5 ghz 1.335v chip I want to buy, but I am so lazy and I think I may just save my $ for Skylake for more hwbot points 

[Curufinwe_wins]

All the chips I see capable of 4.7 are usually doing it at 1.28-1.35v from what I've seen, then occasionally I see one that hits it at a bit lower volts, but very rarely do I see 4.8 or 4.9 at less than 1.35-1.45v.

I have a buddy with a 5 ghz 1.335v chip I want to buy, but I am so lazy and I think I may just save my $ for Skylake for more hwbot points 

I don't think we have too much of a disagreement there.

[Kinda Bottlenecked]

Heh. Tell that to my 4670k. 1.3v and it won't even want to stay stable at 4Ghz. 

[Curufinwe_wins]

Heh. Tell that to my 4670k. 1.3v and it won't even want to stay stable at 4Ghz. 

First things first, have you tried increasing Vin? Secondly, I5's due to the nature of intel's binning process should on average be worse and less consistent overclockers than the i7 line.

[Kinda Bottlenecked]

First things first, have you tried increasing Vin? Secondly, I5's due to the nature of intel's binning process should on average be worse and less consistent overclockers than the i7 line.

 

Yes yes, done everything that could've been done. This is the 4670k that won't overclock. A $100 wasted. 

[Curufinwe_wins]

well... all i can say is..

 

"no shit, sherlock?"

On an aside any clue on why when I try to run userbenchmark it just gets stuck in an infinite loop trying to do 4k mixed IO on my harddrive (I'd run it without that if I could, but this is already one of my least favorite benchmarks already, because I'd argue it's far too short...)

[manikyath]

On an aside any clue on why when I try to run userbenchmark it just gets stuck in an infinite loop trying to do 4k mixed IO on my harddrive (I'd run it without that if I could, but this is already one of my least favorite benchmarks already, because I'd argue it's far too short...)

i've actually never used the benchmark.

 

but from what i've seen from their results, they seem to be mostly accurate, and not biased towards anything. (lets just shun passmark here for the heck of it)

for me the fact they split up the performance differences in different categories (for cpus: single core, quad core and multi-core, both OC and non-OC) is very interesting.

 

they also provide a graph of the "popularity" of user submitted results, rather than (shunning passmark again, because why not) just slapping a number on it.

[Curufinwe_wins]

i've actually never used the benchmark.

 

but from what i've seen from their results, they seem to be mostly accurate, and not biased towards anything. (lets just shun passmark here for the heck of it)

for me the fact they split up the performance differences in different categories (for cpus: single core, quad core and multi-core, both OC and non-OC) is very interesting.

 

they also provide a graph of the "popularity" of user submitted results, rather than (shunning passmark again, because why not) just slapping a number on it.

My issue is that each of the tests is literally supposed to last less than 2 seconds, which makes me question the validity (or rather the variance, which I think is much larger for identical real-world preforming systems than for more robust tests).  (A good example of this is that I am fairly certain the cpu tests are not even long enough to reach max turbo for me, now I could lock turbo to get away from that but eww).

 

I like the distribution numbers ofc, but again I can't seem to get past a 4k mixed IO and I can't deselect it just to see what the userbenchmark says for me in the context of everything else.

 

EDIT: Hell with the 5820k single core even r15 isn't heavy enough to reach max turbo (I've checked, and with forced max speed my single threaded score is ~180+/-5, standard setup is 164+/-5.)

[Speaker1264]

They are clocked based on use of the stock heatsink in a hot/confined environment for maximum compatibility.  Of course if you are in a controlled climate environment with an aftermarket heatsink and optimal airflow you will be able to push them far beyond stock speeds.  This is just as true for the consumer grade CPU's.  Don't even know why this is worth mentioning.  It should be common sense by now.

[Str_Mape]

They are clocked based on use of the stock heatsink in hot/confined environment for maximum compatibility.  Of course if you are in a controlled climate environment with an aftermarket heatsink and optimal airflow you will be able to push them far beyond stock speeds.  This is just as true for the consumer grade CPU's.  Don't even know why this is worth mentioning.

Keep in mind that the X99 CPU's do not come with a heatsink in the box so everyone will use an aftermarket cooler, unless of course you get one of those gimicky 2011 coolers from Intel.

[SeanBond]

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-6700K-vs-Intel-Core-i7-5820K/3502vs2579

 

i'll drop this in here for some actual valuable information.

How is this useful? It's just a slightly better version of cpu boss. Any clown looking to get the more powerful chip should be getting the 5820k unless they really want to sit and forget and only focus on gaming. All this information says is go buy a 6700k because it's better. When it's really not. 

[Curufinwe_wins]

They are clocked based on use of the stock heatsink in a hot/confined environment for maximum compatibility.  Of course if you are in a controlled climate environment with an aftermarket heatsink and optimal airflow you will be able to push them far beyond stock speeds.  This is just as true for the consumer grade CPU's.  Don't even know why this is worth mentioning.  It should be common sense by now.

1. There is no stock heatsink for x99. So that entire argument is invalid.

2. I only mention it because the most recent Vessel video (as I explained) is completely misleading in its utter lack of any sort of discussion including the factors that actually matter in favor of x99. (Don't get me wrong, I don't think everyone should be using it, but I think the conclusions drawn from that video are nearly contrary to the truth due to some imho idiotic constraints.)

[manikyath]

How is this useful? It's just a slightly better version of cpu boss. Any clown looking to get the more powerful chip should be getting the 5820k unless they really want to sit and forget and only focus on gaming. All this information says is go buy a 6700k because it's better. When it's really not. 

then you REALLY didnt look very well. they have single core, quad core and multi-core benchmarks, both overclocked and stock clock.

 

while i agree that they're probably not 100% accurate, they're REALLY good to see the rough differences between parts.

 

as for cpu boss: they base themselves off other benchmarks, one of which is passmark, which is a HUGE red flag for me (once again, picking on passmark, because its fun)