What do you think about Intel coffee lake?

[SpaceGhostC2C]

 

On 8/8/2017 at 7:41 PM, Untitled said:

It's coming soon. On 21 Aug. Will it faster than Ryzen? or it's just another joke from Intel?

 

What do you think?

I think that the great news that Ryzen was sometimes makes us forget what the broad situation is. To begin with, how is it not going to be faster than Ryzen, when Intel's current lineup already is faster than Ryzen core vs core?

I see more of a contest for Threadripper, because the differences in approach for the HEDT do indeed make AMD outright faster in some scenarios, and how these two platforms will evolve over time is really interesting. But in the mainstream the situation is simpler: AMD improved a lot, kept prices down, and just by being not much slower (i.e., not faster) it ruined Intel from the value perspective. Hence, in the mainstream platform, there's nothing preventing Intel from releasing Kaby Lake based CPUs that de-throne Ryzen even in value terms, except for their own calculations on what maximizes their profits. Having the faster core means that it's up to them to come with a product that at the very least is not worse value, while being better in absolute performance terms, than Ryzen, perhaps with the exception of Ryzen 7 for multithreaded use cases if they can't fit 8 cores in 1151. In fact, they could just adjust the pricing of their current CPUs and that would be it. No new development needed.

 

However, Intel has shown at least in the very recent past that they prefer to keep prices for their existing products and do the price adjustment with a new release instead. Plus the difference in core count meant that the price adjustment needed would be too savage, leaving the higher end to AMD for a while. What Coffe Lake will allow them to do, even if there's no improvement in IPC, is to adjust their features and prices to the new, smaller performance margin over AMD. Kaby Lake vs Ryzen isn't the same as Skylake vs. Piledriver; the 2066 lineup already reflects that, and I expect Coffe Lake to reflect that as well.

 

On 8/8/2017 at 8:27 PM, TheCherryKing said:

Kaby Lake and Coffee Lake are a joke! Everything Intel has done this year is a train wreck! Coffee Lake is a superior architecture compared to AMD Ryzen but the pricing will be insane! 

On the contrary, I'm quite sure Coffee Lake's release will be used to realign Intel's offering in terms of performance per dollar. In fact, I suspect the increase in core count will be use to keep the price range more or less unchanged.

But crystal ball can fail  

 

19 minutes ago, done12many2 said:

 

6 cores weren't ready last year when Kaby Lake released.

 

Intel gave it to us and we've been using it since.  

 

 

And even before that, under the name of Skylake  

 

7 minutes ago, done12many2 said:

 

I am saying that it is better to release the tech that is available then to delay it all so that you can include more with more features a year later.  

7 minutes ago, done12many2 said:

 

We have been enjoying the 7700k for the last year seven months.  I would rather had that than to have them not have released it until this year.  

 

But there has to be new tech for that... what is it exactly that they released? A slightly higher overclock for the top 1% chips? I mean, you can disagree, but you have to understand that the main argument behind the "Kaby Lake could have been skipped" claim is precisely that it wasn't really making anything available compared to Skylake, other than artificially baked in DRM, etc.

I think that, at the end of the day, the discussion depends on whether you read that as equivalent to "AMD shouldn't have released Ryzen until they could add features X, Y, Z, just run with Excavator in the meantime", or closer to "AMD should have skipped the FX-9xxx series, and just roll with Piledriver until Ryzen". I'm more inclined to the latter, hence why I tend to agree with @Princess Cadence; if you see it more like the former, you will tend to disagree.

 

[dizmo]

59 minutes ago, Princess Cadence said:

Exactly! The reason why I'll go with the 8700k instead of going locked i7 as always, that base clock is a joke

I can't go with the k SKU in the case I want, so I'm stuck with something else. The leaks did show a 6 core 4.3ghz boost mode though! Which as me interested. If that's the case and it can remain cool then it's a total win. \

I have a feeling Intel will make a solid comeback.

[tom_w141]

1 hour ago, done12many2 said:

 

6 cores weren't ready last year when Kaby Lake released.

 

Intel gave it to us and we've been using it since.  

 

Not sure that they should have delayed it to meet your expectations.  

IK its a grey earlier but personally i'd rather them have not released kaby lake. Would have kept Skylake relevant for longer and wouldn't have created a <1 year product cycle (thats good for no one and you just feel unrewarded for your purchase). 8 months between release and obsolescence is just too short imo. Whereas on the other hand there are some people who would prefer an optimised skylake just to squeeze more so I get both sides. Just saying that personally if I was a kaby lake owner i'd feel a bit cheated that my platform died within a year. Its not even like the motherboard is still useful either. Both products effectively discontinued/obsolete in < 1 year

[done12many2]

29 minutes ago, SpaceGhostC2C said:

But there has to be new tech for that... what is it exactly that they released? A slightly higher overclock for the top 1% chips?

 

To be honest, you're exaggerating to make a point and I get it, but that's a very gross exaggeration.  Kaby Lake is on average faster than Skylake and it's not just the top 1% of the chips.

 

The average low Kaby Lake overclock is the average high Skylake overclock.  

 

With Kaby Lake, 5 GHz (with usable IPC) became a thing.  Not everyone was able to achieve that, but it was far more common than it ever was with Skylake.  

 

Keep in mind when comparing stats between the Kaby Lake statics thread and the Skylake statistics thread that @Dark_wizzie made the stress testing requirements quite a bit harder for Kaby Lake than they were for Skylake, yet the average overclocks remain much higher.  This is to say that not only do they clock higher, but they are more usable with regards to stability at those clocks.  

 

Here on LTT we tend to just make shit up.  On OCN they try to prove things either way, which is why I frequent those threads and use them as references.  

 

OCN Kaby Lake statistics thread:  Avg OC is 5.04 GHz

 

http://www.overclock.net/t/1621347/kaby-lake-overclocking-guide-with-statistics/0_100

 

OCN Skylake statistics thread:  Avg OC is 4.68 GHz

 

http://www.overclock.net/t/1570313/skylake-overclocking-guide-with-statistics/0_100

 

I won't argue if that was worth it as that is completely subjective, but to exaggerate facts to make points with numbers like 1%, I don't agree with.  

 

11 minutes ago, tom_w141 said:

IK its a grey earlier but personally i'd rather them have not released kaby lake. Would have kept Skylake relevant for longer and wouldn't have created a <1 year product cycle (thats good for no one and you just feel unrewarded for your purchase). 8 months between release and obsolescence is just too short imo. Whereas on the other hand there are some people who would prefer an optimised skylake just to squeeze more so I get both sides. Just saying that personally if I was a kaby lake owner i'd feel a bit cheated that my platform died within a year. Its not even like the motherboard is still useful either. Both products effectively discontinued/obsolete in < 1 year

 

I have a 7700k and a x99 platform, both of which have been or are about to be replaced by Intel.  I in no way feel cheated because I understand how things work.  

 

It was my choice to either continue using a 6700k / z170 setup or make the move to the 7700k / z270 setup and I'm glad that I did.  Speedshift, which is a feature that very few people talk about has been a very nice thing to have.  It was also a feature of the 6700k, but it didn't work correctly.  We're at a limit with clocks on Skylake.  We needed 14+ to keep pushing.

 

The additional clocks and a few more things were more than enough for me to justify the switch.  They were also obviously enough for a great deal of people to justify the switch or upgrade to as it was and remains the number #1 selling chip out there.  

 

This probably doesn't make sense from a value standpoint, which is why I understand your points as well, but it was much more than a stop gap and many 7700k owners would argue that the money spent was worth it to them. 

[Jumper118]

1 hour ago, done12many2 said:

 

To be honest, you're exaggerating to make a point and I get it, but that's a very gross exaggeration.  Kaby Lake is on average faster than Skylake and it's not just the top 1% of the chips.

 

The average low Kaby Lake overclock is the average high Skylake overclock.  

 

With Kaby Lake, 5 GHz (with usable IPC) became a thing.  Not everyone was able to achieve that, but it was far more common than it ever was with Skylake.  

 

Keep in mind when comparing stats between the Kaby Lake statics thread and the Skylake statistics thread that @Dark_wizzie made the stress testing requirements quite a bit harder for Kaby Lake than they were for Skylake, yet the average overclocks remain much higher.  This is to say that not only do they clock higher, but they are more usable with regards to stability at those clocks.  

 

Here on LTT we tend to just make shit up.  On OCN they try to prove things either way, which is why I frequent those threads and use them as references.  

 

OCN Kaby Lake statistics thread:  Avg OC is 5.04 GHz

 

http://www.overclock.net/t/1621347/kaby-lake-overclocking-guide-with-statistics/0_100

 

OCN Skylake statistics thread:  Avg OC is 4.68 GHz

 

http://www.overclock.net/t/1570313/skylake-overclocking-guide-with-statistics/0_100

 

I won't argue if that was worth it as that is completely subjective, but to exaggerate facts to make points with numbers like 1%, I don't agree with.  

 

 

I have a 7700k and a x99 platform, both of which have been or are about to be replaced by Intel.  I in no way feel cheated because I understand how things work.  

 

It was my choice to either continue using a 6700k / z170 setup or make the move to the 7700k / z270 setup and I'm glad that I did.  Speedshift, which is a feature that very few people talk about has been a very nice thing to have.  It was also a feature of the 6700k, but it didn't work correctly.  We're at a limit with clocks on Skylake.  We needed 14+ to keep pushing.

 

The additional clocks and a few more things were more than enough for me to justify the switch.  They were also obviously enough for a great deal of people to justify the switch or upgrade to as it was and remains the number #1 selling chip out there.  

 

This probably doesn't make sense from a value standpoint, which is why I understand your points as well, but it was much more than a stop gap and many 7700k owners would argue that the money spent was worth it to them. 

ur rich and know what intel are like, people who saved up for years and upgraded to 7700K just for a miles better cpu with 50% more cores come out a few months later are the people who just got rekt  

[done12many2]

5 minutes ago, Jumper118 said:

ur rich and know what intel are like, people who saved up for years and upgraded to 7700K just for a miles better cpu with 50% more cores come out a few months later are the people who just got rekt  

 

Of course, those would be the same people who should have expected a platform change with the next chip release, right?  Is Intel going to another chipset/socket remap anything new after 2 gens on one platform?

[Jumper118]

7 minutes ago, done12many2 said:

 

Of course, those would be the same people who should have expected a platform change with the next chip release, right?  Is Intel going to another chipset/socket remap anything new after 2 gens on one platform?

they didnt know  

[done12many2]

Just now, Jumper118 said:

they didnt know  

 

I can't help it if people meander through the impact area of life and get blown up.  A little Darwin in action.  

 

Even if they are oblivious to what's going on, they will still have one of the best gaming chips for a good while to come.  It's not like it just stops working or performing great the day that Coffee Lake releases.  Hell, we probably won't be regularly using the extra cores from Coffee Lake for a good while to come.  Until that happens, gaming with a 7700k or a 8700k will remain about the same.  Not the disaster that you guys are trying to paint.  

[TheCherryKing]

It's like Intel never stopped Tick-Tock with the exception that the die size wasn't shrunken for Kaby Lake or Coffee Lake. I thought that the motherboards that were designed for Skylake would be compatible with CPU's for the following two generations because of Intel's new Process, Architecture, Optimization release cycle. This does not seem to be the case. Since the motherboards are compatible with only two generations of CPU's it seems that Intel is still using Tick-Tock.

 

100 Series: Skylake(TOCK) 14nm.

200 Series: Kaby Lake(TICK) 14nm.

 

300 Series: Coffee Lake(TOCK) 14 nm.

400 Series: Cannonlake(TICK) 10 nm.

[SpaceGhostC2C]

2 hours ago, done12many2 said:

 

To be honest, you're exaggerating to make a point and I get it, but that's a very gross exaggeration.  

You got the point, it was obviously an arbitrary random number to make a qualitative, not quantitative point. The rest of this post is therefore besides the point, but just for fun  

2 hours ago, done12many2 said:

Kaby Lake is on average faster than Skylake and it's not just the top 1% of the chips.

 

The average low Kaby Lake overclock is the average high Skylake overclock.  

That is not informative about the distribution, though. A higher average could be driven by the top 1% alone. Or by a single outlier. Or by every part of the distribution being above. So, it's good to have actual numbers, and we would need much higher sample sizes than the one in the link to confidently estimate the 99th percentile, but we can work with what we have if we keep that in mind. But if we were to check the validity of the "top 1%" claim (if there was one, which isn't), we'll need more than to compare the magnitude in the difference in means. I may check that out later, although I don't know if posting the results here would be too much derailing. 

 

2 hours ago, done12many2 said:

I won't argue if that was worth it as that is completely subjective, but to exaggerate facts to make points with numbers like 1%, I don't agree with.  

It is not an exaggeration, it is a figure of speech (an hyperbole). An exaggeration would be if I was trying to make a quantitative statement. Same as saying "no one bought XYZ" is not an exaggeration about the actual number of sales, but a way to express "sales were low". You started the post like you had understood it 

 

[done12many2]

2 hours ago, SpaceGhostC2C said:

But there has to be new tech for that... what is it exactly that they released? A slightly higher overclock for the top 1% chips?

 

7 minutes ago, SpaceGhostC2C said:

You got the point, it was obviously an arbitrary random number to make a qualitative, not quantitative point. The rest of this post is therefore besides the point, but just for fun  

That is not informative about the distribution, though. A higher average could be driven by the top 1% alone. Or by a single outlier. Or by every part of the distribution being above. So, it's good to have actual numbers, and we would need much higher sample sizes than the one in the link to confidently estimate the 99th percentile, but we can work with what we have if we keep that in mind. But if we were to check the validity of the "top 1%" claim (if there was one, which isn't), we'll need more than to compare the magnitude in the difference in means. I may check that out later, although I don't know if posting the results here would be too much derailing. 

 

It is not an exaggeration, it is a figure of speech (an hyperbole). An exaggeration would be if I was trying to make a quantitative statement. Same as saying "no one bought XYZ" is not an exaggeration about the actual number of sales, but a way to express "sales were low". You started the post like you had understood it 

 

 

Actually, anything you said after your initial statement is kinda besides the point.  Your follow up is akin to saying that despite the fact that we are seeing a great deal of higher overclocks out of Kaby Lake when compared to Skylake, we can't say that it's true for the simple fact that we haven't tested every one of them.  Something you know can't happen.

 

Very safe response to an initial statement that contradicts what we know to be true.   

[SpaceGhostC2C]

3 minutes ago, done12many2 said:

 

 

Actually, anything you said after your initial statement is kinda besides the point.  Your follow up is akin to saying that despite the fact that we are seeing a great deal of higher overclocks out of Kaby Lake when compared to Skylake, we can't say that it's true for the simple fact that we haven't tested every one of them.  Something you know can't happen.

 

Very safe response to an initial statement that contradicts what we know to be true.   

No, actually that is a very poor understanding of what I wrote. It has nothing to do with "not having tested every one of them" (what happened with zero tolerance to exaggeration? ).

What I said is basic statistics: information about sample averages is not a way to test whether the difference is at the top of the distribution, the bottom, the middle, or everywhere, but there are ways to test this for a given sample. I even mentioned I would do it. So, no, it has nothing to do with "we need the whole population". I did mention the desirability of a larger sample, simply because the N needed to have a good estimate of the mean isn't the same as the N needed to estimate a top percentile.

By the way, there is an additional problem with the sample, which is its non-random nature: people self-select into submitting results. That is a much more important concern than the size, since we could have the best overclockers at OCN, and therefore not even the 1%, but the top 0.1% choosing to show up and report their nice results. Therefore, it would still no answer the question of how things are at the middle.

Still, I would be willing to assume that these guys are representative of the CPU distribution in general because they haven't tried that many CPUs each, and therefore even if they show a special interest in overclocking, the CPUs they got are not correlated with it. Strong assumption, but not making it would ruin the fun  

 

[NvidiaIntelAMDLoveTriangle]

I think it's not out yet.

[done12many2]

23 minutes ago, SpaceGhostC2C said:

No, actually that is a very poor understanding of what I wrote. It has nothing to do with "not having tested every one of them" (what happened with zero tolerance to exaggeration? ).

What I said is basic statistics: information about sample averages is not a way to test whether the difference is at the top of the distribution, the bottom, the middle, or everywhere, but there are ways to test this for a given sample. I even mentioned I would do it. So, no, it has nothing to do with "we need the whole population". I did mention the desirability of a larger sample, simply because the N needed to have a good estimate of the mean isn't the same as the N needed to estimate a top percentile.

By the way, there is an additional problem with the sample, which is its non-random nature: people self-select into submitting results. That is a much more important concern than the size, since we could have the best overclockers at OCN, and therefore not even the 1%, but the top 0.1% choosing to show up and report their nice results. Therefore, it would still no answer the question of how things are at the middle.

Still, I would be willing to assume that these guys are representative of the CPU distribution in general because they haven't tried that many CPUs each, and therefore even if they show a special interest in overclocking, the CPUs they got are not correlated with it. Strong assumption, but not making it would ruin the fun  

 

 

By the way, how were you able to back your claim that Kaby Lake was only a slightly faster OC if neither of us can prove either way?  Is this one of those deals where you get to say what you want, but the burden of proof lies with me if I say otherwise?

 

I've personally been through dozens and dozens of each chip.  The 7700k is absolutely faster than a 6700k and it's not by a little.  My worst 7700ks would beat out the best of the 6700ks.  

 

Anyways, here's some data from HWBOT in addition to the information I provided earlier from OCN.  I'll just have to agree that we may never know the truth.  

 

6700k vs 7700k air = 279 MHz higher on average for submissions

 

6700k vs 7700k water = 270 MHz higher on average for submissions

 

6700k vs 7700k LN2 = 505 MHz higher on average for submissions

 

The LN2 averages are a great way to see that the silicon speed of Kaby Lake is just flat out faster than that of Skylake by a good amount. 

 

 

 

 

 

 

 

 

 

[SpaceGhostC2C]

1 minute ago, done12many2 said:

 

By the way, how were you able to back your claim that Kaby Lake was only a slightly faster OC if neither of us can prove either way?  Is this one of those deals where you get to say what you want, but the burden of proof lies with me if I say otherwise?

Have you stopped reading altogether? Fine by me, but in that case you may as well stop quoting as well...

[done12many2]

Just now, SpaceGhostC2C said:

Have you stopped reading altogether? Fine by me, but in that case you may as well stop quoting as well...

 

This is priceless.  You do a lot of fancy talking to skirt around things.  

 

There are those that do and those that talk about doing.  

[SpaceGhostC2C]

2 minutes ago, done12many2 said:

 

This is priceless.  You do a lot of fancy talking to skirt around things.  

 

There are those that do and those that talk about doing.  

Unbelievable. You've been arguing with your own shadow for several posts, pretending somehow what I wrote actually meant the opposite, and exactly what you felt like reading and arguing against.

To each its own, I guess. Have fun with it.

 

[done12many2]

3 hours ago, SpaceGhostC2C said:

But there has to be new tech for that... what is it exactly that they released? A slightly higher overclock for the top 1% chips?

15 minutes ago, SpaceGhostC2C said:

Unbelievable. You've been arguing with your own shadow for several posts, pretending somehow what I wrote actually meant the opposite, and exactly what you felt like reading and arguing against.

To each its own, I guess. Have fun with it.

 

 

 

What's unbelievable is that you seems like an intelligent guy, yet you've fallen in lines with the latest LTT trend of posting unsupported claims and hoping for the best.  It's that "let me spit ball this comment and see if I can get away with it" mentality.  Then posting the meme and I have all I need to understand.

 

You made a statement casually implying that Kaby Lake was only slightly faster than Skylake, which is wrong.  You see doing so further supported the rest of your original post.   Everything from that point on was me showing you examples and you giving classes on shit that had nothing to do with supporting your claim.  Spitball and hope it sticks, right?

 

[BoredErica]

It's true that self reported statistics for overclocks are by definition not a totally random sample. But I'm doing all of my work for free, I can't afford to pry overclocks from strangers and get them to run an 8-12 hour stress test, and it's honestly good enough. The person viewing my threads are also a non-random sample; they are people who care enough to look in there in the first place. On a surface level, OCN thread of OCN users for OCN thread viewer makes sense. There are people who post who have bad overclocks and post the results on purpose to try to lower the averages to make the statistic as sane and realistic as possible (then half the time the person turns around and shits on my chart and blames the statistics for being inflated instead of simply accepting they have a bad chip, but trust me, those people want to be heard and will be willing to post their overclock to make it heard). Having read most of the many posts around my threads I can tell you a lot of people have no idea what they are doing, and they didn't even read my OP which is frustrating. I honestly believe the mix of people in my charts is a nice mix of different types of people. If you look at the chart you can see people's cooling solutions all over the place for example.

 

Short of not setting ring bus to a dumb frequency on accident, overclocking better should not net 200mhz improvement or more. (Arguably, if we're just concerned about the ability of the chips instead of the overclockers, we would be looking at the best overclocks to see what they get because there we feel more confident that the users aren't doing anything stupid when they overclock. Take away user error and we are left with what the chip is truly capable of. And if you think Kaby Lake has this weird relationship where the average is basically the same as Skylake but the top chips are quite a bit better, I'm not sure what else to tell you, except that's not how these things work.) Actually, the ring bus error really only affects Kaby Lake since Skylake could do 1:1 core to ring, which if it means anything in this case just means the statistics are less lopsided than they really should be in terms of what the chips are capable of. The ability of the chip is independent of the user, and if we grant that some users might mess up the ring bus and get a lower clock than they otherwise could, that actually pushes up the true average clocks of Kaby Lake up instead of down.

 

I don't understand why people care about IPC. I care about performance, not IPC. I don't care how that performance arrives - with some more and more refined process that clocks higher, or achieving more with the same clocks. It doesn't matter if it's new tech or old tech. How long do I think it will be until the next gen, and what performance improvement from that can I see? Saying whether Kaby Lake was really Skylake, or whatever Intel's intentions were is spin. The facts are the facts: The facts are Kaby Lake is a little bit faster than Skylake when overclocked. You either like it or you don't. Don't buy it if you don't. Buy whatever fits your use case.

Anybody that got Skylake should have forseen something like Kaby Lake, because it was the status quo for a long time. What was harder to foresee was the impact Ryzen had and whatever reaction Intel will now have to that. Even today we struggle to see just what that reaction is. There the situation changes a bit. If Skylake gave me 30fps, Kaby Lake gave me 40fps, and out of nowhere Coffee Lake gave me 60fps, while I do applaud the large gains, it may have to change my plans because this wasn't what I signed up for.

On the other hand, the more single threaded performance can be pushed upwards the better.

I can't speak for DoneTooMany, but I can speak for myself. I have been hearing for half a decade about how people like me who need the fastest single thread performance don't exist. Then on the other side we're getting the short end of the stick from Intel with these slow incremental improvements. Then now maybe we are given the option to upgrade twice in a year or upgrade at the same cadence as before but with greater improvements. I see that as a net positive. To my first point, I understand why people still want to talk about Kaby Lake and how it wasn't a huge improvement. But we all saw that coming. It is what it is. It gets to be a bit annoying after a while. If you don't think it's a good value, just list the pros and your small list will do the talking for you. But if you add bad statistics based from nowhere to your description then some people get upset.

 

Finally to be a little more on topic: I think the extra cores of Coffee Lake will make it quite a bit more competitive with Ryzen than in the past. My guess is (without bothering to pull out a calculator) Ryzen will still win in multithreaded but the case will be less compelling. But I don't really care that much. What I care about is whether Coffee Lake will have higher single thread performance compared to Kaby Lake despite having more cores. If I can have more cores and the usual improvements to single thread performance I will be very happy to upgrade. Time will tell. I want to upgrade to an Apex so I can get some Gskills DDR4 4400 ram running, but the stars have not aligned thus far.

[MyName13]

It wouldn't make sense to not sell 4c8t CPU since it was their flagship since the bulldozer, so does this mean that Pentium will be the old i5, i3 = i7 and everything above hexa cores?This would make almost the entire ryzen lineup obsolete, but I'm afraid that only i7 will have six cores and everything will remain the same.

[VioletGiraffe]

Coffe Lake is 6-core Kaby Lake, right? No architectural improvements should be expected?

[Guest]

On 8/8/2017 at 1:41 PM, Untitled said:

It's coming soon. On 21 Aug. Will it faster than Ryzen? or it's just another joke from Intel?

 

What do you think?

I personally think as a Ryzen owner 6 core coffee lake is going to cause Amd to drop prices 

[MyName13]

18 minutes ago, VioletGiraffe said:

Coffe Lake is 6-core Kaby Lake, right? No architectural improvements should be expected?

We still don't know much.

 

6 minutes ago, jdwii said:

I personally think as a Ryzen owner 6 core coffee lake is going to cause Amd to drop prices 

If i3 gets 2 cores then they will really have to drop 1400 and 1500x to ryzen 3 and 1600 to 1400 prices, if this happens I wouldn't be surprised if hexa cores become the new minimum in 3 years.

[DarkSmith2]

from:

7700k 4/8T with 4,2GHz BaseClock, 4,5GHz TurboClock - 91W TDP

 

to:

8700k 6C/12T with 3,7GHz BaseClock, 4,3GHz TurboClock - 95W TDP

(with 4,7GHz Single Core)

 

It looks like a damn nice improvement in stock performance to me. Well most gamers i know dont overclock their chips for several reasons, but most gamers i know also arent deep into tech stuff from begin with, so they probably wouldnt even invest the time trying to learn how overclocking is done. They do buy the K-Series CPUs because they come with higher stock clocks. (Basically poeple that would buy a r7 1700x or 1800x instead of a 1700 because they think its much faster)

 

For the others "improvement" relies on OC. How far will it go? How much power will it draw? How much heat will it produce? I personally wouldnt expect any major IPC improvements... Having two more cores at nearly the same TDP is already damn impressive. (IF the IPC will stay at the same level)

 

[atavax]

On 8/14/2017 at 0:00 PM, Noirgheos said:

All you have to do is take a look at a recent 7900X review to know that's not true. It pulls ahead of the 7700K in most games, and it doesn't have a clockspeed advantage either. 

 

Intel won't suddenly price mainstream models at HEDT. It makes no sense for a new generation and release. The 8700K will likely be $350 USD, like every other mainstream i7 before it.

 

where are you seeing benchmarks that shows the 7900X beat the 7700K in most games? I got the opposite impressions, ashes of the singularity and hitman being the only games where the 7900X seems to beat the 7700K http://www.gamersnexus.net/hwreviews/2961-intel-i9-7900x-review-benchmarks-game-streaming-vr-premiere/page-6

also

https://arstechnica.com/gadgets/2017/07/intel-core-i9-fastest-chip-but-too-darn-expensive/3/