Rumor Mill: Intel Core-i9 CPUs to Appear Soon - Lots of Cores

So I guess your ignoring the AMD 16 core 32 thread that is coming with the nickname thread ripper? More cores only plays to AMD's advantage as their process is a low power design and keep the speed up without going nuclear. Should be a interesting time in CPU's, but no one is destroying the other.
And add that in a battle of high core counts, SMT/HT becomes another focus and AMD has quite the advantage enough there to bridge a slight IPC deficit, hence its performance against a 4.3Ghz 6900k.
 
Here's hoping for ECC support...

693.gif
 
  • Like
Reactions: N4CR
like this
I can see it now....

i9 7920X (12c/24t) = $2,449.99
i9 7900X (10c/20t) = $1,649.99
i9 7820X (8c/16t) = $1,049.99
i9 7800X (6c/12t) = $549.99

:banghead:

Intel have always been greedy when it comes to CPU prices, whether you like them or not, AMD have done us all a favor by releasing Ryzen. It will help to keep Intel honest with their pricing.
 
Intel could erase Ryzen if they priced these right. Ryzen can't stand up to kabby lake core for core, but you can't get kabby in more than 4 cores, only the older broadwell. An 8 core kabby lake will demolish Ryzen. Intel most likely will price these in the stratosphere though making it moot.
 
Intel could erase Ryzen if they priced these right. Ryzen can't stand up to kabby lake core for core, but you can't get kabby in more than 4 cores, only the older broadwell. An 8 core kabby lake will demolish Ryzen. Intel most likely will price these in the stratosphere though making it moot.


Only if they get their termals in check and out clock the Ryzen. the per cycle IPC isn't great enough to make up for a slower clock yet. And from what we have seen so far they are not doing it.
 
  • Like
Reactions: fs123
like this
Only if they get their termals in check and out clock the Ryzen. the per cycle IPC isn't great enough to make up for a slower clock yet. And from what we have seen so far they are not doing it.

I don't follow your logic. Last I looked an 6950x OC to ~4.2 to 4.3. That is 10 cores... update to the next process I would expect to hit ~4.4 and at least this report shows turbo 3 hitting 4.5. One is happy to get an 8 core Ryzen get over 4 ghz. So what are you going on about? Sure these wont hit 5 ghz, but no >4 core cpu does.
 
They are planning to release the i7-7740K which has the exact same specs as the 7700K, except its TDP and that it is for LGA-2066 with X299.
So, I expect X299 to make the difference. The number of lanes will certainly do at the very least. I wonder what an X299 board will cost...

I'm still on Sandy Bridge. I can wait a month for when the NDA lifts and we will get to know more about features. Based on those we will probably be able to tell if we can expect a big rise in performance or not from these.
I'm also curious about the general fan noise, because of the higher TDPs.
 
This is the textbook example of an anticompetitive business practice. You can't abuse your market position to drive competitors out of business.
But you can, can't you? That is exactly what a free market means. And there are plenty of people out there advocating for a completely free market, without any government interference
 
It's the combination of clock speed and IPC that matters. Either in isolation of the other is completely meaningless.

Core count? Count me out. Unless you render/encode there is no point to more than ~4 to 6 cores.

All it does is result in more expensive, hotter designs without any real benefits, and sometimes drawbacks as the lower clocked many core parts perform worse in most tasks.

Only true because we devs don't bother trying to take advantage of multi-threading. It's hard making sure things like race conditions don't blow the entire thing up.

No, I don't agree with the sentiment (the power is there, USE IT), but there it is.
 
(For those buy for the future people....why not buy for the present, and when the future does happen, buy for the present again at that point?)

Because you have to buy everything again when the socket(s) change for minor performance gain.

I just posted the same in the AMD rumor thread. Any idea why Intel is giving these chips only 1.375MB of L3 cache per core? Sandy and Ivy had 2-2.5 while Broadwell and Haswell had 2.5 per core. Nearly halving the L3 seems odd. Cost savings? Heat dissipation issues? Considering the die size, I don't get all the empty space.

Intel hasn't released chips with much in the way of IPC gains for several years now.

If these are cut Xeons then it may have much more stuff for certain companies that we can't access, hence the unapparent 'inefficiency' of area.
 
The cache structure in SKL-X/SKL-SP is vastly different to SKL/KBL.

L2 have increased from 256KB to 1MB. And the L3 while smaller, is a lot faster.

Core IPC is still the same besides AVX512. However those cache changes will make quite the difference.
 
7900X is looking to be my next upgrade if the price is right.
 
No way Intel would have ever offered these without AMD releasing Ryzen. I seriously doubt any of these will be affordable. They really should take a different approach, modernize their whole stack. Nobody should have to be buying a dual core anymore.
i3 --> 4 cores, 8 threads
i5 --> 6 cores, 12 threads
i7 --> 8 cores, 16 threads
i9 --> 10+cores, 20+ threads
 
If it drops prices on most, if not all of the rest of lineup, cool. If not, meh, whatever.

Extreme line CPUs have never effected the prices of main stream CPUs, don't hold your breath.

"could compete with AMD's Ryzen"

Look I'm no Intel fanboy, but Intel is going to pulverize Ryzen. Undoubtedly. You guys all know this.

The CPU it self walk all over the current AMD lineup? Sure, but most people price matters as well, you have to consider bang for the buck, which the extreme line has never been, it's been the flagship premium line. So if these see a large drop in price to a similar Xeon CPU, sure, however the Extreme line has always been for those people who don't care about bang for buck and just want the absolute best.

No way Intel would have ever offered these without AMD releasing Ryzen. I seriously doubt any of these will be affordable. They really should take a different approach, modernize their whole stack. Nobody should have to be buying a dual core anymore.
i3 --> 4 cores, 8 threads
i5 --> 6 cores, 12 threads
i7 --> 8 cores, 16 threads
i9 --> 10+cores, 20+ threads

What? Intel has released an Extreme line every generation, so why exactly would Intel not have released these without AMD, when everyone was already expecting these? That reality distortion field is strong with you.
 
i feel very close to finally upgrading my main rig. not sure if it will be a 16c32t naples at 4GHz or a 12c24t i9 at 4.5Ghz, but clearly this Itel move is more related to Naples than Ryzen.
 
I really was planning on a Ryzen rig but if Intel prices the 7900X at a $1000 or less price point that's what I will most likely get. Shame Ryzen had issues with memory speed when all memory slots were full or I would probably already have built a Ryzen rig. Now I am waiting to see how the 7900X performs.
 
And here I am considering finally moving on from my 2600k... LOL and yeah, bang for the buck matters. Whatever I get will be in the $300-$500 range as I just can't really justify any more. This chip cost me $250 and frankly, even allowing for inflation, I would just like something that would last as long and be as great as this one has been, and maybe for as long. I overclock it, but I also consider it overclocking my wallet. AMD or Intel doesn't matter to me either, although motherboard stability DOES. I'm considering myself the market these guys need to win, and the possibility of Intel releasing something that may be competitive to AMD's current lineup is something that I am in the "wait and see" stage as far as prices go. I was REALLY considering AMD this time, and was about to begin researching, but now I may wait just a little longer.

Note, even the 1080 video cards have dropped a lot it seems in the past few months. It'll be interesting to see what AMD responds with... and yes, this competition is great for us, maybe not so great for AMD. I haven't forgotten that AMD hosted the Hardocp party I went to in Dallas and I would love to see more of them. They do get points from me for that.
 
And here I am considering finally moving on from my 2600k... LOL and yeah, bang for the buck matters. Whatever I get will be in the $300-$500 range as I just can't really justify any more. This chip cost me $250 and frankly, even allowing for inflation, I would just like something that would last as long and be as great as this one has been, and maybe for as long. I overclock it, but I also consider it overclocking my wallet. AMD or Intel doesn't matter to me either, although motherboard stability DOES. I'm considering myself the market these guys need to win, and the possibility of Intel releasing something that may be competitive to AMD's current lineup is something that I am in the "wait and see" stage as far as prices go. I was REALLY considering AMD this time, and was about to begin researching, but now I may wait just a little longer.

Note, even the 1080 video cards have dropped a lot it seems in the past few months. It'll be interesting to see what AMD responds with... and yes, this competition is great for us, maybe not so great for AMD. I haven't forgotten that AMD hosted the Hardocp party I went to in Dallas and I would love to see more of them. They do get points from me for that.

If you are looking to spend $250, don't wait on these chips, they are high end chips, expect at best just under $1k
 
Will this plug into current Kaby Lake motherboards, such as my Asus TUFF Z270 ?
 
Only true because we devs don't bother trying to take advantage of multi-threading. It's hard making sure things like race conditions don't blow the entire thing up.

No, I don't agree with the sentiment (the power is there, USE IT), but there it is.


In many cases so hard that it is impossible.

Not all code lends itself well to multithreading. I'd argue that we've gotten about as good as we are going to get on that front, not because of lazy developers, but because of the fundamentals of computer science.
 
Intel is a business first. You are fooling yourself if you think AMD wouldn't do the same thing if the roles were reversed. AMD is not some charity. I guarantee you AMD wouldn't be selling Ryzen at the current price if it crushed Intel.
Reading comprehension failed you.
 
In many cases so hard that it is impossible.

Not all code lends itself well to multithreading. I'd argue that we've gotten about as good as we are going to get on that front, not because of lazy developers, but because of the fundamentals of computer science.

I disagree. Code that doesn't lend itself well to multithreading is only that way due to poor architecture and design.

I can not tell you how much incredibly bad code goes to production due to poor development practices.
 
Let us all thank AMD for making the shitbag monopoly Intel start selling new products.
Fortunately I don't have to buy AMDs comparatively less powerful chips to benefit from them. ;)
 
Last edited:
I disagree. Code that doesn't lend itself well to multithreading is only that way due to poor architecture and design.

I can not tell you how much incredibly bad code goes to production due to poor development practices.


I believe you about bad code, and bad development practices. While it's not my area of expertise, I live it every day in trying to validate software for life supporting medical devices. (Though in this circumstance we care more about reliability than performance optimization)

That being said, I know quite a few academic experts in computer science. Back in ~2011 I used to be on the bandwagon for accusing those "lazy devs" for holding AMD's amazing many core designs back. They set me straight in a hurry. There is only so much you can do when it comes to designing an architecture that lends itself to parallelism. Particularly in game engines, a ton of calculations depend on the calculations made just before them, and attempting to multithread these actions will just result in nasty thread lock situations, which can be worked around, but the workarounds usually slow the code down to where it is less efficient than if it were just single threaded to begin with.
 
If Intel has no competition whats the measurement to determine if they are overcharging?

BTW Im not blind to your point but I don't agree with it based solely on the acronym you are using. HEDT or most importantly high end. It's a luxury market that demands its own n price

First, it's not a luxury market as it's not like Intel can only make a few hundred or thousand of these chips a year, but a high volume consumer product especially when HEDT are often Xeons that couldn't make the cut. Three easy ways amongst others: 1) Knowing that Intel is setting the price on higher chips that can be bought which come with the "privilege" of not having more and more of chip purposefully detuned/damaged/fused off/ turned off; 2) Charging very large premium increases for small gains in performance above the point in the point in the product stack where there is no competition; and 3) Continuing to raise prices year after year simply because there is no competition for that product.

I may be buying higher end parts as a sort of toy, but for many people who use these chips for their professional careers it has been obvious for the past several years that Intel is charging a significant and artificial premium simply because the corporate giant can. The last two times Intel faced real competition from AMD it, back in the 90s, spurred better chips and in the 2000s spurred the company to engage in anticompetitive practices that crippled AMD.
 
Back
Top