Intel's Future Chips: News, Rumours & Reviews

[jimmysmitty]

Its hard to trust a roadmap without context or where it came from. I am sure Intel will announce something in the near future.

 

[-Fran-]

Its hard to trust a roadmap without context or where it came from. I am sure Intel will announce something in the near future.

I can also add: I'm sure it will rain before the end of the year. I won't compromise on places though! That might be giving away too much information.

Cheers! 😀

 

[goldstone77]

Intel launches Xeon E-2100 workstation line
If you were looking for a desktop part with ECC, it is here
Jul 12, 2018 by Charlie Demerjian

https://www.semiaccurate.com/2018/07/12/intel-launches-xeon-e-2100-workstation-line/

Note: Intel added PCI Express 3.0 Lanes (CPU+Chipset), which include the lanes on the motherboard.

 

[goldstone77]

Intel CPU Roadmap 2018-2020: Intel Sticking With 14 nm, 10 nm Intel CPUs Pushed To 2020
https://segmentnext.com/2018/07/10/intel-cpu-roadmap-10-nm-2020/

The slide is hard to see, but it shows several interesting things:

• 
  - Canon lake has vanished from the roadmap.
  - Only 10nm parts shown in the slide are Icelake and, later on, Tigerlake
  - 10 nm server Icelake is expected in Q3 2020
  - 10 nm desktop Icelake expected in Q4 2019
  - 10 nm laptop (H series) Icelake expected Q2 2020
  - The earliest 10 nm parts are expected in Q2 2019 in the ultramobile segment. The same that canonlake targeted. I wonder if they will launch low volume too to act as pipecleaner for the 10 nm process.

I don't know if the slide is legit, but it's consistent with all rumors I heard.

Intel responded to this:

Speaking of Intel, reportedly the company is shutting down its Extreme Edition CPU brand.

This is according to former principal Engineer and Performance Architect at Intel, Francois Piednoel, who revealed that Intel is killing its high-end CPU brand which is a big mistake on Intel’s part.

The Extreme Edition CPU brand has been an important part of Intel’s dominance in the market and with Intel killing it we won’t be seeing any Extreme Edition CPUs in the market in the future if the report is true.

This story isn't true, François Piednoël typically just spouts of stuff that isn't true. He is as pro Intel as any single human can possibly be, for future reference.

"There is no change to the branding of the Intel Core Extreme Edition processor and Intel Core X-series processor family," Intel said.

Edit: I have to add that the roadmap was a speculative roadmap made by Dayman:
https://twitter.com/Dayman58/status/1016143133283930112

 

[aldaia]

Edit: I have to add that the roadmap was a speculative roadmap made by Dayman:
https://twitter.com/Dayman58/status/1016143133283930112

No wonder it was consistent with the rumors. It was actually made putting all rumors together 🙂

 

[YoAndy]

FYI Intel’s 9th Gen Core i9-9900K 8-Core CPU Could Arrive In August

https://fossbytes.com/intel-9th-gen-coffee-lake-refresh-8-core-soldered-ihs-cpu/

 

[-Fran-]

That's an interesting rumour... Higher clocks and 2 extra cores will most definitely push power consumption to the sky!

I know Intel's process is really good; 14nm+ is proof of that and they're even going back to soldering for mainstream (FRIGGIN' FINALLY, INTEL). I would imagine it's really safe to just assume these will be in 14nm++ and speed-wise they will most likely be at the very top of what they can do; kind of what AMD has done with the 2700X.

Interesting times.

The only doubt that I have is if they will display the new MoBos / chipset for these new CPUs soon or not. I'm expecting the mainstream i9 will really really need a 125W TDP envelope.

Cheers!

 

[jimmysmitty]

That's an interesting rumour... Higher clocks and 2 extra cores will most definitely push power consumption to the sky!

I know Intel's process is really good; 14nm+ is proof of that and they're even going back to soldering for mainstream (FRIGGIN' FINALLY, INTEL). I would imagine it's really safe to just assume these will be in 14nm++ and speed-wise they will most likely be at the very top of what they can do; kind of what AMD has done with the 2700X.

Interesting times.

The only doubt that I have is if they will display the new MoBos / chipset for these new CPUs soon or not. I'm expecting the mainstream i9 will really really need a 125W TDP envelope.

Cheers!

We can probably look at the i7 7820x for reference. It has a 140W TDP. However it also has more features, quad channel RAM and the new UPI vs QPI while mainstream uses DMI, slower but still fast. I would assume 125 could be correct but still dropping two lanes for the IMC and a slower lower power interconnect might allow it to drop below that.

Plus I don't think temps will be that bad. With 2 additional cores on the same uArch it will have a larger surface area to dissipate heat and might keep temps similar to current 6 cores.

 

[goldstone77]

Look at the power draw differential of the 7700K and 8700K. Also look at it under load at 5GHz to get an idea of what to expect.

https://www.tomshardware.com/reviews/intel-coffee-lake-i7-8700k-cpu,5252-12.html

 

[goldstone77]

i7-7700K
Processor Base Frequency
4.20 GHz
Max Turbo Frequency
4.50 GHz
TDP
91 W
https://ark.intel.com/products/97129/Intel-Core-i7-7700K-Processor-8M-Cache-up-to-4_50-GHz


i7-8700K
Processor Base Frequency
3.70 GHz
Max Turbo Frequency
4.70 GHz
TDP
95 W
https://ark.intel.com/products/126684/Intel-Core-i7-8700K-Processor-12M-Cache-up-to-4_70-GHz

TDP
Thermal Design Power (TDP) represents the average power, in watts, the processor dissipates when operating at Base Frequency with all cores active under an Intel-defined, high-complexity workload. Refer to Datasheet for thermal solution requirements.

My take away is that Intel could do what they did before, and reduce base frequency to maintain a good looking TDP. You could potentially see ~200W+ power draw in a torture test with 8 cores at 5.5GHz.

 

[jimmysmitty]

i7-7700K
Processor Base Frequency
4.20 GHz
Max Turbo Frequency
4.50 GHz
TDP
91 W
https://ark.intel.com/products/97129/Intel-Core-i7-7700K-Processor-8M-Cache-up-to-4_50-GHz


i7-8700K
Processor Base Frequency
3.70 GHz
Max Turbo Frequency
4.70 GHz
TDP
95 W
https://ark.intel.com/products/126684/Intel-Core-i7-8700K-Processor-12M-Cache-up-to-4_70-GHz

TDP
Thermal Design Power (TDP) represents the average power, in watts, the processor dissipates when operating at Base Frequency with all cores active under an Intel-defined, high-complexity workload. Refer to Datasheet for thermal solution requirements.

My take away is that Intel could do what they did before, and reduce base frequency to maintain a good looking TDP. You could potentially see ~200W+ power draw in a torture test with 8 cores at 5.5GHz.

But do we use overclocked or stock settings to determine what the TDP is? I would go by stock clocks. Overclocking anything is going to go beyond the normal TDP so tat is expected.

I would though be impressed to see an 8 core chip at 5.5GHz so long as it is not using LN2. Doubt it though.

 

[goldstone77]

But do we use overclocked or stock settings to determine what the TDP is? I would go by stock clocks. Overclocking anything is going to go beyond the normal TDP so tat is expected.

I would though be impressed to see an 8 core chip at 5.5GHz so long as it is not using LN2. Doubt it though.

Stock settings, TDP by Intel's definition would be at base clock, which they could make a ~3.2GHz base at 95W TDP. 5.5GHz could be similar to the 8086K, single core turbo boot of 5GHz. This is all speculation anyway.

 

[-Fran-]

But if Intel goes that route to keep the TDP numbers low, the CPU won't be able to pull away from Zen *at TDP*. That is to say, much like it's happening now, it will be another 7900X shenanigan from Intel. They need to put the TDP higher so the base clocks are also higher enough to pull away from Zen.

Every single reviewer will make 1:1 comparisons this time using the same cooling solutions and the conclusions will be really interesting to draw now.

Cheers!

 

[jimmysmitty]

But if Intel goes that route to keep the TDP numbers low, the CPU won't be able to pull away from Zen *at TDP*. That is to say, much like it's happening now, it will be another 7900X shenanigan from Intel. They need to put the TDP higher so the base clocks are also higher enough to pull away from Zen.

Every single reviewer will make 1:1 comparisons this time using the same cooling solutions and the conclusions will be really interesting to draw now.

Cheers!

Reviewers will but there are people who like with Bulldozer will ignore TDP or per clock etc.

it depends on the person. Personally I don't care about TDP unless its insane (like 9590 insane) but I still doubt its going to be insanely higher.

 

[-Fran-]

Reviewers will but there are people who like with Bulldozer will ignore TDP or per clock etc.

it depends on the person. Personally I don't care about TDP unless its insane (like 9590 insane) but I still doubt its going to be insanely higher.

Well, enthusiasts will just use the best cooling they can afford regardless of what they buy, so I don't disagree. My point was more in-line with how now the conclusions will be far more easier to draw, just like they were with Phenom II and Core2Quad days.

From what we know: AMD is behind IPC, but has an upper hand in communication intra-core (I'd say at least). Intel has the upper hand in process (higher clocks, mainly; I'll just take this as granted), AMD has an edge with core scaling and Intel has an advantage with certain instruction sets flows (including AVX family).

On a very high and early impression, clock for clock, I think Intel still has the advantage, but I'll be willing to say AMD will be able to scale better. Intel having the clock advantage will add to their overall prowess.

AMD needs to pull a rabbit out of their process hat to pull away from Intel this time around. I'm willing to say that is the real deciding factor of the next gen.

Cheers!

 

[jimmysmitty]

Well they can scale right now with MCM but I don't think anything above 8 core will scale well TDP or power wise, I think they will run into a lot of the same issues Intel does.

Guess we will see although I think Intel needs to pull a damn human from their hat to get back into their lead.

 

[goldstone77]

Intel will at least go far enough to pass Ryzen performance comparatively at base clock. I personally, would crank up base clocks across the board as far as it can go like AMD does, and let TDP numbers fall where they may, unless they need to keep them for certain product lines.

 

[goldstone77]

The latest rumor:

https://videocardz.com/76808/intel-core-i9-9900k-i7-9700k-and-i5-9600k-possible-specifications-emerge

 

[jimmysmitty]

I almost feel like Intel should just do HT across the board for the top end and in mid to low end drop it. So i9/i7 and top end i5s with HT and mid end i5s and lower no HT.

What feels weird though is the cache between the i7 and i9. Why would they disable 33% of it.

 

[-Fran-]

I almost feel like Intel should just do HT across the board for the top end and in mid to low end drop it. So i9/i7 and top end i5s with HT and mid end i5s and lower no HT.

What feels weird though is the cache between the i7 and i9. Why would they disable 33% of it.

Maybe 14nm++ is not yielding as much as Intel would hope to?

Cache is very extensive on the die, so I would imagine that 33% is because of ~33%+ of chips are not qualified to be i9s and they have to downgrade them to i7s. I would rather they keep the active cores in them with less cache (if what I'm thinking is true).

Cheers!

 

[jimmysmitty]

I almost feel like Intel should just do HT across the board for the top end and in mid to low end drop it. So i9/i7 and top end i5s with HT and mid end i5s and lower no HT.

What feels weird though is the cache between the i7 and i9. Why would they disable 33% of it.

Maybe 14nm++ is not yielding as much as Intel would hope to?

Cache is very extensive on the die, so I would imagine that 33% is because of ~33%+ of chips are not qualified to be i9s and they have to downgrade them to i7s. I would rather they keep the active cores in them with less cache (if what I'm thinking is true).

Cheers!

I highly doubt that enhancements to 14nm would kill yields that much. If anything the yields should be better with better power numbers.

Still it will be interesting to see if anyone delids one to see why the cache is cut. 33% is an odd number though.

 

[goldstone77]

Only reasons I can think of cutting cache is to limit performance for segmentation. We need to take a look a synthetic benchmarks to check for performance limited by threads! Having the 8/16 and 8/8 perform nearly the same for a wide range of test would be problematic! I'm sure the process is exceeding good, but yields could, potentially, be a problem with the larger Die.

 

[jimmysmitty]

That makes me wish they would keep it i7, i5 and i3. Simple. i7 has 8/16, i5 has 6/12 and i3 has 4/8. Performance still stays balanced between low to high end. Makes it simple for people to know what they can get and want. Leave i9 for HEDT with 8/16 or more, up to whatever they can squeeze in.

 

[goldstone77]

That makes me wish they would keep it i7, i5 and i3. Simple. i7 has 8/16, i5 has 6/12 and i3 has 4/8. Performance still stays balanced between low to high end. Makes it simple for people to know what they can get and want. Leave i9 for HEDT with 8/16 or more, up to whatever they can squeeze in.

I think the naming scheme is also about pricing. Let's see how much they charge for the i9. I was thinking 8/16K $499; 8/16 $449; 8/8 $399.

 

[jimmysmitty]

Probably correct about that. Although I wouldn't mind if Intel decided to price cut AMD a bit to push prices down more. I still liked it when it was simple but thats just me. Simplicity.

Instead now we will have i7/i9 for mainstream and HEDT and then i5/i3 for desktop with varying cores and threads. Hard for me to remember all of this.