Discussion Intel current and future Lakes & Rapids thread

[ondma]

i7-1068NG7. What a name...

Did Intel get model number advice from HP? That Ice Lake x360's model number: Convertible 13-aw0xxx, following in HP's MD5 hash consumer model name tradition, e.g., the HP Pavilion 13-an0010nr.

That i7-1065G7 might be the "4.1 GHz" SKU: since Whiskey Lake, i7 advertised "Max Turbo Frequency" clocks are not Turbo Boost, but some wholly and utterly unrealistic "Thermal Velocity Boost" clocks. So, 3.9 GHz Turbo Boost & 4.1 GHz TVB.

I fully expect TVB shenanigans with all Intel 10nm and <10nm SKUs, not just i7/i9 SKUs.

The marketing spiel from Coffee Lake R / Whiskey Lake:



(emphasis mine)

You mean like AMD is already doing with PBO?

 

[Guru]

By the time Intel release anything decent on 10nm, AMD will release their Zen 2 refresh, likely on 6nm as TSMC advertised its capable of and uses the same process, so no need for any major change by AMD. If 6nm can bring in 400-500mhz higher operating frequencies over 7nm and what AMD are able to squeeze out now, its going to match and even beat Intel on stock frequencies.

I'm never buying Intel, unless they come up with much faster, much more energy efficient and much cheaper products than what amd is offering and they would have to win in gaming AND creativity/office workloads, while being cheaper and more energy efficient and have equal or same cores as amd.

 

[birdie]

The latest i7-1065G7 GeekBench 4 results:

CPU: https://browser.geekbench.com/v4/cpu/13954489 (Single: 5566, Multi: 17857)
GPU: https://browser.geekbench.com/v4/compute/4312201 (OpenCL score: 62885)

The base frequency has been upped from 1.3 to 1.5GHz (I still wonder why it's so low).

 

[IntelUser2000]

As for UN there are some Icelake SKUs floating around with this branding, one is called i7-1068NG7. .

By the way,

4.1GHz isn't that low. The top dual core(which translates into higher base clocks) Skylake at 15W was at 2.6GHz Base and 3.4GHz Turbo. The top 28W Skylake was at 3.3GHz Base and 3.6GHz Turbo.

The 10nm chips are beating Tock of 14nm already in clock speeds.

For Kabylake, the top 15W chip was a 2.5GHz Base 4GHz Turbo. Top 28W one was 3.5GHz Base and 4GHz Turbo. The spread is very low.

The quad core version, or Kabylake-R was at 1.9GHz Base and 4.2GHz Turbo.

If 6nm can bring in 400-500mhz higher operating frequencies over 7nm and what AMD are able to squeeze out now, its going to match and even beat Intel on stock frequencies.

Not only you are saying something that's to rile up people, its wishful thinking.

I think I said before those wishing for 6GHz frequencies that its never going to happen. All you need is to look at is history, with the world overclocking record charts.

We could see even with the 9900K. Its sold as a 5GHz chip. But it could at best do 200MHz more. If they want to take it just 200MHz beyond that, it starts requiring exotic cooling.

It took them 4 iterations of 14nm to reach the 5GHz point. Why do people still believe in the crazy idea that clock scaling will continue?

 

[jpiniero]

4.1GHz isn't that low.

It is when they will be competing with a 4.6-4.9 Comet Lake and likely 4.4-4.5 Renoir.
And I figure the 4.1 is going to be an ultra rare model you'll never see; so it's more in practice 3.9 tops.

 

[Guru]

By the way,

4.1GHz isn't that low. The top dual core(which translates into higher base clocks) Skylake at 15W was at 2.6GHz Base and 3.4GHz Turbo. The top 28W Skylake was at 3.3GHz Base and 3.6GHz Turbo.

The 10nm chips are beating Tock of 14nm already in clock speeds.

For Kabylake, the top 15W chip was a 2.5GHz Base 4GHz Turbo. Top 28W one was 3.5GHz Base and 4GHz Turbo. The spread is very low.

The quad core version, or Kabylake-R was at 1.9GHz Base and 4.2GHz Turbo.



Not only you are saying something that's to rile up people, its wishful thinking.

I think I said before those wishing for 6GHz frequencies that its never going to happen. All you need is to look at is history, with the world overclocking record charts.

We could see even with the 9900K. Its sold as a 5GHz chip. But it could at best do 200MHz more. If they want to take it just 200MHz beyond that, it starts requiring exotic cooling.

It took them 4 iterations of 14nm to reach the 5GHz point. Why do people still believe in the crazy idea that clock scaling will continue?

That is what people said about 4GHz, even Intel at the time dropped plans for Pentium 5 and 5GHz and went with Duo Core design.

While I don't think 6GHz is realistic anytime in the near future, 400MHz for AMD would be running 5GHz for most of their processors. Only the unreleased R9 3950x will run 3.7GHz, meaning it will run up to 5.1GHz is 6nm is able to provide that kind of speed bump. I mean people here thought that Zen 2 cpu's will regress in terms of clock speeds, most were saying clock speeds in the 3.8-4.3GHz range in terms of boost, what we got is 4.2-4.7GHz boost in their FIRST 7nm generation. I said back then, you can go through my posts and find 5/6 posts talking about this, that while I don't expect AMD to hit 5GHz, I expect them to run close enough to it, and that is exactly what they did. 3950X is running at 4.7GHz

The next step for AMD and their Zen 2 refresh would be to improve frequencies and 6nm will allow them to do that. I think TSMC CEO just said that it delivers an 18 per cent logic density gain and cost advantages. So we'll see. AMD have also shown themselves very competent in the CPU area, so I'm expecting some small, but specific improvements in the cpu design that will gain 2-4% more IPC, coupled with 300-500MHz more clock speeds and we are likely looking at about 8-12% performance improvements.

 

[jpiniero]

I have doubts AMD is using 6 nm; at least for the CPUs. Likely more 5 nm.

 

[TheGiant]

That is what people said about 4GHz, even Intel at the time dropped plans for Pentium 5 and 5GHz and went with Duo Core design.

While I don't think 6GHz is realistic anytime in the near future, 400MHz for AMD would be running 5GHz for most of their processors. Only the unreleased R9 3950x will run 3.7GHz, meaning it will run up to 5.1GHz is 6nm is able to provide that kind of speed bump. I mean people here thought that Zen 2 cpu's will regress in terms of clock speeds, most were saying clock speeds in the 3.8-4.3GHz range in terms of boost, what we got is 4.2-4.7GHz boost in their FIRST 7nm generation. I said back then, you can go through my posts and find 5/6 posts talking about this, that while I don't expect AMD to hit 5GHz, I expect them to run close enough to it, and that is exactly what they did. 3950X is running at 4.7GHz

The next step for AMD and their Zen 2 refresh would be to improve frequencies and 6nm will allow them to do that. I think TSMC CEO just said that it delivers an 18 per cent logic density gain and cost advantages. So we'll see. AMD have also shown themselves very competent in the CPU area, so I'm expecting some small, but specific improvements in the cpu design that will gain 2-4% more IPC, coupled with 300-500MHz more clock speeds and we are likely looking at about 8-12% performance improvements.

6 or 5nm will make things worse from freq standpoint because of heat flow density
TSMC process is just Apple process, they dont care bout others

by 3rd of august I have info dell xps with icelake should be out...any first hand infos>?

 

[IntelUser2000]

That is what people said about 4GHz, even Intel at the time dropped plans for Pentium 5 and 5GHz and went with Duo Core design.

Yes. Even 4GHz is on the not sane front.

It's not that top clocks have really advanced, they are just getting better at taking what used to be overclocking headroom, and selling it as stock.

Skylake did worse than Sandy Bridge for overclocks. Since 2011, by refining the same core and process for nearly 5 years they got a 10% increase.

Sure Ryzen 3K can reach 4.6GHz, but with 1 core auto boost enabled. Manual overclock is same as 2K! The top overclock records usually have SMT and most cores disabled and only run on 1 core.

Both Ryzen and Coffeelake chips are blowing their power budgets reaching stock performance. The 2002 then Intel CTO Pat Gelsinger prediction that such processors will have insane heat density is true today. The HSF combo we have on modern day processors are monstrosities and we also have lot more people running water cooling setups.

You might think its trivial because all you have to do is watch, but these engineers are basically trying to add a floor on the world's tallest skyscraper, or trying to make the fastest ever plane go 100km/h faster.

It's the same line of thinking people like Ray Kurzweil has. Just plot a straight line and assume it'll continue!

 

[mikk]

It is when they will be competing with a 4.6-4.9 Comet Lake and likely 4.4-4.5 Renoir.
And I figure the 4.1 is going to be an ultra rare model you'll never see; so it's more in practice 3.9 tops.

Renoir is going against Tigerlake. There is nothing for AMD against Icelake this year.

 

[IntelUser2000]

Renoir is going against Tigerlake. There is nothing for AMD against Icelake this year.

Icelake should have a 6 months lead over Renoir, and Renoir 6 months lead over Tigerlake.

 

[mikk]

https://downloadmirror.intel.com/28982/a08/Test_Driver.zip

There is a 26.20.16.9999 test driver with code names for some unannounced products.

Spoiler

; GLV HW
iGLVGT2 = "Intel Glenview UHD Graphics" "gfx-driver-ci-master-2624"

; TGL HW
INTEL_DEV_9A49 = "Intel(R) UHD Graphics, Gen12 LP" "gfx-driver-ci-master-2624"
INTEL_DEV_9A40 = "Intel(R) UHD Graphics, Gen12 LP" "gfx-driver-ci-master-2624"
INTEL_DEV_9A59 = "Intel(R) UHD Graphics, Gen12 LP" "gfx-driver-ci-master-2624"
INTEL_DEV_9A60 = "Intel(R) UHD Graphics, Gen12 LP" "gfx-driver-ci-master-2624"
INTEL_DEV_9A68 = "Intel(R) UHD Graphics, Gen12 LP" "gfx-driver-ci-master-2624"
INTEL_DEV_9A70 = "Intel(R) UHD Graphics, Gen12 LP" "gfx-driver-ci-master-2624"
INTEL_DEV_9A78 = "Intel(R) UHD Graphics, Gen12 LP" "gfx-driver-ci-master-2624"
INTEL_DEV_9A7F = "Intel(R) UHD Graphics, Gen12 LP" "gfx-driver-ci-master-2624"

; DG1 HW
iDG1LPDEV = "Intel(R) UHD Graphics, Gen12 LP DG1" "gfx-driver-ci-master-2624"

; LKF-R HW
INTEL_DEV_CAFE = "Intel(R) UHD Graphics, LKF-R" "gfx-driver-ci-master-2624"

; ATS HW
iATSHPDEV = "Intel(R) UHD Graphics, Gen12 HP ATS" "gfx-driver-ci-master-2624"

; LKF HW
INTEL_DEV_9840 = "Intel(R) UHD Graphics, LKF" "gfx-driver-ci-master-2624"
INTEL_DEV_9841 = "Intel(R) UHD Graphics, LKF" "gfx-driver-ci-master-2624"
INTEL_DEV_9842 = "Intel(R) UHD Graphics, LKF" "gfx-driver-ci-master-2624"
INTEL_DEV_9850 = "Intel(R) UHD Graphics, LKF" "gfx-driver-ci-master-2624"

; JSL Simulation
iJSLSIM = "Intel(R) UHD Graphics, JSL" "gfx-driver-ci-master-2624"

; EHL HW
iEHLLPDEVMOB1 = "Intel(R) UHD Graphics, EHL" "gfx-driver-ci-master-2624"
iEHLLPDEVMOB2 = "Intel(R) UHD Graphics, EHL" "gfx-driver-ci-master-2624"
iEHLLPDEVMOB3 = "Intel(R) UHD Graphics, EHL" "gfx-driver-ci-master-2624"
iEHLLPVAL = "Intel(R) UHD Graphics, EHL" "gfx-driver-ci-master-2624"

; RYF HW
iRYFGT2 = "Intel(R) UHD Graphics, RYF" "gfx-driver-ci-master-2624"

; RKL HW
iRKLLPGT1H32 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT1HPro32 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT1S32 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT0P5S16 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT1U32 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT0P5U16 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT0 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"

; DG2 HW
iDG2HP512 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
iDG2HP256 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
iDG2HP128 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"

; ADLS HW
INTEL_DEV_4680 = "Intel(R) UHD Graphics, ADL-S" "gfx-driver-ci-master-2624"
INTEL_DEV_46FF = "Intel(R) UHD Graphics, ADL-S" "gfx-driver-ci-master-2624"

One of the more interesting probably Rocket Lake:

; RKL HW
iRKLLPGT1H32 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT1HPro32 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT1S32 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT0P5S16 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT1U32 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT0P5U16 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"
iRKLLPGT0 = "Intel(R) UHD Graphics, RKL" "gfx-driver-ci-master-2624"

There is only GT0, GT0.5 and GT1, it doesn't mean it won't come with GT2 because this listing is super early.

GT1S32....GT1 with 32 EUs and S for socket I believe. This more or less confirmes it is Gen11 based because it is exactly half of GT2 64EU ICL-U.


Also interesting in regards to dedicated graphics:

; DG1 HW
iDG1LPDEV = "Intel(R) UHD Graphics, Gen12 LP DG1" "gfx-driver-ci-master-2624"

; DG2 HW
iDG2HP512 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
iDG2HP256 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
iDG2HP128 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"

 

[jpiniero]

There is only GT0, GT0.5 and GT1, it doesn't mean it won't come with GT2 because this listing is super early.

The leak suggested that only some Rocket Lake mobile was getting the 10 nm chiplet, the rest was getting a 14 nm one. Even 32 EU on 14 nm is going to be pretty big.

 

[Dayman1225]

; ADLS HW
INTEL_DEV_4680 = "Intel(R) UHD Graphics, ADL-S" "gfx-driver-ci-master-2624"
INTEL_DEV_46FF = "Intel(R) UHD Graphics, ADL-S" "gfx-driver-ci-master-2624"

Alder Lake - S? Seems a bit early for that to show up

 

[Dayman1225]

Some info from Q2 call:

• Snow Ridge production in early 2020
• Icelake Client shipping in volume
• 10nm FPGAs shipping in H2'19
• Two 10nm fabs in full prodcution
• Icelake SP production in H1'20, volume ramp in H2'20 - Yield ahead of schedule
• Cascade Lake is one of their fastest ramping products ever

 

[IntelUser2000]

JSL = Jasper Lake. But it might not be a product since it says simulation.

I'm guessing,
ATS = Probably Arctic Sound. A dGPU MCM successor to Kabylake-G.

Also the dGPUs might be clocked beyond 1.5GHz, with some reaching 2GHz.

The leak suggested that only some Rocket Lake mobile was getting the 10 nm chiplet, the rest was getting a 14 nm one. Even 32 EU on 14 nm is going to be pretty big.

The 14nm is probably canned and the GPU is all on 10nm. 32EU isn't Gen 9 but Gen 11.

There is only GT0, GT0.5 and GT1, it doesn't mean it won't come with GT2 because this listing is super early.

Best graphics for mobile and rest for desktops. Same tactic used in the Sandy Bridge generation.

 

[jpiniero]

The 14nm is probably canned and the GPU is all on 10nm. 32EU isn't Gen 9 but Gen 11.

What I mean is that it would be Gen11, but cut in half to compensate for the chiplet being on 14 nm. Maybe Alder will be the same deal, so desktop would get 48 EUs.

 

[IntelUser2000]

What I mean is that it would be Gen11, but cut in half to compensate for the chiplet being on 14 nm. Maybe Alder will be the same deal, so desktop would get 48 EUs.

They are not going to port Gen 11 to 14nm for an extremely small volume product. That's based on the SIPP roadmap and we already know details have changed on the leaked roadmaps.

 

[jpiniero]

They are not going to port Gen 11 to 14nm for an extremely small volume product.

The entire mainstream desktop lineup plus most of mobile is extremely small volume?

 

[IntelUser2000]

The entire mainstream desktop lineup plus most of mobile is extremely small volume?

How do you figure "most of mobile"? Even the client mobile roadmap had the graphics all at 10nm. It's only a segment in SIPP that showed 14nm.

Yes, I understand its fashionable to bash Intel, but I'm just simply stating reality. Either it'll be 14nm Gen 9 or 10nm Gen 11. "Porting" to a new process isn't like copying and pasting an icon in windows. When you move a particular uarch to a new process, it no longer is the same chip, because you have to account for the differences when designing it.

Any company that don't die will recover. It's called a business cycle. Some people believe it doesn't exist, or they believe the trend will only go in a straight line.

 

[jpiniero]

Maybe it'd make more sense if Rocket's IGP was Gen12 and not Gen11. So the 10 nm version will use Tigerlake's IGP chiplet while the 14 nm version is just a cut down version, 32 after all is a third of 96. If you are saying why they would do this, the obvious answer is that they want the entire lineup using an updated IGP but won't have the capacity to deliver due to the bad 10 nm yield.

 

[mikk]

How do you figure "most of mobile"? Even the client mobile roadmap had the graphics all at 10nm. It's only a segment in SIPP that showed 14nm.

The only 14nm +10nm gfx Rocket Lake is placed in the H/G segment. The U-series comes with 14nm graphics, The entire desktop lineup isn't even a chiplet, it's plain 14nm, this is for sure not extremely small volume. 32 EUs for GT1 must be something Gen11 based. Obviously Gen11 can be ported to 14nm.

Any thoughts for Gen12 HP DG2? 512= 512 EUs???

; DG2 HW
iDG2HP512 = "Intel(R) UHD Graphics, Gen12 HP DG2"
iDG2HP256 = "Intel(R) UHD Graphics, Gen12 HP DG2"
iDG2HP128 = "Intel(R) UHD Graphics, Gen12 HP DG2"

 

[zinfamous]

Absolutely not. I just cannot wrap my head around the fact that no one among Intel management had the guts to stand up against the company being actively destroyed/mismanaged for so long. Intel perfectly knew about Zen/Ryzen. Intel knew about TSMC/Samsung 7nm nodes. They still kept dragging their 14nm node when it was obvious they didn't even had the capacity to mass-produce many-cores CPUs not to mention that 14nm doesn't quite work for such a scenario 'cause you end up with CPUs which have an insane TDP.

when you start off by firing all your critics and hiring only yes men and women, why is it so hard to believe? It's no mystery how such an environment breeds utter incompetence at either the corporate level and elsewhere. Hell, there are some rather public-facing "very big" examples of this kind of broken "leadership" breeding incompetence on a daily basis.

 

[IntelUser2000]

The only 14nm +10nm gfx Rocket Lake is placed in the H/G segment. The U-series comes with 14nm graphics,

Actually, according to that really old roadmap, U has both 14+14 and 14+10. But the former is only for SIPP. H/G has no such successor.

The entire desktop lineup isn't even a chiplet, it's plain 14nm, this is for sure not extremely small volume. 32 EUs for GT1 must be something Gen11 based. Obviously Gen11 can be ported to 14nm.

If its Rocket Lake, its very much likely a chiplet. Again, look back at that roadmap.

Before you call moving uarch to a different process a "port", in actuality the specifications will change quite a bit. Smaller processes offer lower power, even if it may not be able to reach higher clock speeds. But the complications of newer processes make it only true on certain levels, so the behavior becomes very different.

Any thoughts for Gen12 HP DG2? 512= 512 EUs???

With high clocks. At 1.8GHz it'll be at 14TFlops and still occupy the high end, assuming it scales(unlike Gen 9 GT4e). Architectures seem to be converging in a way with high-level specifications being similar across multiple ISAs, because of diminishing returns in all areas.

The 128EU version will try to displace MX250-like setup for larger ultrabooks, and be faster than Tigerlake's GPU. The 256EU version and its variants go in the mid-range market. The 512EU is where you try to make your mark.

 

[LightningZ71]

Are you sure that the 128, 256, and 512 reference the EU count and not an L4 or graphics working memory cache of some sort similar to the old Iris-Pro setup? That's an awful lot of EUs for a highly memory constrained environment...