Discussion Intel current and future Lakes & Rapids thread

[Asterox]

Courtesy of VC:

View attachment 37564

"Greetings to the new gaming king"+if anyone mentions power consumption they will be shot.

 

[Abwx]

It happened at 7:30. Petrifying.

Viral marketing in daylight, at CES...

 

[Hitman928]

It happened at 7:30. Petrifying.

Gotcha, Yes, terrifying, lol.

 

[Asterox]

Viral marketing in daylight, at CES...

Great comedy or Intel and security, .......................but in reality.

Intel vs AMD Processor Security: Who Makes the Safest CPUs?

Who has the most secure processors, AMD or Intel?

www.tomshardware.com

"Intel currently has 242 publicly disclosed vulnerabilities, while AMD has only 16. That’s a 15:1 difference in AMD’s favor. The gap is just too large to ignore."

 

[Bouowmx]

Is Rocket Lake coming now or later? Because Z590 is out (in the hands of reviews).
https://wccftech.com/msi-z590-gaming-pro-carbon-wifi-motherboard-preview/ (no, this is not reposting another forum's pics, but I believe they are author's own pics)
Pics also confirm Rocket Lake is 11th Gen only Core i5 and up.

 

[jpiniero]

Is Rocket Lake coming now or later? Because Z590 is out (in the hands of reviews).
https://wccftech.com/msi-z590-gaming-pro-carbon-wifi-motherboard-preview/ (no, this is not reposting another forum's pics, but I believe they are author's own pics)
Pics also confirm Rocket Lake is 11th Gen only Core i5 and up.

That says preview, not review. The rumors did say that Z590 was coming first/before Rocket Lake.

 

[lobz]

Is Rocket Lake coming now or later? Because Z590 is out (in the hands of reviews).
https://wccftech.com/msi-z590-gaming-pro-carbon-wifi-motherboard-preview/ (no, this is not reposting another forum's pics, but I believe they are author's own pics)
Pics also confirm Rocket Lake is 11th Gen only Core i5 and up.

Mid-March

 

[Hulk]

Things are getting really interesting. Intel reports some gaming benchmarks vs the 5800x that show Intel has a 4% average better result. This is just about the clockspeed difference between these two parts.

IPC-wise I think these two cores are going to trade blows pretty evenly.

Of course you can't get more than 8 cores with Intel. And then there's the fact that you need a nuke (and cooling towers) for the Intel.

 

[majord]

intel are citing 'up to' 19% . wonder whattheir average is? , @ something less than 19% , IPC should still be a bit below Zen 3

 

[Hulk]

intel are citing 'up to' 19% . wonder whattheir average is? , @ something less than 19% , IPC should still be a bit below Zen 3

The game benches were directly compared to Zen 3. Zen 3 and RL are going to be close. But AMD will own the core count and the power/efficiency and the price premium for the high end parts.

Basically Intel has to clean up the scraps the way AMD used to. Funny turn of events.

 

[eek2121]

I don't participate here much anymore for reasons, but I figured I'd drop a few tidbits of info:

NOTE: before I start, for those of you that are new here, I don't own stock in Intel. I don't have a single Intel CPU or motherboard in my house, etc. In other words, I'm not a fanboy.

1. Many people think 10nm has yield issues. They have capacity issues. Also, a corporation like Intel does NOT divert from their internal roadmaps. Shoot, you'll find that AMD doesn't either. These products have a LONG lifecycle. It takes years to bring them from the concept stage to production stage. The bottom line is that even if Intel had 10nm ready for the desktop, Rocket Lake is still going out the door.
2. Intel is positioning itself to be extremely competitive with AMD. AMD is paying almost no attention to client based ML (machine learning) (as far as we know, and if they are, they haven't leaked a thing). Intel is rolling out Xe alongside Rocket Lake, Tiger Lake, and future generations. This is an early attempt at bringing intelligent ML to mainstream platforms. Apple has also done this with the M1 (and other chips) with their "neural engine". It baffles me why AMD is letting themselves be blindsided by this. Xe graphics were never about gaming, they were always about getting stuff done. Machine learning. Yes, HPG is launching later this year, but once again: Intel users will have Xe graphics onboard to handle things like AI upscaling, etc. What will AMD have? It remains to be seen if they will stumble or if they can adapt. The lack of leaks in this regard are pretty scary. They don't even have a decent video encoder or decoder right now...
3. Rocket lake was originally planned for March. Off the top of my head I believe it was March 17th (but I'd have to dig for the exact date and quite frankly I don't care). It was never slated for January. That leak was fake. There is an IPC increase, and yes, it will be Zen 3 chips *some of the time*. In non gaming tasks (or in gaming tasks that properly utilize cores) Zen 3 will still come out on top. The leaks involving the 5900X weren't necessarily fake or biased, but there are caveats that we'll all find out later on.

One thing I did NOT know, but rather, I learned: The decrease in core count wasn't due to heat or power, it was due to size. The drop in core count was to keep margins consistent. Nothing more. I was under the assumption that it was all about both heat and power. (EDIT: point of clarification: Intel would have pushed chips with higher power consumption or worse thermal performance out the door regardless. It was all about the chip size.)

Anyway, happy commenting. I'm going back into my hole for another 6 months or so (At least. Those of you that know me know where to find me). Happy new year to all!

 

[Kocicak]

Many people think 10nm has yield issues. They have capacity issues. Also, a corporation like Intel does NOT divert from their internal roadmaps. Shoot, you'll find that AMD doesn't either. These products have a LONG lifecycle. It takes years to bring them from the concept stage to production stage. The bottom line is that even if Intel had 10nm ready for the desktop, Rocket Lake is still going out the door.

The problem is, that Intel CPUs are now so weak compared to AMD products, that they can compete only when insanely clocked, consuming huge amounts of energy.

Rocket lake is not on 10nm, because it cannot clock so high and 14 nm can.

Intel producing CPUs consuming 250W and wanting customers to throw peltier water cooler on it consuming 350W MORE, is a comedy act at this moment.

 

[jj109]

Tiger Lake H is going to clock on 5GHz on multiple cores, already confirmed.
Rocket Lake will be easier to cool than Comet Lake because the core is 20% bigger at the same max power levels.
The up to 19% claim is based on SPEC2017 ST based on their fine print.

It is a little wild how the same people who were explaining why Zen3 was only a few % ahead of Zen2 in 1080p max settings gaming benchmarks have forgotten all of that, and are now using that as a bludgeon against Intel's IPC claims.

 

[Det0x]

I find this very disappointing, 14nm rocket lake can't even beat 10 core skylake, let alone compete with Zen3

Intel Core i9-11900K qualification sample compared with Core i9-10900K at 5.2 GHz @ https://videocardz.com/newz/intel-core-i9-11900k-qualification-sample-compared-with-core-i9-10900k-at-5-2-ghz

The testing platform features ASUS ROG STRIX RTX 3090, Z490 motherboard, and memory clocked at 3600 MHz (CL16). Both processors have been overclocked to 5.2 GHz and are cooled down by an industrial CPU chiller. The preview has been shared by Bibibili creator 二斤啦啦啦啦.

CPU-Z
In CPU-Z the Core i9-11900K scores 693 points which are around 11% faster than 10900K, however. In the multi-threaded test, the CPU scores 6723 points which means that Comet Lake-S CPU is still 13% faster. This is of course due to the fact that the 10th Gen Core CPU offers more threads (20) than the 11th Gen Core (16).




Cinebench
The CPU has been tested in all three popular versions of Cinbench. It looks that the 11th Gen Core processor has a clear advantage in the single-threaded benchmark while falling to compete with the current flagship in the multi-threaded benchmark.

	Cinebench R15	Cinebench R20	Cinebench R23
Core i9-11900K	250 / 2356	610 / 5672	1582 / 13864
Core i9-10900K	223 / 2655	526 / 6374	1388 / 16535

3DMark

Combined Scores	Fire Strike	Fire Strike Extreme	Time Spy	Time Spy Extreme
Core i9-11900K	31294	19303	11339	8289
Core i9-10900K	30225	19421	11406	8525

 

[Panino Manino]

Why Intel compared this 8 Core with AMD 12 Core? For games?
Shouldn't the comparison be with the 5800X? It's equal or faster than the 5900X and costs U$200 less.

 

[Kocicak]

Because they want to price it like 5900X?

 

[Abwx]

Because they want to price it like 5900X?

Because they want people to believe that it require 12 Zen 3 cores to match 8 RKL cores, otherwise they would had used a 5800X as comparison since they could had displayed better numbers..

 

[exquisitechar]

I find this very disappointing, 14nm rocket lake can't even beat 10 core skylake, let alone compete with Zen3

Intel Core i9-11900K qualification sample compared with Core i9-10900K at 5.2 GHz @ https://videocardz.com/newz/intel-core-i9-11900k-qualification-sample-compared-with-core-i9-10900k-at-5-2-ghz

The testing platform features ASUS ROG STRIX RTX 3090, Z490 motherboard, and memory clocked at 3600 MHz (CL16). Both processors have been overclocked to 5.2 GHz and are cooled down by an industrial CPU chiller. The preview has been shared by Bibibili creator 二斤啦啦啦啦.

CPU-Z
In CPU-Z the Core i9-11900K scores 693 points which are around 11% faster than 10900K, however. In the multi-threaded test, the CPU scores 6723 points which means that Comet Lake-S CPU is still 13% faster. This is of course due to the fact that the 10th Gen Core CPU offers more threads (20) than the 11th Gen Core (16).
View attachment 37630

View attachment 37631

Cinebench
The CPU has been tested in all three popular versions of Cinbench. It looks that the 11th Gen Core processor has a clear advantage in the single-threaded benchmark while falling to compete with the current flagship in the multi-threaded benchmark.

	Cinebench R15	Cinebench R20	Cinebench R23
Core i9-11900K	250 / 2356	610 / 5672	1582 / 13864
Core i9-10900K	223 / 2655	526 / 6374	1388 / 16535

View attachment 37632View attachment 37633View attachment 37634

3DMark

Combined Scores	Fire Strike	Fire Strike Extreme	Time Spy	Time Spy Extreme
Core i9-11900K	31294	19303	11339	8289
Core i9-10900K	30225	19421	11406	8525

Several things are off here, best wait for proper reviews.

 

[Det0x]

Several things are off here, best wait for proper reviews.

Like what ?

i see some people say it don't match the previous leaked 700+ CPU-Z score, but 693/5.2 -> *5.3 = 706 points

 

[Hulk]

I don't participate here much anymore for reasons, but I figured I'd drop a few tidbits of info:

NOTE: before I start, for those of you that are new here, I don't own stock in Intel. I don't have a single Intel CPU or motherboard in my house, etc. In other words, I'm not a fanboy.

1. Many people think 10nm has yield issues. They have capacity issues. Also, a corporation like Intel does NOT divert from their internal roadmaps. Shoot, you'll find that AMD doesn't either. These products have a LONG lifecycle. It takes years to bring them from the concept stage to production stage. The bottom line is that even if Intel had 10nm ready for the desktop, Rocket Lake is still going out the door.
2. Intel is positioning itself to be extremely competitive with AMD. AMD is paying almost no attention to client based ML (machine learning) (as far as we know, and if they are, they haven't leaked a thing). Intel is rolling out Xe alongside Rocket Lake, Tiger Lake, and future generations. This is an early attempt at bringing intelligent ML to mainstream platforms. Apple has also done this with the M1 (and other chips) with their "neural engine". It baffles me why AMD is letting themselves be blindsided by this. Xe graphics were never about gaming, they were always about getting stuff done. Machine learning. Yes, HPG is launching later this year, but once again: Intel users will have Xe graphics onboard to handle things like AI upscaling, etc. What will AMD have? It remains to be seen if they will stumble or if they can adapt. The lack of leaks in this regard are pretty scary. They don't even have a decent video encoder or decoder right now...
3. Rocket lake was originally planned for March. Off the top of my head I believe it was March 17th (but I'd have to dig for the exact date and quite frankly I don't care). It was never slated for January. That leak was fake. There is an IPC increase, and yes, it will be Zen 3 chips *some of the time*. In non gaming tasks (or in gaming tasks that properly utilize cores) Zen 3 will still come out on top. The leaks involving the 5900X weren't necessarily fake or biased, but there are caveats that we'll all find out later on.

One thing I did NOT know, but rather, I learned: The decrease in core count wasn't due to heat or power, it was due to size. The drop in core count was to keep margins consistent. Nothing more. I was under the assumption that it was all about both heat and power. (EDIT: point of clarification: Intel would have pushed chips with higher power consumption or worse thermal performance out the door regardless. It was all about the chip size.)

Anyway, happy commenting. I'm going back into my hole for another 6 months or so (At least. Those of you that know me know where to find me). Happy new year to all!

Thanks for posting this. I have a few follow up questions if you don't mind.

To a non-expert like myself capacity issues and yield issues seem closely related. Are you implying yields are good they just don't have the fabs to get out enough 10nm?

Perhaps AMD isn't focused on ML (future) because they need to survive now? First deal with getting something out the door they can sell now and then deal with ML once they have some breathing room?

I don't think January vs March is a big deal. I think most people think that RL should have probably been out a year ago and at 10nm for Intel to have retained it's normal dominance.

 

[inf64]

Looking at those scores @ 5.2Ghz that OCed Rocketlake part cannot match 5800X @ stock in those benchmarks (both ST or MT). It is clear that Zen3 has noticeable IPC advantage that +500Mhz clock cannot compensate. Gaming numbers that intel shared are also very close between 5900X and Rocketlake, so there won't be any noticeable difference.No mention of power draw- 5800X could be drawing HALF of Rocketlake's power while performing better in majority of tasks. The only saving grace is availability as AMD migh struggle supplying Zen3 parts to the market.

 

[dmens]

Many people think 10nm has yield issues. They have capacity issues.

LOL, nice spin. One leads to the other. You should be able to figure out which way.

Also, a corporation like Intel does NOT divert from their internal roadmaps. Shoot, you'll find that AMD doesn't either. These products have a LONG lifecycle. It takes years to bring them from the concept stage to production stage. The bottom line is that even if Intel had 10nm ready for the desktop, Rocket Lake is still going out the door.

Anyone seen a Cannonlake lately? 10nm desktop parts were planned and abandoned before the Rocketlake backport was even conceptualized as a reactionary measure. Saying that Intel would have bothered with a 14nm backport if they had 10nm parts ready in 2017 is absurd.

 

[Hulk]

Looking at those scores @ 5.2Ghz that OCed Rocketlake part cannot match 5800X @ stock in those benchmarks (both ST or MT). It is clear that Zen3 has noticeable IPC advantage that +500Mhz clock cannot compensate. Gaming numbers that intel shared are also very close between 5900X and Rocketlake, so there won't be any noticeable difference.No mention of power draw- 5800X could be drawing HALF of Rocketlake's power while performing better in majority of tasks. The only saving grace is availability as AMD migh struggle supplying Zen3 parts to the market.

I'm sooo curious as to the following:
1. Will AMD catch up with Zen 3 demand by the time RL arrives.
2. RL pricing and will it be affected by #1? If Zen 3 is available then Intel is going to have to get competitive on price.

All of those partnerships AMD was showing today at CES. They're saying "We're here!"

 

[ondma]

Because they want to price it like 5900X?

5900x usually is the best binned, will have the highest single core performance. 5800x is generally equal or slightly slower than 5900x in gaming.

 

[ondma]

I find this very disappointing, 14nm rocket lake can't even beat 10 core skylake, let alone compete with Zen3

Intel Core i9-11900K qualification sample compared with Core i9-10900K at 5.2 GHz @ https://videocardz.com/newz/intel-core-i9-11900k-qualification-sample-compared-with-core-i9-10900k-at-5-2-ghz

The testing platform features ASUS ROG STRIX RTX 3090, Z490 motherboard, and memory clocked at 3600 MHz (CL16). Both processors have been overclocked to 5.2 GHz and are cooled down by an industrial CPU chiller. The preview has been shared by Bibibili creator 二斤啦啦啦啦.

CPU-Z
In CPU-Z the Core i9-11900K scores 693 points which are around 11% faster than 10900K, however. In the multi-threaded test, the CPU scores 6723 points which means that Comet Lake-S CPU is still 13% faster. This is of course due to the fact that the 10th Gen Core CPU offers more threads (20) than the 11th Gen Core (16).
View attachment 37630

View attachment 37631

Cinebench
The CPU has been tested in all three popular versions of Cinbench. It looks that the 11th Gen Core processor has a clear advantage in the single-threaded benchmark while falling to compete with the current flagship in the multi-threaded benchmark.

	Cinebench R15	Cinebench R20	Cinebench R23
Core i9-11900K	250 / 2356	610 / 5672	1582 / 13864
Core i9-10900K	223 / 2655	526 / 6374	1388 / 16535

View attachment 37632View attachment 37633View attachment 37634

3DMark

Combined Scores	Fire Strike	Fire Strike Extreme	Time Spy	Time Spy Extreme
Core i9-11900K	31294	19303	11339	8289
Core i9-10900K	30225	19421	11406	8525

Wow, 10 cores are faster than 8 in multi threaded benchmarks. Who would ever have thought that.
You do realize, I hope, that RL would have to have 25% higher IPC to equal CL in multi threaded benchmarks.