News Intel Makes it Official: Eight Core Tiger Lake Chips Coming

[Admin]

Intel confirmed via a blog post that eight-core Tiger Lake processors are coming to market.

Intel Makes it Official: Eight Core Tiger Lake Chips Coming : Read more

 

[Dsplover]

Great news. AMD Vermeer will be great, their Cezanne sounds great, and Tiger Lake sounds great.

We are spoiled rotten from competition.

 

[watzupken]

I am still awaiting independent Tiger Lake U reviews.

 

[ThermalPaste]

In terms of multi-core, Intel is way behind. I'm sure Tiger Lake won't be a slouch, but at the same time, I doubt it'll be anywhere near the versatility of AMD's monster Ryzen CPU's, especially the 4000 series.

 

[simba2020]

Hard to believe 8 core tigerlake is coming any time in the foreseeable future given yield issues on 10nm. "More detail at a later date" does not inspire any sort of confidence.

 

[Rdslw]

Great news. AMD Vermeer will be great, their Cezanne sounds great, and Tiger Lake sounds great.

We are spoiled rotten from competition.

Finally, I would say, for a while my laptop wanted to retire but new stuff was just not compelling to do the jump. Gains were like 20% + nvme from 5y old laptop.
and now with ryzen 4k its like 300%, and new stuff that's coming it will be even more.
I cant wait for both nvidia gpu's and new cpu's drip down to low power stuff where I dwell.

 

[setx]

"Soon" is likely 2022, maybe even 2022 Q4.

 

[st379]

It is not clear from the article are those the 7-25 watt or 45 watt?

I think Intel stay with pathetic quad cores on the 7-25 watt.

Good luck for Intel closing up to 200% performance gap between Ice Lake 10nm+++++++++++++ and Ryzen 4800u with a quad core.

 

[leeenghuat]

soon Intel will release 7nm chip. THEY ARE NOW DOING IT. I AM INTEL WORKER

 

[GetSmart]

Intel Tiger Lake U first tests https://zhuanlan.zhihu.com/p/214334613

There are also a bunch of leaks on Geekbench. For example, Intel Core i7-1185G7 https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=i7-1185G7&dir=desc&sort=multicore_score many of which are very recent (few days ago).. 😉

Thus do expect 8-core Intel Tiger Lake H to be roughly twice faster than Intel Tiger Lake U.. 🐯

 

[st379]

Intel Tiger Lake U first tests https://zhuanlan.zhihu.com/p/214334613

There are also a bunch of leaks on Geekbench. For example, Intel Core i7-1185G7 https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=i7-1185G7&dir=desc&sort=multicore_score many of which are very recent (few days ago).. 😉

Thus do expect 8-core Intel Tiger Lake H to be roughly twice faster than Intel Tiger Lake U.. 🐯

Thank you. Very interesting. You deserve a like .

From the benchmarks of Cinebench the r5 4500u (2615) almost score the same as the Tiger Lake(2626).

The 4600u (3292), 4700u (3509) and 4800u(4328) are a lot stronger.

It is fair to say that Tiger Like is DEAD ON ARRIVAL.

Here is Ice Lake 10nm++++++ vs ryzen 4800u for reference at 13:50:

View: https://youtu.be/hFYdHkvRs2c?t=830

 

[PCWarrior]

https://zhuanlan.zhihu.com/p/214334613
(a) The 1185G7 is not in the charts. The second-ranked 1165G7 is.

(b) Absolutely amazing that even the second-ranked sku has a single-core score in Cinebench R20 of 589. That is 11% faster than even an i9 10900K (which has the highest Cinebench score so far with 531 and that of course is achieved with a frequency of 5.3GHz). I imagine that the 1186G7 breaks the 600 boundary. Also in multicore the 1165G7 scores higher than a desktop 9600K. Not bad at all.

(c) Absolutely hilarious that the “15W” 4800U draws a whooping 63W during Cinebench. As I said and on another thread AMD is using smoke and mirrors with Renoir. Plug-in performance use unlimited power, then in battery mode where no reviewer benchmarks for scores drop clocks and power to 15W so that it looks good on battery benchmarks and we fool everyone that it was indeed 15W perfomance all along. Good trick, seems to be working on ignorant fanboys. Taking cheating on benchmarks to another level.

(d) Cinebench performance is largely academic anyway. Not only because the percentage of people using Cinema 4d is less than 0.001% but because CPU-only 3D rendering is becoming a thing of the past. Most other similar applications (e.g. Blender) have moved to offer a GPU-only rendering option that is vastly faster than even the fastest cpu-only rendering time. I haven’t seen a single 3D artist not using gpu-acceleration when provided. Moreover, even Maxon with the latest Cinema 4D R21 now offers natively the Maya Arnold renderer and this is the renderer of choice between 3D artists because it offers superior quality over Maxon’s own renderer (which is what is used in Cinebench R20). And Arnold on Maya now comes with GPU acceleration too. Even a year ago the GPU-only rendering quality was indistinguishable from the cpu-only rendering quality. See articles below.
https://techgage.com/news/autodesk-announces-arnold-renderer-gpu-beta/
https://techgage.com/article/autodesk-arnold-render-gpu-beta-performance/

--------------------------------------------
https://browser.geekbench.com/v5/cpu/3710380
Geekbench 5 is a solid CPU benchmark. It’s a proper all-round benchmark testing CPU performance in a variety of workloads. The 1185G7 scores 1610 in single core and 6113 in multicore. Again that single-core score is the highest we have ever seen so far sans extreme overclocking with liquid nitrogen. For reference the 10900K with its 5.3GHz single-core boost does around 1400. So that’s 15% faster than the 10900K. Also, the 4800U does only around 1150-1200 single core and 6400-6600 multicore. So, the 1185G7 is over 1/3 faster in single-core (and pretty much per core so it also has the faster 1T, 2T, 3T, 4T and even 5T-5.5T performance). The pure multithreaded performance is only 4.5%-7.5% lower. Not bad given it consumes over 30% less power.

 

[st379]

https://zhuanlan.zhihu.com/p/214334613
(a) The 1185G7 is not in the charts. The second-ranked 1165G7 is.

(b) Absolutely amazing that even the second-ranked sku has a single-core score in Cinebench R20 of 589. That is 11% faster than even an i9 10900K (which has the highest Cinebench score so far with 531 and that of course is achieved with a frequency of 5.3GHz). I imagine that the 1186G7 breaks the 600 boundary. Also in multicore the 1165G7 scores higher than a desktop 9600K. Not bad at all.

(c) Absolutely hilarious that the “15W” 4800U draws a whooping 63W during Cinebench. As I said and on another thread AMD is using smoke and mirrors with Renoir. Plug-in performance use unlimited power, then in battery mode where no reviewer benchmarks for scores drop clocks and power to 15W so that it looks good on battery benchmarks and we fool everyone that it was indeed 15W perfomance all along. Good trick, seems to be working on ignorant fanboys. Taking cheating on benchmarks to another level.

(d) Cinebench performance is largely academic anyway. Not only because the percentage of people using Cinema 4d is less than 0.001% but because CPU-only 3D rendering is becoming a thing of the past. Most other similar applications (e.g. Blender) have moved to offer a GPU-only rendering option that is vastly faster than even the fastest cpu-only rendering time. I haven’t seen a single 3D artist not using gpu-acceleration when provided. Moreover, even Maxon with the latest Cinema 4D R21 now offers natively the Maya Arnold renderer and this is the renderer of choice between 3D artists because it offers superior quality over Maxon’s own renderer (which is what is used in Cinebench R20). And Arnold on Maya now comes with GPU acceleration too. Even a year ago the GPU-only rendering quality was indistinguishable from the cpu-only rendering quality. See articles below.
https://techgage.com/news/autodesk-announces-arnold-renderer-gpu-beta/
https://techgage.com/article/autodesk-arnold-render-gpu-beta-performance/

--------------------------------------------
https://browser.geekbench.com/v5/cpu/3710380
Geekbench 5 is a solid CPU benchmark. It’s a proper all-round benchmark testing CPU performance in a variety of workloads. The 1185G7 scores 1610 in single core and 6113 in multicore. Again that single-core score is the highest we have ever seen so far sans extreme overclocking with liquid nitrogen. For reference the 10900K with its 5.3GHz single-core boost does around 1400. So that’s 15% faster than the 10900K. Also, the 4800U does only around 1150-1200 single core and 6400-6600 multicore. So, the 1185G7 is over 1/3 faster in single-core (and pretty much per core so it also has the faster 1T, 2T, 3T, 4T and even 5T-5.5T performance). The pure multithreaded performance is only 4.5%-7.5% lower. Not bad given it consumes over 30% less power.

<<Inappropriate content removed by moderator>>. Enjoy these 2 videos.

4700u destroy Ice lake 10nm+++++++++++:

View: https://www.youtube.com/watch?v=dsvat45M5nU


4800u destroy Ice Lake 10nm++++++++++:

View: https://www.youtube.com/watch?v=hFYdHkvRs2c

 

[Shadowclash10]

In terms of multi-core, Intel is way behind. I'm sure Tiger Lake won't be a slouch, but at the same time, I doubt it'll be anywhere near the versatility of AMD's monster Ryzen CPU's, especially the 4000 series.

And the fact that they didn't push the fact that they'll have 8 cores, and then they said "oh yeah, 4 cores are great" means that 8 cores won't be what they're pushing.

 

[GetSmart]

It is fair to say that Tiger Like is DEAD ON ARRIVAL.

Intel Tiger Lake performance is much stronger than Intel Ice Lake, on top of much higher clockspeeds as well (up to 4.8GHz) that contributed very high single core score (which even surpasses desktop CPUs). Thus likely Intel Tiger Lake-U would be tops in the 4C/8T category. And looking at the Cinebench scores even matches AMD Ryzen 5 4500U with 6C/6T. 😎

Hence my prediction that Intel Tiger Lake-H with 8 cores (and higher TDP due to being H-series part) would be twice faster than the recently launched Intel Tiger Lake-U. Additionally would be likely faster than AMD Ryzen 7 4800H as well, considering 4C/8T Intel Tiger Lke already can match 6C/6T AMD Renoir. Things will get even more interesting when LPDDR5 arrives... 🐯

Edit: That new Intel Xe integrated GPU now has discrete GPU level performance. Thus integrated GPU gaming may favor Intel Tiger Lake...

 

[PCWarrior]

IThat new Intel Xe integrated GPU now has discrete GPU level performance. Thus integrated GPU gaming may favor Intel Tiger Lake...

Yep. If you look at that Chinese review the performance in graphics benchmarks for the igpu of the 1065G7 is 50% higher than that of the 4800U and the MX150. It is even 17.2% higher than that of the MX350 and only 4.5-9.5% slower than new MX450.

 

[GetSmart]

Found another one https://notebook.cz/clanky/recenze-notebook/2020/11-generace-intel-core-acer-swift-5-sf514-55t

Also 50% to 60% faster in the integrated GPU department according to that review. Would be interesting to see how the H-series will perform with 32 EUs..

 

[spongiemaster]

Found another one https://notebook.cz/clanky/recenze-notebook/2020/11-generace-intel-core-acer-swift-5-sf514-55t

Also 50% to 60% faster in the integrated GPU department according to that review. Would be interesting to see how the H-series will perform with 32 EUs..

The cinebench R20 single core score of 583 is something else. That's not just slightly ahead, that's more than 10% higher than anything Intel or AMD is currently selling. When was the last time the per core performance crown was held by a low power mobile CPU?

 

[GetSmart]

The cinebench R20 single core score of 583 is something else. That's not just slightly ahead, that's more than 10% higher than anything Intel or AMD is currently selling. When was the last time the per core performance crown was held by a low power mobile CPU?

Now you've mentioned it, this may be probably the first time a low power mobile CPU core outperformed any desktop CPU core.

Years ago, probably Intel's Yonah (used in Intel Core Duo mobile CPUs) would be closest to compete against desktop CPUs https://techreport.com/review/9794/core-duo-on-the-desktop/ and https://www.anandtech.com/show/1900

 

[st379]

Latest and greatest from Intel marketing department :

1. Using industrial water chiller in a demonstartion.
https://www.tomshardware.com/news/intel-28-core-cpu-5ghz,37244.html

2. Using different gpu while claiming their cpus are better
https://www.techpowerup.com/268888/...-to-stake-gaming-performance-leadership-claim

3. Using different ram, os, TDP, cooling and storage in Tiger Lake review.

11th Generation Intel® Core™ Mobile Processors - 1 | Performance Index

edc.intel.com

4. Saying that Intel TDP in notebooks is accurate.

"Running Prime95 reveals more notable differences between the Ryzen 7 4800U and the competing Core i7-1065G7 or Core i7-10710U as shown by our graphs below. Power consumption would spike to 57.5 W on our Lenovo before falling and stabilizing at 49.6 W compared to 64.4 W and 42.8 W on the Ice Lake-powered XPS 13 2-in-1. " (notebbokcheck.net)

5. Saying that geekbench is more accurate than R20.

R20 support AVX instructions including the latest from Intel the AVX 512. It is very good benchmark to give you general idea in Adobe, handbrake, blender, 7-zip and other apps that support that instruction. Of course you should always look for the specific app that you are going to use to make sure that you are getting the performance you want.

On the other hand geekbench showed the Apple A12 (5 watt TDP?) better in multi threading than 300 watt cpu. Yes, sound legit to me.


<<Inappropriate content removed by moderator>>

 

[GetSmart]

Heck the strong single core score also shows up in Cinebench R15.

Compared to this (from https://www.anandtech.com/bench/CPU-2020/2772 )

And Cinebench R15 doesn't use AVX512...

 

[st379]

Heck the strong single core score also shows up in Cinebench R15.

Compared to this (from https://www.anandtech.com/bench/CPU-2020/2772 )

And Cinebench R15 doesn't use AVX512...

The moderator once again edit my post because I used the L word to describe these "benchmarks" in my last post......

Anyway back to the topic......

From the czech site using google translate:
"
Cinebench R20 - after 15 minutes of intense load

• Rendering (Single CPU): 577 cb
• Rendering (Multiple CPU): 1687 cb

This i7 can be clocked up to 4.7 GHz (20 W consumption) and held for five minutes. Then the clock will drop slightly to 4.6 GHz, reducing power consumption to an indefinitely sustainable 17 W. However, it's good to add that these six- and eight-core processors from AMD consume only 9 watts at a single-core load - that third of the power cost Intel twice the power consumption.
"
Not the best translation but the point is twice the power with far less than twice the score.
And really bad multi core score less than 4500u.

 

[GetSmart]

Results will vary depending on the cooling capability (and manufacturer's power settings) of the laptop being tested. That is a well known factor. The one from Chinese website probably had better cooling and more power headroom.

Most of those benchmark scores online are usually either on one run (best case) or average of multiple runs (typical case), and not after "15 minutes of intense load" which would have the laptop already heated up and having less thermal headroom. 🔥

The point here is that despite clockspeed disadvantage (compared to top desktop parts capable of 5.3GHz) and on a mobile platform (with typically less power headroom, limited cooling, higher memory latencies, etc), Intel's new Tiger Lake core shows a much higher IPC (ie. per clock) than all other current architectures from either Intel and AMD. 🤠

Thus imagine the performance of 8-cores in Intel's upcoming Tiger Lake H instead of just only 4-cores in current Intel Tiger Lake U chips...:

 

[st379]

Results will vary depending on the cooling capability (and manufacturer's power settings) of the laptop being tested. That is a well known factor. The one from Chinese website probably had better cooling and more power headroom.

Most of those benchmark scores online are usually either on one run (best case) or average of multiple runs (typical case), and not after "15 minutes of intense load" which would have the laptop already heated up and having less thermal headroom. 🔥

The point here is that despite clockspeed disadvantage (compared to top desktop parts capable of 5.3GHz) and on a mobile platform (with typically less power headroom, limited cooling, higher memory latencies, etc), Intel's new Tiger Lake core shows a much higher IPC (ie. per clock) than all other current architectures from either Intel and AMD. 🤠

Thus imagine the performance of 8-cores in Intel's upcoming Tiger Lake H instead of just only 4-cores in current Intel Tiger Lake U chips...:

Can't compare IPC with different clocks.
Intel use 20 watt Amd use 9 watt for single core. Another point that make it hard to compare.

I will give you that I am impressed with the 4.7 ghz for five minutes.

Maybe Intel should have aim for less IPC and smaller cores with better all cores boost and mt performance?

 

[GetSmart]

And on the desktop with far better cooling, typically higher clockspeeds can be maintained for longer time. Desktop CPUs also benefit from higher performance desktop DDR4 memory with lower latency. As most have already know, besides memory bandwidth, latency also affects performance.

IPC is simple, at the same clockspeed the CPU with the higher result has the better IPC. Likewise if a CPU had lower clockspeed with higher result then its IPC is better than another CPU at higher clockspeed but lower result. Lots of example from Intel's Conroe and later AMD's Bulldozer era. Thus roughly "relativeIPC=score/clockspeed"... 🆒

Perhaps Intel's newer Gracemont cores (rumored to include AVX and 30% higher IPC than previous generation Intel Atom lineage cores) in Intel Alder Lake may provide some extra performance alongside the main higher performance big cores, but I'm not a fan of big.LITTLE on the desktop...