Tiger Lake - Sample Review Discussion

[Dave2150]

Anandtech review of the i7-1185G7 has been released > https://www.anandtech.com/show/16084/intel-tiger-lake-review-deep-dive-core-11th-gen

This is a mobile 28w 4 core 8 thread CPU, on a 10nm+ process, which is clocking to 4.8Ghz.
Dr. Ian Cutress describes the single thread performance between Intel/AMD as so:

Against the x86 competition, Tiger Lake leaves AMD’s Zen2-based Renoir in the dust when it comes to single-threaded performance. "

Source: https://www.anandtech.com/show/16084/intel-tiger-lake-review-deep-dive-core-11th-gen/8



Rocket lake, the desktop equivalent of this CPU, should be very interesting. 4.8Ghz for a laptop chip isn't too bad, with a 28w power target. Imagine a 120 TDP version of this CPU's single target performance in games!

Considering Intel's 10th gen CPU's (which are using 5 year old Skylake and 5.5 year old 14nm) are already slightly faster than AMD's Zen2 CPU's, when Rocket Lake launches, the lead Intel take in gaming is going to be huge!

That said, I'm probably only going to upgrade when I can buy either an AM5 or LGA1700 platform next year, so it will be between Intel's Alder Lake and AMD's Zen4 in 2021, it will be interesting to see how Zen3 improves over Zen2 in a few months.

 

[kitfit1]

Mmmmmmmmm, perhaps you should have read the complete review before posting

"The new Tiger Lake stills falls down against the competition when we start discussing raw throughput tests. Intel was keen to promote professional workflows with Tiger Lake, or gaming workflows such as streaming, particularly at 28 W rather than at 15 W. Despite this we can easily see that the 15 W Renoir options with eight cores can blow past Tiger Lake in a like-for-like scenario in our rendering tests and our scalable workloads. The only times Intel scores a win is due to accelerator support (AVX-512, DP4a, DL Boost). On top of that, Renoir laptops in the market are likely to be in a cheaper price bracket than what Intel seems to be targeting."

 

[Dave2150]

Mmmmmmmmm, perhaps you should have read the complete review before posting

"The new Tiger Lake stills falls down against the competition when we start discussing raw throughput tests. Intel was keen to promote professional workflows with Tiger Lake, or gaming workflows such as streaming, particularly at 28 W rather than at 15 W. Despite this we can easily see that the 15 W Renoir options with eight cores can blow past Tiger Lake in a like-for-like scenario in our rendering tests and our scalable workloads. The only times Intel scores a win is due to accelerator support (AVX-512, DP4a, DL Boost). On top of that, Renoir laptops in the market are likely to be in a cheaper price bracket than what Intel seems to be targeting."

Hmm, perhaps you should have read where I'm excited about the single thread (and thus gaming performance) of Intel's desktop chips, based on the same cores?

I'd still buy a renoir CPU from AMD over the Intel chip, as it's clearly the better option for laptops when you consider the overall package. I'm just talking about the core single thread performance improvements, that will translate over to the upcoming desktop CPU's.

 

[FoxEye]

Rocket Lake still on 14nm?

 

[Journey]

Hmm, perhaps you should have read where I'm excited about the single thread (and thus gaming performance) of Intel's desktop chips, based on the same cores?

Not sure how many games make use of AVX-512.

 

[Dave2150]

Rocket Lake still on 14nm?

Yeah, but with the new architecture back ported to 14nm.

I'm guessing 20-30% more IPC over Comet Lake (which is still Skylake). All depends what clocks they can get to.

Alder Lake is where they'll likely completely leapfrog AMD in gaming, as that's gonna be 10nm, with the big and little core design (chiplets).

 

[Dave2150]

Not sure how many games make use of AVX-512.

? Even without AVX-512, these new CPU's have huge IPC increases over Skylake.

Remember, Comet Lake is still a 5 year old architecture (Skylake). We're going to see a really nice gaming performance uplift!

 

[FoxEye]

20-30% would be enormous, after years of single-digit gains per gen.

I think it's fair to be a bit sceptical until we see it in front of us

 

[Journey]

? Even without AVX-512, these new CPU's have huge IPC increases over Skylake.

Remember, Comet Lake is still a 5 year old architecture (Skylake). We're going to see a really nice gaming performance uplift!

Sorry to break this to you, buy Cypress Cove (14nm Willow Cove) is going to be frequency limited, so any IPC increase. which is roughly going to be 10% against Skylake due to node losses and cache redesign required to fit the core/threads into a reasonable die size and power envelope.

The is a likelihood that they could be pushing 180w+ if they try to get the frequency they are seeing now with 10th Gen. Once they get it onto 7nm in 2023 then I would expect a big uplift, but don't hold your breath until then.

 

[elina08]

"As an engineer, genuine clock-for-clock performance gains get me excited. Unfortunately Tiger Lake doesn't deliver much on this front, and in some cases, we see regressions due to the rearranged cache depending on the workload used. This metric ignores power - but power is the metric on which Tiger Lake wins. Intel hasn't really been wanting to talk about the raw clock-for-clock performance, and perhaps understandably so (from a pure end-user product point of view at any rate)."

https://www.anandtech.com/show/16084/intel-tiger-lake-review-deep-dive-core-11th-gen/19

 

[humbug]

Yeah, but with the new architecture back ported to 14nm.

I'm guessing 20-30% more IPC over Comet Lake (which is still Skylake). All depends what clocks they can get to.

Alder Lake is where they'll likely completely leapfrog AMD in gaming, as that's gonna be 10nm, with the big and little core design (chiplets).

You don't need to guess, Anand has the chip, they tested it. Its in your own link, this is their conclusion.

IPC improvements of Willow Cove are quite mixed. In some rare workloads which can fully take advantage of the cache increases we’re seeing 9-10% improvements, but these are more of an exception rather than the rule. In other workloads we saw some quite odd performance regressions, especially in tests with high memory pressure where the design saw ~5-12% regressions. As a geometric mean across all the SPEC workloads and normalised for frequency, Tiger Lake showed 97% of the performance per clock of Ice Lake.

In a competitive landscape where AMD is set to make regular +15% generational IPC improvements and Arm now has an aggressive roadmap with yearly +30% IPC upgrades, Intel’s Willow Cove, although it does deliver great performance, seems to be a rather uninspiring microarchitecture.

Its an overall regression in IPC.

The only reason it beat Zen 2 in single threaded is the 4.8Ghz vs 4.1Ghz of the 4800U.

IPC is way behind, Multicore performance is laughably behind.

A couple of examples from the same review... the 4800U is a 15 Watt 8 core CPU, Tigerlake at the same 15 Watts is LESS than half the performance. 15 Watt Intel's latest and greatest can't even get to half the performance of the old soon to be end of life Zen 2.

Tigerlake is junk. @Dave2150 its crap mate, Intel have lost it.... They are going backwards.

 

[KompuKare]

I thought that Anandtech article struck me as a deep dive in very restrictive sense.
Okay it is in laptop so we can't expect dGPU results, but it's almost like Intel PR set out terms like no Cinebench ST vs MT scores and similar things.

I don't SPEC scores aren't a great comparison tool.

 

[Freddie1980]

Meh the big news from the Tiger Lake demo was the performance of the Xe graphics chip not the CPU performance which only has 4 cores.

 

[humbug]

I thought that Anandtech article struck me as a deep dive in very restrictive sense.
Okay it is in laptop so we can't expect dGPU results, but it's almost like Intel PR set out terms like no Cinebench ST vs MT scores and similar things.

I don't SPEC scores aren't a great comparison tool.

Intel lose and lose badly in any real workload, so they have been trying to redefine what real is to something you or i have never seen, heard of, let alone use.

I Edit Videos, do 2D and 3D rendering as a hobby, compile code..... all the things Intel get crushed in.

So Intel dig out synthetic benchmarking tools from 2006 that use ancient instruction sets no one uses anymore and are only available in Intel CPU's.... and then say don't look at all that, don't look at the benchmarks that are relevant to your use case, look at this synthetic obscure junk we have over here.

 

[KompuKare]

SPEC benches are common for server comparison, far less so for consumer workloads.
However, AFAIK aren't they distributed as source code which you have compile:
"As the SPEC benchmarks are distributed as source code, it is up to the party performing the test to compile this code. "
From the Wikipedia
https://en.m.wikipedia.org/wiki/SPECint
Okay, well we all know Intel are good at using their compilers to win benchmarks.
Which is mostly legit fur SPEC, but doesn't tell us much except that the Intel compiler is good at generating code which runs well on their latest CPU, probably uses avx512 heavily, and is probably not representative of most programs.
Hey, if Intel wanted most code to use AVX512 as heavily as possible they should not have segmented the market on AVX feature set for the last decade and had been willing to spend the transistors even on Celeron and Pentium.

 

[Davedree]

A couple of examples from the same review... the 4800U is a 15 Watt 8 core CPU, Tigerlake at the same 15 Watts is LESS than half the performance. 15 Watt Intel's latest and greatest can't even get to half the performance of the old soon to be end of life Zen 2.

Tigerlake is junk. @Dave2150 its crap mate, Intel have lost it.... They are going backwards.

Unfortunately the Yoga slim 7 is not a 15w configuration that you are implying.
It uses a lot more than 15w, with 3 stages of power targets vs thermal limits, that's not to say that the Intel doesn't consume more either. Not disputing Amd's performance over intel, Amd are all over Intel it's fantastic bar they can't manufacture enough of these beauties.

Coming from a r5 2500u and recently R3 4300U owner.

 

[Destination]

Hmm, perhaps you should have read where I'm excited about the single thread (and thus gaming performance) of Intel's desktop chips, based on the same cores?
.

Very odd results.
Many of those results have it beating the intel flagship desktop processor at 5.3GHz.
Did they downvolt it to 28 watts also?

 

[Destination]

Intel lose and lose badly in any real workload, so they have been trying to redefine what real is to something you or i have never seen, heard of, let alone use.

I Edit Videos, do 2D and 3D rendering as a hobby, compile code..... all the things Intel get crushed in.

So Intel dig out synthetic benchmarking tools from 2006 that use ancient instruction sets no one uses anymore and are only available in Intel CPU's.... and then say don't look at all that, don't look at the benchmarks that are relevant to your use case, look at this synthetic obscure junk we have over here.

Shrouting?

 

[Boot_Loop]

This is one of the worst threads I've seen in a while. Tigerlake, far from having 'huge' IPC gains, actually has on average LESS IPC than Ice Lake. It's a terrible release.

Not only that, it's 10nm, desktop Rocket Lake is still, you guessed it, 14nm...

 

[Grim5]

Not sure how many games make use of AVX-512.

probably none

avx-512 is a power draw black hole and should be banned