Intel Rocket Lake-S CPU Benchmarked: Up to 22% Faster Compared to the Previous Generation

[owen10578]

userbenchmark...nuff said. take this with an ocean of salt.

 

[londiste]

userbenchmark...nuff said. take this with an ocean of salt.

That is a very short-sighted statement. Their benchmark results have been rather solid on almost all fronts. All the controversy is about interpreting and presenting them.

 

[mastrdrver]

That is a very short-sighted statement. Their benchmark results have been rather solid on almost all fronts. All the controversy is about interpreting and presenting them.

The professionals say otherwise:

Userbenchmark adjusting CPU 'rankings', pandering to Intel?

Userbenchmark changes their CPU rating algorithm to push huge focus on 1T performance while nerfing nT performance because 'Ryzen is too fast'

www.tweaktown.com

UserBenchmark offers explanation for changes to CPU score weights

This past week a portion of the enthusiast community was lit aflame with controversy over UserBenchmark's decision to adjust its CPU scoring system. While not a popular...

www.techspot.com

Updated: Userbenchmark Responds to Criticism Over Score Weighing Revisions

Userbenchmark's new scoring mechanism has stoked controversy in the enthusiast community.

www.tomshardware.com

UserBenchmark Adjusts Algorthym to Favor Single-Thread Performance, Causing Strange Results

UserBenchmark.com has adjusted the algorithm for how the overall performance of a CPU is calculated. Previously, single-core performance made up 30% of the result, quad-core made up 60%, and the final 10% was reserved for "multicore" performance. This was calculated so that an Intel Core i7-7700K...

linustechtips.com

Even the Intel subreddit has banned Userbenchmark as a source.

https://www.reddit.com/r/intel/comments/g36a2a/userbenchmark_has_been_banned_from_rintel/

Their scoring is unreliable and as independent and trustworthy as a suggestion from a fanboy.

 

[Athlonite]

I thought Intel didn't like benchmarks claiming they were an unreliable way to measure real world performance but here they are using an intel favouring benchmark program and trying to tell us gee look at all our newer faster whizbang cpu speed yeah fuck intel off ya hypocrites either use real world aps or GTFO you can't have it both ways and expect us to believe a word your saying

 

[okbuddy]

only mobile still?

 

[londiste]

The professionals say otherwise:

Userbenchmark adjusting CPU 'rankings', pandering to Intel?

Userbenchmark changes their CPU rating algorithm to push huge focus on 1T performance while nerfing nT performance because 'Ryzen is too fast'

www.tweaktown.com

UserBenchmark offers explanation for changes to CPU score weights

This past week a portion of the enthusiast community was lit aflame with controversy over UserBenchmark's decision to adjust its CPU scoring system. While not a popular...

www.techspot.com

Updated: Userbenchmark Responds to Criticism Over Score Weighing Revisions

Userbenchmark's new scoring mechanism has stoked controversy in the enthusiast community.

www.tomshardware.com

UserBenchmark Adjusts Algorthym to Favor Single-Thread Performance, Causing Strange Results

UserBenchmark.com has adjusted the algorithm for how the overall performance of a CPU is calculated. Previously, single-core performance made up 30% of the result, quad-core made up 60%, and the final 10% was reserved for "multicore" performance. This was calculated so that an Intel Core i7-7700K...

linustechtips.com

Even the Intel subreddit has banned Userbenchmark as a source.

https://www.reddit.com/r/intel/comments/g36a2a/userbenchmark_has_been_banned_from_rintel/

Their scoring is unreliable and as independent and trustworthy as a suggestion from a fanboy.

Now, please read again what I wrote and then the links.
I said results vs interpreting and presenting them. Links talk about CPU rankings and CPU score weights - all about interpretation and presentation.
From all the controversies I do not remember critique about the results themselves.

Basically when you run that benchmark the results you get for 1-core, 2-core, 4-core, 8-core and 64-core have been the same all this time, never changed and that was never the problem.

only mobile still?

Rocket Lake is desktop, LGA1200, Q1-2021 (March, probably).

 

[Mouth of Sauron]

That is a very short-sighted statement. Their benchmark results have been rather solid on almost all fronts. All the controversy is about interpreting and presenting them.

Well, so far Intel wasn't exactly representing them good. Maybe they are representative in general, maybe not.

All the following hardware needs good and solid 3rd party testing, apart from producers claim, too:
- Zen3 (this we'll hopefully get very soon - days, I hope)
- RDNA2
- 3600 Ti
- Intel Xe (all variants)

 

[Dave65]

Is that with the chiller hidden behind the wall on or off?

 

[TheLostSwede]

Is that with the chiller hidden behind the wall on or off?

Chiller? No, no, that's for noobs, this is on liquid Helium...

 

[Bytales]

179 / 4.2 = 42.619
152 / 5.3 = 28.679

equals to 48.6% IPC improvement.

I guess the research days in area 51 finally pays off.

Only goes to show what could have been possible a couple of generations ago. They made this move now, only because AMD came strong from the back. Otherwise you would have gotten your 1 to 3% perfomance improvement between generations. This is the reason i despise Intel and NVIDIA. theyre only purpose is to milk the plebs. not to advance technology or whatever theyr trying to sell us. No, pure milking as usual.

See the example in GPU. AMD made a GPU that they will sell for 999USD and has the same performance as the Nvidia 1500 dolar gpu. And please remeber they are going to make profit even selling the cheap 999 dolars cpu. whats above that is simply greed. Which is even more compeling on those multi thousand enterprise gpus, when they should cost but a couple of hundred dollars at most.

Capitalism at its best. If we want this trend to dissapear, wed need to make money dissapear. Just like in star trek. Otherwise this will alwayshappen

 

[ahenriquedsj]

Even the 5600x is faster than the 10700k LOL

 

[LemmingOverlord]

Is it just me or is 1GB of system RAM reserved for Xe graphics?

 

[efikkan]

179 / 4.2 = 42.619
152 / 5.3 = 28.679
equals to 48.6% IPC improvement.
I guess the research days in area 51 finally pays off.

It doesn't work that way.
The clock speed needs to be locked (and far away from power limits) in order to approximate IPC. Reported clock speeds of engineering samples are all over the place, and even for retail products rated clock speeds are not actual clock speeds.

Intel was talking about double digits IPC gains, this is around 50% more. This sounds highly unlikely, otherwise Intel would of course hinted to something like this and not to 10%+.

Rocket Lake is likely a Sunny Cove derivative, which probably will result in IPC gains at ~18%. (unless it has improvements beyond Sunny Cove)

AMD Drivers are still crap

My primary OS is Debian:
CPU = sometimes it cant boot, sometimes the OS crash if the SSD is on the Asmedia Chip
GPU = Vulcan dont work right, RX 5700 have a Hardwarebug with Audio via HDMI

There is only one reason to switch from Intel/Nvidia to AMD, u wanna pay a lot of cash to be a Beta tester.

Unfortunately, this is true for Linux support, and is why my next workstation will be Intel/Nvidia based.
But I'll probably get an AMD system too, since I'm down one machine. But I'll wait for good reviews of motherboards, as I don't have time to be a beta tester.

 

[mastrdrver]

Now, please read again what I wrote and then the links.
I said results vs interpreting and presenting them. Links talk about CPU rankings and CPU score weights - all about interpretation and presentation.
From all the controversies I do not remember critique about the results themselves.

Actually they do say it's about the tests:
"Here you see that the previous rating was split with 60% of the score being tests of quad-core level performance, while the single-core comes in at 30% and higher than quad-core labeled 'multi-core' being 10%. Now we move right and find that the scoring has been adjusted as follows. Quad-core level performance drops to 58%, single-core jumps to 40%, while the multi-core falls to a mere 2%. This would make sense several years ago when frequencies were still growing, but now that it is increasingly more difficult to increase frequencies, scaling in cores has been having over the past 3+ years. I guess you could say that since Ryzen came onto the scene it has started the shift to many more cores on consumer-level CPUs."

If the results are biased, what is there to interpret? How it's not biased?

 

[londiste]

Actually they do say it's about the tests:
"Here you see that the previous rating was split with 60% of the score being tests of quad-core level performance, while the single-core comes in at 30% and higher than quad-core labeled 'multi-core' being 10%. Now we move right and find that the scoring has been adjusted as follows. Quad-core level performance drops to 58%, single-core jumps to 40%, while the multi-core falls to a mere 2%. This would make sense several years ago when frequencies were still growing, but now that it is increasingly more difficult to increase frequencies, scaling in cores has been having over the past 3+ years. I guess you could say that since Ryzen came onto the scene it has started the shift to many more cores on consumer-level CPUs."

If the results are biased, what is there to interpret? How it's not biased?

Test result is from single-core performance or quad-core performance or other tests, represented by specific measured number. These have not changed.
As you point out, aggregation has changed, the way they put together their own performance index or whatever they call it these days. This is interpretation and presentation.
The results of individual tests are not biased.

 

[Makaveli]

That is not a correct way to calculate IPC, yeah it scores higher results at lower clocks, however, we have to wait for benchmarks to see how much IPC increase is there actually.

Plus take into account that this may be a heavily optimized benchmark so not a real-world performance representation.

Not to mention don't take anything seriously from userbenchmark.

I will wait for an offical review in march 2021.

 

[londiste]

Not to mention don't take anything seriously from userbenchmark.

Again, very short-sighted statement.
You obviously did not understand what the problem with Userbenchmark was or what can or cannot be taken seriously from them.
Userbenchmark has been a pretty valuable resource for leaks and their measured numbers are not bad.

 

[Makaveli]

Again, very short-sighted statement.
You obviously did not understand what the problem with Userbenchmark was or what can or cannot be taken seriously from them.
Userbenchmark has been a pretty valuable resource for leaks and their measured numbers are not bad.

There is nothing you are going to post that will change my opinion on this.

Mastrdrver links which I read before he posted them speaks volumes on this topic.

If you want to use them all the power to you. I will wait for official reviews.

 

[EarthDog]

And what cooling solution are they using for that increase? What is the tdp? Unless we see that the rest is quite meaningless.

Air or an AIO...

Again, very short-sighted statement.
You obviously did not understand what the problem with Userbenchmark was or what can or cannot be taken seriously from them.
Userbenchmark has been a pretty valuable resource for leaks and their measured numbers are not bad.

Userbenchmark is MEH at best bud. Even with all the other nonsense Mak is talking about. I haven't touched it since a shit thread popped up here trying to use it, but I don't recall it showing actual clockspeeds and it was tough to compare. Also, results were not consistent in some things from run to run.

I'd stay away from it. Perhaps it's good in horseshoes, hand grenades, and nuclear warfare...but not sniping.

 

[londiste]

Userbenchmark is MEH at best bud. Even with all the other nonsense Mak is talking about. I haven't touched it since a shit thread popped up here trying to use it, but I don't recall it showing actual clockspeeds and it was tough to compare. Also, results were not consistent in some things from run to run.

Well, that is a reasonable standpoint. I don't exactly agree but at least you are judging this on merits (or flaws) of benchmark itself

 

[Shatun_Bear]

Still on 14nm in 2021....

The power draw and heat of these is going to be just as horrible as Comet Lake.

 

[EarthDog]

Well, that is a reasonable standpoint. I don't exactly agree but at least you are judging this on merits (or flaws) of benchmark itself

People also don't agree the Earth is round...

...just saying.

 

[Makaveli]

Still on 14nm in 2021....

The power draw and heat of these is going to be just as horrible as Comet Lake.

That is why they are only going up to 8 c / 16 T.

And even if Rocket lake has equal or greater performance to Zen 3. They won't have an answer for the 12 and 16 cores parts.

 

[EarthDog]

That is why they are only going up to 8 c / 16 T.

And even if Rocket lake has equal or greater performance to Zen 3. They won't have an answer for the 12 and 16 cores parts.

They are going backwards? There is a 10c/20t part already on the mainstream from intel...

 

[mastrdrver]

Test result is from single-core performance or quad-core performance or other tests, represented by specific measured number. These have not changed.

But they have by the very quote I used in my last post.

Why are you trying to make a distinction where there is not one?