• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Samsung Exy 7420 Benchmarks Reveals Superfast Mobile Chip

Status
Not open for further replies.
A

ARMD Supremacy

Member
The Samsung Exynos 7420 chipset has been tested in the Geekbench 3 benchmark, likely residing in a prototype Galaxy S6 unit. The SoC aced the benchmark test scoring 5478 and 1520 points in the multi-core and single-core tests, respectively.
Click to expand...

First benchmarks of Samsung Exynos 7420 64-bit 14nm (unconfirmed) chip as benchmarked and compared to new Intel Cherryview 14nm mobile:

http://browser.primatelabs.com/geekbench3/compare/1754567?baseline=1780313

Intel Cherryview Single: 837 Samsung Exynos 7420: 1520

Intel Cherryview Multi: 2817 Samsung Exynos Multi: 5478


http://www.gsmarena.com/samsung_exynos_7420_aces_geekbench_30-news-10921.php

Duplicate posts are not allowed.
Markfw900
 
Last edited by a moderator:
Not as good as I was hoping considering Apple will undoubtedly have a new SoC ready soon as well.

Also, this seems to coincide with the 1.2x boost ARM predicted for their A57 running in 64-bit workloads (vs 32-bit). The Exynos 5433 could score around 1250 to 1300 points in single thread tests.
 
Afraid to compare to Broadwell, aren't you? Atom is still in its infancy: the architecture you're comparing your Samsung chip with is from 2013 and only the first iteration of Atom.
 
witeken said:
Afraid to compare to Broadwell, aren't you? Atom is still in its infancy: the architecture you're comparing your Samsung chip with is from 2013 and only the first iteration of Atom.
Click to expand...

To be fair, it kills Core M in multi-core. It actually scores the same as many of the u series of Core. Of course, a lot of that is due to the SHA score. And I don't actually believe these processors perform anywhere near Core M or an I5 u series.

That being said, no doubt this thing crushes Bay Trail. As for Cherryview, I'm going to assume that bench is no where near where a final processor would be. I mean, it actually scores lower than the Z3770.
 
To be fair, Cherryview atom is the latest iteration of that chip, and is on 14nm. And I would hardly say atom is in its "infancy". More accurately is has just suffered severe neglect until the last couple of years.

That said though, I view all these comparisons of x86 and ARM directly with a grain of salt. I have always said it will be difficult for intel to compete directly with ARM on apps that dont use x86. They need to take advantage of the ability to run x86 in order to really compete. All the delays and somewhat (so far at least) disappointing performance of 14nm has not helped either.
 
witeken said:
Afraid to compare to Broadwell, aren't you? Atom is still in its infancy: the architecture you're comparing your Samsung chip with is from 2013 and only the first iteration of Atom.
Click to expand...
Atom has been with us since 2008. It is at least the second major micro architecture. Atom has been using three nodes without counting 14nm. At that time ARM had just announced Cortex-A9. Want to compare how Atom and ARM chips evolved since then and that despite Intel process advantage?

Please don't make the typical Intel fan answer that back then Intel wasn't serious 😉
 
Nothingness said:
Atom has been with us since 2008. It is at least the second major micro architecture. Atom has been using three nodes without counting 14nm. At that time ARM had just announced Cortex-A9. Want to compare how Atom and ARM chips evolved since then and that despite Intel process advantage?

Please don't make the typical Intel fan answer that back then Intel wasn't serious 😉
Click to expand...

ARM has really been impressive here. In the time it has taken to go from Silvermont to Airmont, ARM has put out:

- Cortex A57
- Cortex A17
- Cortex A53
 
I wonder if the 14nm process that samsung is using is just that good. The 5433 was in Apple A7 territory in single thread even on 20nm, but this is equal to A8 in single and demolishes it in multicore.


That multicore score... wow. I was impressed with A8X multicore but this is just craaazy.
 
waffleironhead said:
TIL a chip series from 2008 is still in its infancy.
Click to expand...

It was originally an architecture from 2004 put on a small process node to make cheap and slow netbooks. It has been put on a 5 year release cadence by Paul Otellini, so the first iteration was 2013. Sure, it's got some platform changes and after netbooks it was put into tablets and smartphones, but nothing to write home about. What's worse, it that its successor was also meant for cheap laptops, and not so much for tablets.

You can't call platform that has a BOM delta and compensation associated with it and only 1 major silicon iteration mature.

Just look at how long it's token Nvidia to get a decent chip out of Tegra.
 
mikk said:
This CherryView comparison is such a nonsense.
Click to expand...

Best case Geekbench 3 score for the production Cherryview SoC 2.7GHz) is probably ~1100 ST score.

Still not really that great compared to an ARM A57 at 2GHz+.

That said, taking a single benchmark in isolation is never wise or really all that useful.
 
Arachnotronic said:
Best case Geekbench 3 score for the production Cherryview SoC 2.7GHz) is probably ~1100 ST score.

Still not really that great compared to an ARM A57 at 2GHz+.

That said, taking a single benchmark in isolation is never wise or really all that useful.
Click to expand...

Airmont is a dual-issue architecture. No surprises there.
 
witeken said:
Airmont is a dual-issue architecture. No surprises there.
Click to expand...

I think Intel really needs to up the ante with Goldmont. It's not enough for Intel even to be on par with the ARMy, especially given its vaunted process lead, it needs to be unequivocally better.

Right now, Silvermont performs at best on-par with an Exynos 5433 (3DMark Physics) and at worst well below it (Geekbench).

Now, don't get me wrong: Silvermont is a good architecture, but Intel delivered SoCs based on it quite late to market. Silvermont should have been an early-to-mid 2013 deal in both tablets and phones, and Intel should have had 14nm parts by mid-2014 if it truly wanted to capture the performance lead in mobile.
 
Arachnotronic said:
I think Intel really needs to up the ante with Goldmont. It's not enough for Intel even to be on par with the ARMy, especially given its vaunted process lead, it needs to be unequivocally better.
Click to expand...

Do you think Goldmont has been delayed to 2016 to make some competitive changes?
 
witeken said:
Do you think Goldmont has been delayed to 2016 to make some competitive changes?
Click to expand...

The Goldmont is set in stone at this point; making modifications to a pre-existing core design is extremely difficult, so Intel had better hope it made the right architectural decisions there.

Where Intel can make modifications is on the SoC level. They could swap in other IPs, add in other IPs if those IP blocks are already done and validated.
 
Status
Not open for further replies.

TRENDING THREADS

Back
Top