AMD Raven Ridge 'Zen APU' Thread

Page 59 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
PeterScott

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
mikk said:
It's slower.

3657 vs 4347 points. Keep in mind your comparison is x32 vs x64 of Windows.
Click to expand...

I was talking about the Android score (4451), hence why I was wondering if the Mobile version of Geekbench was different from the desktop version.
 
DrMrLordX

DrMrLordX

Lifer
Apr 27, 2000
22,741
12,728
136
Only concern I would have about Raven Ridge Chromebooks is that, typically, Android apps with native code require instruction set translation to run on x86 CPUs, which degrades performance. Which is why Intel Android tablets were never as good as they could be.

Though I think that if Google gives their direct blessing to x86 by using it in their branded products, that it may steer a lot of app developers to switch to native x86 instructions instead of native ARM/NEON instructions.
 
NewFatMike

NewFatMike

Junior Member
Jun 24, 2017
22
1
11
I mean the x86 Chromebooks seem to be working fine on the apps front. There has been some hubbub about the ARM based Samsung Chromebook Plus having marginally better Android app performance than the x86/Intel chipped Chromebook Pro, but the standard ChromeOS functions are better on the Pro.

It's probably not enough difference to make the daily use bat an eye, though.

Sent from my Pixel using Tapatalk
 
R

Ryzenenthusiast

Member
Feb 19, 2017
40
63
51
The Android 7.1.2 Geekbench 4 results of Raven Ridge are highly probably Chromebooks. If you check some Chromebook Geekbench results, they are also being reported as Android 7.1.2.

The thing I wonder is; There was an in-company event of AMD couple of days ago. Anyone who is planning to leak the slide deck? :)
 
S

Spartak

Senior member
Jul 4, 2015
353
266
136
here you can see the AMD 2700U Android vs Windows: http://browser.geekbench.com/v4/cpu/search?utf8=✓&q=AMD+Ryzen+7+2700U+

Single core score is 3495 on Windows (seems in line with Ryzen scores) vs 4451/4323 on Android. That's about 25% performance difference cross platform. Makes me wonder how much further optimised iOS is. Wouldn't be surprised if that platform is optimised another 20% vs Android, if not more. Unfortunately we cannot compare, but it makes all those "A11 IPC outperforms Kaby Lake" claims rather suspicious.
 
DrMrLordX

DrMrLordX

Lifer
Apr 27, 2000
22,741
12,728
136
Spartak said:
it makes all those "A11 IPC outperforms Kaby Lake" claims rather suspicious.
Click to expand...

Maybe so, which is why we should have some relatively OS-agnostic tests run on A11 to learn more about that. I'm still looking for SPEC numbers (lulz).

But hey, different day, different thread . . .
 
L

Lodix

Senior member
Jun 24, 2016
340
116
116
DrMrLordX said:
Maybe so, which is why we should have some relatively OS-agnostic tests run on A11 to learn more about that. I'm still looking for SPEC numbers (lulz).

But hey, different day, different thread . . .
Click to expand...
There are SPEC benchmarks for some Android SOCs and they represent a similar result to Geekbecnh 4. The info of the benchmark is open to take a look.

Edit: there is a SPEC benchmark of the iPad pro with the A9X.

https://www.anandtech.com/show/9766/the-apple-ipad-pro-review/4
 
S

Spartak

Senior member
Jul 4, 2015
353
266
136
Lodix said:
There are SPEC benchmarks for some Android SOCs and they represent a similar result to Geekbench 4. The info of the benchmark is open to take a look.

Edit: there is a SPEC benchmark of the iPad pro with the A9X.

https://www.anandtech.com/show/9766/the-apple-ipad-pro-review/4
Click to expand...


What do you mean with similar results? If you mean Geekbench is platform agnostic like SPEC, then it can't be similar results.

For example, the Core m3-6Y30 had an average SPECint_base2006 of +67% vs the A9X clock for clock in that review. The Geekbench result of the same m3-6Y30 is about ~2700 vs ~3070 or -15%. that's a performance gap of *1,9, or rather: geekbench results for x86 processors vs Ax are almost half the specint results.
 
Last edited:
L

Lodix

Senior member
Jun 24, 2016
340
116
116
Spartak said:
What do you mean with similar results? If you mean Geekbench is platform agnostic like SPEC, then it can't be similar results.

For example, the Core m3-6Y30 had an average SPECint_base2006 of +67% vs the A9X clock for clock in that review. The Geekbench result of the same m3-6Y30 is about ~2700 vs ~3070 or -15%. that's a performance gap of *1,9, or rather: geekbench results for x86 processors vs Ax are almost half the specint results.
Click to expand...
That was Geekbench 3. Anyway just take a look of the SPECint and now take in mind that from A9 to A11 had been a huge IPC increase and you get the proof that Apple has similar IPC to Intel right now.
 
PeterScott

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Spartak said:
here you can see the AMD 2700U Android vs Windows: http://browser.geekbench.com/v4/cpu/search?utf8=✓&q=AMD+Ryzen+7+2700U+

Single core score is 3495 on Windows (seems in line with Ryzen scores) vs 4451/4323 on Android. That's about 25% performance difference cross platform. Makes me wonder how much further optimised iOS is. Wouldn't be surprised if that platform is optimised another 20% vs Android, if not more. Unfortunately we cannot compare, but it makes all those "A11 IPC outperforms Kaby Lake" claims rather suspicious.
Click to expand...

Another problem with browsing geekbench results is that they are highly inconsistent. The same exact setup will show different results by up to a factor of two, or even more.
 
S

Spartak

Senior member
Jul 4, 2015
353
266
136
PeterScott said:
Another problem with browsing geekbench results is that they are highly inconsistent. The same exact setup will show different results by up to a factor of two, or even more.
Click to expand...

There are multiple explanations for that: mostly ultrabooks vary wildly, and that's due to thermal throttling. So you should really only look for the higher results. With desktops, the higher results are mostly due to overclocking. The really low results are mostly showing PC's with issues.

So it's not that inconsistent when you look past those factors, the higher end average should be normative.
 
S

Spartak

Senior member
Jul 4, 2015
353
266
136
Lodix said:
That was Geekbench 3. Anyway just take a look of the SPECint and now take in mind that from A9 to A11 had been a huge IPC increase and you get the proof that Apple has similar IPC to Intel right now.
Click to expand...

No I compared Geekbench 4 results. The specint results were from the article. The skew on geekbench 3 is even higher, with x86 scoring lower and Apple's Ax chips scoring higher.
 
AtenRa

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
https://www.techpowerup.com/237900/hp-envy-x360-15-bq101na-could-be-first-raven-ridge-implementation

HP ENVY x360 15-bq101na Could be First "Raven Ridge" Implementation

rnQQn1eqsw4PVpiK.jpg
 
  • Like
Reactions: Drazick and lightmanek
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom