Benchmark tests show new iPad Pro models outperform MacBook Pro in some CPU & GPU tasks

[Drak3]

1 hour ago, Droidbot said:

want a fucking Z Canvas refresh so bad. Just. Add. TB3

I wouldn't mind a refresh either. I'd like to see in i5hq/hk variant to them, and tb3 without ditching other IO would be nice. A desktop counterpart would be cool too.

[Doobeedoo]

Yeah geekbench everywhere lately for everything upcoming...

[crystal6tak]

On 6/16/2017 at 0:53 PM, Đỗ Đức Huy said:

Sees Geekbench

"Fake bench more like it"

Basically it have a very unusual way of weighing score. Too high emphasis on security hash functions which most SOCs have dedicated ICs for, while General CPUs don't.

 It's like taking an Ant miner and showing its bench in Hashing compared to an 6950x and then claiming it to be more powerful than the Extreme edition.

GeekBench should die in a fire

Care to elaborate?

 

On their documentation cryptography workloads only contain AES, which I'm pretty sure intel can do with hardware acceleration (AES-NI). Besides, the documentation also states cryptography weighing to be 5% only.

[Zodiark1593]

On 6/16/2017 at 0:51 AM, Coaxialgamer said:

I know this isn't really related to the topic but i had to say something. 

X86 isn't a CISC core. Hasn't been since the pentium pro. It's a RISC core with a CISC wrapper to maintain compatibility with 8086,286,386 and 486. 

Plus, the whole CISC vs RISC argument hasn't been relevant since the 90's. They are design philosophies more than anything, not indicators of performance. 

Very much agree here, two differing means of achieving the same goal. 

 

Also, does anyone know whether Geekbench leverages newer instruction sets on both ARM and x86? While x86 has progressed slowly as far as legacy code, the big gains are to be had with AVX(2) and FMA3. To not have both processors operating at their best would hardly be called a comparison of their capability.