Please anandtech test Spec2006 on A8X

[tempestglen]

http://www.anandtech.com/show/8554/the-iphone-6-review/3

For our first high level benchmark we turn to SPECint2000. Developed by the Standard Performance Evaluation Corporation, SPECint2000 is the integer component of their larger SPEC CPU2000 benchmark. Designed around the turn of the century, officially SPEC CPU2000 has been retired for PC processors, but with mobile processors roughly a decade behind their PC counterparts in performance, SPEC CPU2000 is currently a very good fit for the capabilities of Cyclone and Enhanced Cyclone.
SPECint2000 is composed of 12 benchmarks which are then used to compute a final peak score. Though in our case we’re more interested in the individual results.
SPECint2000 - Estimated Scores A8 A7 % Advantage 164.gzip

842​

757​

11%​

175.vpr

1228​

1046​

17%​

176.gcc

1810​

1466​

23%​

181.mcf

1420​

915​

55%​

186.crafty

2021​

1687​

19%​

197.parser

1129​

947​

19%​

252.eon

1933​

1641​

17%​

253.perlbmk

1666​

1349​

23%​

254.gap

1821​

1459​

24%​

255.vortex

1716​

1431​

19%​

256.bzip2

1234​

1034​

19%​

300.twolf

1633​

1473​

10%​

Keeping in mind that A8 is clocked 100MHz (~7.7%) higher than A7, all of the SPECint2000 benchmarks show performance gains above and beyond the clock speed increase, indicating that every benchmark has benefited in some way. Of these benchmarks MCF, GCC, PerlBmk and GAP in particular show the greatest gains, at anywhere between 20% and 55%. Roughly speaking anything that is potentially branch-heavy sees some of the smallest gains while anything that plays into the multiplication changes benefits more.
MCF, a combinatorial optimization benchmark, ends up being the outlier here by far. Given that these are all integer benchmarks, it may very well be that MCF benefits from the integer multiplication improvements the most, as its performance comes very close to tracking the 2X increase in multiplication throughput. This also bodes well for any other kind of work that is similarly bounded by integer multiplication performance, though such workloads are not particularly common in the real world of smartphone use.

Now A8X has 2G memory and satisfy the requirement of Spec2006.

Please anandtech give us a gift!:biggrin:

 

[FwFred]

...

 

[raghu78]

yeah a TK1 Denver (28nm) vs A8X (20nm) vs Core M (14nm) comparison would be very helpful. With 16FF+ I am sure Apple A9X will get close to Core M on perf and power efficiency. Nvidia's next gen Erista also could really pack a lot of performance.

 

[asendra]

yeah a TK1 Denver (28nm) vs A8X (20nm) vs Core M (14nm) comparison would be very helpful. With 16FF+ I am sure Apple A9X will get close to Core M on perf and power efficiency. Nvidia's next gen Erista also could really pack a lot of performance.

If rumors hold, Apple will be skipping 16FF+ and use directly 14FF with Samsung fabs.

http://www.imore.com/samsung-reportedly-providing-apple-14nm-apple-a9-chips-iphone-7-2015

 

[ancientarcher]

If rumors hold, Apple will be skipping 16FF+ and use directly 14FF with Samsung fabs.

http://www.imore.com/samsung-reportedly-providing-apple-14nm-apple-a9-chips-iphone-7-2015

Practically 16FF (from TSMC) and 14FF (from Samsung, also used by GloFo) is the same thing. TSMC uses the 16nm instead of 14nm designation because of the chinese association of 4 (and 14, 24, 34, you get the picture) with really bad things. 4 is the chinese version of the western aversion to 13

so, you can't 'skip' 16FF+ to go to 14FF. Firstly, they are equivalent technologies offered by different companies. And secondly, 16FF+ comes a year after (in 2016) 14FF (in 2015)

 

[asendra]

Practically 16FF (from TSMC) and 14FF (from Samsung, also used by GloFo) is the same thing. TSMC uses the 16nm instead of 14nm designation because of the chinese association of 4 (and 14, 24, 34, you get the picture) with really bad things. 4 is the chinese version of the western aversion to 13

so, you can't 'skip' 16FF+ to go to 14FF. Firstly, they are equivalent technologies offered by different companies. And secondly, 16FF+ comes a year after (in 2016) 14FF (in 2015)

Yeah, I imagined they were about the same tech just name differently, though I didn't know that was the reason, I was thinking more along the lines of marketing reasons by Samsung because 14 is less than 16.

By skipping I meant they seem to have decided to not wait for TMSC to have it ready (and repeat 20nm next year?) and instead go with Samsung and their version of it (14FF). Sorry I wasn't more clear, not my first language!

 

[Nothingness]

http://www.anandtech.com/show/8554/the-iphone-6-review/3



Now A8X has 2G memory and satisfy the requirement of Spec2006.

Please anandtech give us a gift!:biggrin:

If this is to read such thing as:

MCF, a combinatorial optimization benchmark, ends up being the outlier here by far. Given that these are all integer benchmarks, it may very well be that MCF benefits from the integer multiplication improvements the most, as its performance comes very close to tracking the 2X increase in multiplication throughput. This also bodes well for any other kind of work that is similarly bounded by integer multiplication performance, though such workloads are not particularly common in the real world of smartphone use.

The analysis is completely wrong: in 181.mcf the number of integer mul is close to 0 (<3M out of >45B instructions on AArch64). mcf is heavy on data cache and TLB. What we are seeing here are improvements to the memory subsystem in general.

 

[Phynaz]

Practically 16FF (from TSMC) and 14FF (from Samsung, also used by GloFo) is the same thing. TSMC uses the 16nm instead of 14nm designation because of the chinese association of 4 (and 14, 24, 34, you get the picture) with really bad things. 4 is the chinese version of the western aversion to 13

Not correct.

I live in Taiwan (republic of china) and the whole "chinese people hate/avoid the number 4" thing is way overdone in terms of western impressions in my opinion.

Our buildings have 4th floors, we use coins and make change that tallies up to increments of 4's (4, 40, 44, 400, etc.) absolutely no one avoids the number "4" as if it were the plague, contrary to the way it is portrayed and captured in some westernized impressions of Asian culture.

We don't skip the 4th step when climbing stairs or ladders, we don't shun the 4th room on a given floor, etc. etc.

I can absolutely guarantee you not a single person here gives two shts whether their mobile phone has 3 cores or 4 cores or 5 cores. They care about price, features, and prestige (bling factor). That's it. In other words, no different than every other culture across the globe.

 

[Dufus]

I remember buying a local SIM card while working in mainland China. There were several prices so I asked what the difference was expecting some to have more services than the others such as GPRS, voicemail, international calling and so on. To my surprise the only difference was the numbers being thought unlucky, in between and lucky! IMHO it's a very personal thing though and should not be used to stereotype.

 

[EightySix Four]

I remember buying a local SIM card while working in mainland China. There were several prices so I asked what the difference was expecting some to have more services than the others such as GPRS, voicemail, international calling and so on. To my surprise the only difference was the numbers being thought unlucky, in between and lucky! IMHO it's a very personal thing though and should not be used to stereotype.

It is the difference between vanity and functionality though. For something functional no one cares, for something vanity related (cell phone number) people are more likely to care. At least that was my experience when I was there.

 

[tempestglen]

yeah a TK1 Denver (28nm) vs A8X (20nm) vs Core M (14nm) comparison would be very helpful. With 16FF+ I am sure Apple A9X will get close to Core M on perf and power efficiency. Nvidia's next gen Erista also could really pack a lot of performance.

http://www.brightsideofnews.com/2014/10/23/soft-machines-unveils-visc-cpu-architecture-linley/

 

[lefty2]

Is Spec2006 even available for iOS? I can't find it in the App Store anyway.

 

[Nothingness]

Is Spec2006 even available for iOS? I can't find it in the App Store anyway.

No. If you have $800 and the tools to build it, you can order it here

 

[paffinity]

Does anyone know how Anandtech is running CPU2000 on iPhone? Did they create an app, or using a jailbroken device via command line?

CPU2006 version of mcf will be difficult to run with only 2GB RAM. The way these phone/tablet OS's segment off memory (without swap) the devices tend to crash and reboot. At least that's what I see.