Bay Trail benchmark appears online, crushes fastest Snapdragon ARM SoC

Page 7 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Sweepr

Diamond Member
May 12, 2006
5,148
1,142
131
Intel will launch Avoton/Rangeley server SoCs with up to 8 Silvermont cores (2.4GHz, 2.7GHz Turbo) @ 5-20W TDP this year. They will put Silvermont cores everywhere to battle ARM.
 
Last edited:

Imouto

Golden Member
Jul 6, 2011
1,241
2
81
What bench and what ARM competition?

OP's bench. Looks like you didn't even read those links and proceeded to cheerleading.

Is an 8 Core PD always faster than a 4 core i7? Answer: no.

This time they're showing off an embarrasingly parallel benchmark score boasting a 30% performance lead over a 4 core S800. Krait cores that perform about the same as the A15s in the Huawei chip.

It's been proven many times that AnTuTu scores mean nothing anyway. It may be the same as the previous Atom scores and have an outstanding memory score while sucking at CPU. It may be burst performance. It may be the Intel compiler. It may be reporting its speed wrong. It may be a combination of all that.
 

Exophase

Diamond Member
Apr 19, 2012
4,439
9
81
So, in AnTuTu quad 2-wide out of order Silvermont cores will perform better than dual 4-wide out of order Haswell cores with hyper threading.

According to the following link, a Core i5-4200U clocked at 1600 MHz (2600 MHz Turbo Frequency) scores 54,861 in AnTuTu:

http://vr-zone.com/articles/samsung...d-core-i5-scores-over-54000-points/40677.html

Given the excellent scaling of the AnTuTu benchmark, a quad core Silvermont running at 1600 MHz will score over 63,000 points in AnTuTu rendering it 15% better than a Haswell i5-4200U.

The excellent scaling of the AnTuTu benchmark appears to be less than credible, but a quad core Silvermont running at 2100 MHz will score over 82,000 points in AnTuTu. This would render it 50% better than a Haswell i5-4200U (Tray price $287).

Interesting, but doesn't this look strange?

AnTuTu's overall score is an aggregate of several tests, including graphics and storage ones. The CPU part scales very well with core count because they just replicate a very old single-threaded benchmark to every core. But the entire score obviously does not scale with CPU count or clock.

If you want to talk CPU performance you should just look at the integer and floating point scores, and to some extent memory score. Or at least that would be the case if it were a fair or good benchmark for those categories.
 

simboss

Member
Jan 4, 2013
47
0
66
It's been proven many times that AnTuTu scores mean nothing anyway. It may be the same as the previous Atom scores and have an outstanding memory score while sucking at CPU.

Every benchmark has one (or even several) bias, you can't really find the perfect benchmark, that in itself is not a problem specific to AnTuTu.

It may be burst performance.
It may be reporting its speed wrong.

That is clearly not acceptable for a benchmark, it needs to give consistent results across runs.

If you want to test the burst performance, you can do it, but do it in a way that makes it clear it is the burst performance, and report the max speed you have used, not the base one that does not mean anything.

It may be the Intel compiler.

That again is totally acceptable, you can't use the same compiler anyway, it makes perfect sense to use the best you can.

It may be a combination of all that.

THIS is the problem with AnTuTu: it is a closed source, not documented benchmark that nobody can understand and verify.

What was used as a "toy" by people on their phones (looking at the AnTuTu website is quite telling about its lack of technical relevance) is now used as an industry standard to try to (or even succeed) make users/reviewers/investors believe a platform is the best.


an extract from the App description itself that shows how easy it is to trick the benchmark:
NOTE: Some manufacturers promote they products for profits by using fake benchmarks. In fact, some users are unable to run official version of AnTuTu Benchmark on their devices for manufacturers have modified their products, thus the cheating behavior will not be discovered. These manufacturers passed the buck to AnTuTu, claimed that it is because of the incompatibility of AnTuTu Benchmark. AnTuTu make a solemn statement here that there is no compatibility problems on AnTuTu Benchmark. If AnTuTu Benchmark can not run on your device, or there is no readable words on the interface of AnTuTu Benchmark, and your device is made by Chinese manufacturer, then you should know this is cheating.

and later:
Is AnTuTu Benchmark accurate? It is hard to answer the questions regards to the accuracy. This can only judged by users themselves. The method is quite simple: If a new device's benchmark exceed more than 50% than equal class devices released at the same time, then the result is questionable.
 

FwFred

Member
Sep 8, 2011
149
7
81
AnTuTu's overall score is an aggregate of several tests, including graphics and storage ones. The CPU part scales very well with core count because they just replicate a very old single-threaded benchmark to every core. But the entire score obviously does not scale with CPU count or clock.

I think it would be a mistake for AnTuTu (full score) to fully scale with CPU count.

Also a very good reason why AnTuTu favors quad core chips over dual core Clovertrail+ (yes, 4 threads help). This test really does cut both ways. I prefer benchmarks based on real world phone or tablet usages. Linpack, Quadrant, Geekbench also are not good phone benchmarks. Too many benchmarks fully scale with core count, which is very rare on real world usages. Single thread performance is still critical, especially when it comes to UI responsiveness and browser usage.
 
Mar 10, 2006
11,715
2,012
126
Intel's biggest challenge is not absolute performance, or even performance/watt. The engineers and scientists have delivered. Now it is up to the management and marketing department to price Silvermont right. As good as Silvermont is, if they demand too much of a premium over nearest ARM rival, most ODMs would pick ARM.

As we have seen the evolution of 'good enough' desktop, so we will see the rise of 'good enough' smartphones and tablets. And this transition to good-enough devices will happen quicker with mobile devices than with desktop. It might have already started.

So Intel's biggtest threat is not ARM or any of their licensees, but their own internal desire to jack up profit margin as high as possible.

Do you think Intel is dumb enough to price themselves out of this market, particularly when participation is so vital to their long term future?

It's really this simple:

Either Intel sells a Bay Trail chip for $30-$50 (whatever these chips go for), or they sell nothing. With PC sales waning, it's better to sell something profitably, than to sell nothing.
 
Last edited:

monstercameron

Diamond Member
Feb 12, 2013
3,818
1
0
Do you think Intel is dumb enough to price themselves out of this market, particularly when participation is so vital to their long term future?

It's really this simple:

Either Intel sells a Bay Trail chip for $30-$50 (whatever these chips go for), or they sell nothing. With PC sales waning, it's better to sell something profitably, than to sell nothing.

we are talking about intel, right?
 

Exophase

Diamond Member
Apr 19, 2012
4,439
9
81
Do you think Intel is dumb enough to price themselves out of this market, particularly when participation is so vital to their long term future?

It's really this simple:

Either Intel sells a Bay Trail chip for $30-$50 (whatever these chips go for), or they sell nothing. With PC sales waning, it's better to sell something profitably, than to sell nothing.

I don't think Intel is that dumb.

I also don't think they're dumb enough to disable 64-bit on BayTrail-T or Merrifield, and AVX2 on Celeron or Pentium Haswells. If that's what ends up happening I won't have a very good explanation for why. Don't even ask me about TSX today
 

24601

Golden Member
Jun 10, 2007
1,683
39
86
Intel has ALWAYS rather sold nothing than sell something for less than 60% margins.

Something tells me this time isn't any different.
 

erunion

Senior member
Jan 20, 2013
765
0
0
Bay Trail wouldn't crush the fastest ARM competition...

in this bench.

Considering that licensed ARM cores scale to a max of 4, that octo huawei is a big.little configuration.

But it is telling that you are so willing to jump to impossible conclusions to disprove a point.
 
Last edited:

blackened23

Diamond Member
Jul 26, 2011
8,548
2
0
I wouldn't underestimate the business minds at intel. In prior years they just didn't have the focus on super portable products, but now that they do - they will price themselves appropriate to the market. Now they won't be dirt cheap as some ARM SOCs as I imagine the Bay Trail will be better by every metric (debatable, but my opinion). But, intel's new CEO in particular knows what has to be done and he will make it a competetive offering even in terms of price.

Again, don't underestimate the business side of Intel - they have some of the best and brightest minds available. They want domination of this market, in fact they're hellbent on it.
 

moonbogg

Lifer
Jan 8, 2011
10,637
3,095
136
Wow this is so exciting. Can't wait to put this in my phone and oc the crap out of it. Imagine the fps that fruit ninja will get.
 

mrmt

Diamond Member
Aug 18, 2012
3,974
0
76
I'm talking about targets, not actuality.

It's a mentality thing.

No, you are not talking about targets. In fact, it's almost thread crapping.

Intel runs a behemoth server business, and there target margins are sky high. At the same time, you can't get huge margins on Celeron or Atom, so target margins should be lower.
 

Cerb

Elite Member
Aug 26, 2000
17,484
33
86
I don't think Intel is that dumb.

I also don't think they're dumb enough to disable 64-bit on BayTrail-T or Merrifield, and AVX2 on Celeron or Pentium Haswells. If that's what ends up happening I won't have a very good explanation for why. Don't even ask me about TSX today
What about TSX, today? Nobody will ever want optimistic lock bypassing with a Core CPU (RTM might be mostly friendly to many threads, but HLE should pretty much be a benefit to each thread at a time). If you might, that'll be another $800+ (a whole new notebook, for anyone wanting to counter with $400), and your current machine might reach a landfill quicker.
 

Exophase

Diamond Member
Apr 19, 2012
4,439
9
81
Considering that licensed ARM cores scale to a max of 4, that octo huawei is a big.little configuration.

ARM clusters scale to a max of 4. ARM's interconnects support two clusters. Those clusters can be the normal big.LITTLE A7 + A15, or they can be A7 + A7 or A15 + A15. MediaTek appears to be doing something with the former.

Nothing stops an SoC from running both clusters and hence all eight cores simultaneously, so long as they didn't break the interconnect like Samsung appears to have with Exynos 5410.

Not that running 8 cores at the same time is very useful, but being able to simultaneously run cores from two separate voltage and clock domains is (that's what using two clusters can give you).

What about TSX, today? Nobody will ever want optimistic lock bypassing with a Core CPU (RTM might be mostly friendly to many threads, but HLE should pretty much be a benefit to each thread at a time). If you might, that'll be another $800+ (a whole new notebook, for anyone wanting to counter with $400), and your current machine might reach a landfill quicker.

I said not to ask
 

24601

Golden Member
Jun 10, 2007
1,683
39
86
No, you are not talking about targets. In fact, it's almost thread crapping.

Intel runs a behemoth server business, and there target margins are sky high. At the same time, you can't get huge margins on Celeron or Atom, so target margins should be lower.

Nice trying to mix one with the other.

Back that up with some numbers, or that's almost thread crapping.

Their target margin in Atom has always been high, and they have almost always succeeded in meeting it.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |