Bay Trail benchmark appears online, crushes fastest Snapdragon ARM SoC

[Vesku]

Going to need more information before any sort of picture of Bay Trail can be constructed.

 

[Hans de Vries]

Not really. This is what Intel told us:

The same is true for this diagram:

Watch out to not get fooled by the usual benchmark manipulations.
Marketeers won't lie outside the lawyer determined boundaries.
They'll let the journalists and naive fans do all the lying for them.

For instance: 4.4x lower power

(22nm BayTrail versus current 32nm Atom at "iso-perf")

What this says is that a 2GHz quadcore BayTrail will use 4.4x times
less power as a dual core 32nm Atom clocked at more then 4GHz.
The overclock is needed to give the latter the same performance
(iso-perf) as BayTrail...

Hans

 

[Arachnotronic]

Watch out to not get fooled by the usual benchmark manipulations.
Marketeers won't lie outside the lawyer determined boundaries.
They'll let the journalists and naive fans do all the lying for them.

AnTuTu and Quadrant came into focus when Microsoft choose Nvidia's
Tegra3 for surface tablets as a direct competitor.

For instance this the Atom in the Motorola Razr-i scores better as the
Nvidia Tegra3 in Quadrant but if you go one step further and look at
the subscores in detail then you'll see that the Atom CPU score is just
1/3 of that of the Quad A9 Tegra3.....

source.

The 2GHz Atom with hyper-threading score is second in the list and the
CPU sub-score is the blue piece of the bar.

Furthermore. A benchmark as AnTuTu can easily be abused with
processors that can run in burst-mode at 5+ times or so their
(advertised) TDP. The benchmark is to short to warm up the phone
if you run at a far higher burst mode TDP.

Here's a simple example how the score degrades if you run AnTuTu
multiple times:

source.

And this is with an Arm device without all the high TDP super bursting
features of the Baytrail SOC...

Hans

Hans,

It's a good thing that we're not talking about a single-core, 5 year old Atom uArch and instead the redesigned, not-at-all crappy "Silvermont" in a quad core config

 

[Arachnotronic]

The same is true for this diagram:

Watch out to not get fooled by the usual benchmark manipulations.
Marketeers won't lie outside the lawyer determined boundaries.
They'll let the journalists and naive fans do all the lying for them.

For instance: 4.4x lower power

(22nm BayTrail versus current 32nm Atom at "iso-perf")

What this says is that a 2GHz quadcore BayTrail will use 4.4x times
less power as a dual core 32nm Atom clocked at more then 4GHz.
The overclock is needed to give the latter the same performance
(iso-perf) as BayTrail...

Hans

Uhhhh, no? How about a downclocked Bay Trail?

 

[SOFTengCOMPelec]

When you got a 250M$ R&D budget for multiple core designs. This is what you get.
http://arm.com/files/pdf/Earnings_Tables_2012.pdf

At the end of the day, SUCCESS is typically measured by MARKET SHARE.

Hence articles like This

Which say stuff like this quote from it :-

ARM Holdings Plc, which has sprinted ahead of Intel Corp. in the market for mobile chips

So, if they can do the above with a considerably smaller budget than Intel's, I take my hat off to them, bend down and say, "wow!".

 

[Vesku]

I think his point about "not a lie but not the truth either, tada it's called marketing" still stands.

 

[sontin]

So, a 1100MHz QuadCore Silvermont will be faster than a 1900MHz A15 QuadCore? Sure, lol.

 

[ShintaiDK]

At the end of the day, SUCCESS is typically measured by MARKET SHARE.

Hence articles like This

Which say stuff like this quote from it :-



So, if they can do the above with a considerably smaller budget than Intel's, I take my hat off to them, bend down and say, "wow!".

I assume you dont want to talk big.LITTLE anymore then?

You never answered my questions.

If these numbers are true, then ARM is a goner in all highend phones and tablets.

 

[Silverforce11]

At the end of the day, SUCCESS is typically measured by MARKET SHARE.

No, in business, success is measured in profits.

 

[Sweepr]

Bay Trail will have an egde over Temash/Kabini in power consumption while this could be gone with "Beema" that arrives in 2014.

as for Bay Trail T scores, does the benchmarking program detects when the chip is using boost/turbo and show it or does it not?

Since Bay Trail T chip could have turbo/boost mode.!

Temash barely beats old Ivy Bridge ULV efficiency. It wouldnt surprise me if Haswell ULT wipes the floor with Temash (normalized battery life) as it already puts it to shame in both CPU and GPU performance (especially the GT3 versions). No wonder why they have to sell it so cheap. Considering Bay Trail is much lower performance/power than that, AMD doenst even deserve to be mentioned next to this Intel/ARM solutions that fit smartphones/small tablets.

 

[blackened23]

So, a 1100MHz QuadCore Silvermont will be faster than a 1900MHz A15 QuadCore? Sure, lol.

You sound really worried there! It's going to happen. Intel has superior technology and now that they're focused on efficiency, we're seeing them crush ARM SOCs, as expected. It's okay though. ARM SOCs can compete on price, now that they're going to lose by huge margins in both performance and efficiency.

I find it all too funny if only because ARM Holdings have been so arrogant over the years. "Intel will never catch us". Allllrighty. Toooo funny.

 

[SOFTengCOMPelec]

No, in business, success is measured in profits.

What happened to taking the LONG TERM view of a business.

Perhaps if Intel charged $10,000 for each of their chips, they would make the most PROFITS, both this year, and next year.

But, would they exist in twenty years time ?

If your market share is continually falling, in the markets that your company majors in, then even if you are still making healthy profits, surely your company is going to disappear in the longer term ?

I get the impression, Intel DON'T want profits at ANY cost. They want to survive for a long period of time as well, that does not necessarily involve making the MOST profits today, at the expense of disappearing in the future.

If Intel spend $0 R&D, they may maximise profits for the next year or so, but eventually their stuff would get overtaken, and Intel would probably disappear.

 

[frozentundra123456]

I predicted many months ago this would happen, once intel focused on the task at hand - they produced a part which destroys all ARM SOCs within a lower power envelope on their 22nm process. It's pretty hilarious considering that the ULV Haswell gets 13 hours of battery life in a macbook air while the iPad 4 gets what? 10 hours on a custom ARM SOC? So based on that, I wonder how much battery life bay trail will get? 15+? Probably.

Personally, I think it's freakin hilarious that ARM's hubris is about to catch up to them. They said intel would never ever beat them in efficiency. Hilarious. I can't wait to see the silly damage control posts full of people doubting intel.

Nothing will shut ARM up faster than seeing their best ARM SOCs get destroyed, easily, by intel's bay trail.

Actually, you are seeing it already from some in this forum. Seems as if every marketing powerpoint slide for AMD is taken as gospel, while admittedly very preliminary benchmarks favorable to intel are debunked and/or claimed to not really matter.

I definitely think it is too early to say how well this chip will perform, but the data looks promising. All I am saying is that it is amazing how far some will go to bash one company and accept every claim from another.

 

[podspi]

I wonder how far away we are from mobiles that can dock into fully capable (if not bleeding-edge) desktops? I really hope somebody actually does this well. It is also why I think ChromeOS is a bad idea. Google should try to make Android more desktop/laptop ready, as opposed to having two different OSes with different ecosystems.

 

[SOFTengCOMPelec]

I assume you dont want to talk big.LITTLE anymore then?

You never answered my questions.

If these numbers are true, then ARM is a goner in all highend phones and tablets.

I was trying to counter some peoples apparent overestimation of the capabilities of Intel chips, over Arm's chips, rather than get into a huge technical discussion of the merits of one architecture, over another.

As to "being a goner in all high end phones" ...

As you get closer and closer to what really is what was a desktop chip of a few years ago (or longer), put into a tiny battery powered, handheld device, then Intel really does have some aces up their sleeve.
Intel have huge experience, existing designs and R&D teams, patents, one of (if not) the best fab plant capabilities and all sorts of other advantages to bring to the table.

So, I agree with you, there IS big.little.trouble.in.China (joke, oops, I mean England, Cambridge vs Intel locations worldwide).

 

[Arachnotronic]

So, does anybody smell a repeat of what happened to AMD when Conroe came out? lol

I remember all of the denial back then, too...

What was it Charlie D. said? The AMD engineers were "dancing in the halls" re: Barcelona? LOL!

No wonder Warren East is "retiring"...that guy saw this coming from a mile away and didn't want to be around when Intel proved him wrong.

 

[ShintaiDK]

What happened to taking the LONG TERM view of a business.

Perhaps if Intel charged $10,000 for each of their chips, they would make the most PROFITS, both this year, and next year.

But, would they exist in twenty years time ?

If your market share is continually falling, in the markets that your company majors in, then even if you are still making healthy profits, surely your company is going to disappear in the longer term ?

I get the impression, Intel DON'T want profits at ANY cost. They want to survive for a long period of time as well, that does not necessarily involve making the MOST profits today, at the expense of disappearing in the future.

If Intel spend $0 R&D, they may maximise profits for the next year or so, but eventually their stuff would get overtaken, and Intel would probably disappear.

Your theories are completely nonsense. 10000$ chips would not make any profits at all. Higher price, the lower volume. But at a certain point lower price will not raise the volume. And at another certain point you get perfect profit compared to the price/volume ratio.

 

[SiliconWars]

Actually, you are seeing it already from some in this forum. Seems as if every marketing powerpoint slide for AMD is taken as gospel, while admittedly very preliminary benchmarks favorable to intel are debunked and/or claimed to not really matter.

Intel knows every marketing trick in the book and is currently inventing more as we speak. "iso perf" and "SDP" being two of the latest.

It's generally quite easy to tell when we're being had. The claims are outrageous, stuff like 30% faster at half the clock speed for example (yes that's the one we're currently supposed to believe).

Are you seriously buying this? You know what they say about the bigger the lie...

 

[SiliconWars]

So, does anybody smell a repeat of what happened to AMD when Conroe came out? lol

I remember reading Anand's last years Atom (Medfield) article, and I can see how much market share Intel gained since then when they were supposedly smashing the competition.

No wonder Warren East is "retiring"...that guy saw this coming from a mile away and didn't want to be around when Intel proved him wrong.

And the reason Otellini is retiring is...?

 

[ShintaiDK]

I was trying to counter some peoples apparent overestimation of the capabilities of Intel chips, over Arm's chips, rather than get into a huge technical discussion of the merits of one architecture, over another.

As to "being a goner in all high end phones" ...

As you get closer and closer to what really is what was a desktop chip of a few years ago (or longer), put into a tiny battery powered, handheld device, then Intel really does have some aces up their sleeve.
Intel have huge experience, existing designs and R&D teams, patents, one of (if not) the best fab plant capabilities and all sorts of other advantages to bring to the table.

So, I agree with you, there IS big.little.trouble.in.China (joke, oops, I mean England, Cambridge vs Intel locations worldwide).

It seems to be you wastly overestimating the capabilities of ARM Holding. The 2 biggest custom SOC makers of ARM are rejecting the big.LITTLE concept. That should be a huge wakeup call for you.

 

[mrmt]

And the reason Otellini is retiring is...?

Enlighten us, please.

 

[SOFTengCOMPelec]

Your theories are completely nonsense. 10000$ chips would not make any profits at all. Higher price, the lower volume. But at a certain point lower price will not raise the volume. And at another certain point you get perfect profit compared to the price/volume ratio.

In practice you are right. But, I was trying to explain the concept, of profit, not being the only thing a business needs to keep their eye on, in order to succeed in the long term. The $10,000 was not meant as a serious thing (it was to make a point/example).
Some of Intels top chips are not that much less than $10,000. I think you once posted a price sheet, where the top chip was something like $3995 ?

So, that would make a dual cpu chip system, what, $8,000, and a (i.e. a 4 way) quad cpu chip system $16,000, for the Intel chips alone.

 

[mrmt]

Actually, you are seeing it already from some in this forum. Seems as if every marketing powerpoint slide for AMD is taken as gospel, while admittedly very preliminary benchmarks favorable to intel are debunked and/or claimed to not really matter.

The fanbois want AMD to succeed, so whenever they claim a pie-in-the-sky improvement that fits the fanboi's *wishes*, then they take it at face value. With Intel, it's not that the fanbois are neutral. They *don't* want Intel to succeed because it will hurt AMD even more, so whenever they see whatever evidence of improvement they will deny it because it hurts their wishes.

That's confirmation bias on its crudest form.

 

[Chiropteran]

Does this even matter? Intel already had a phone processor, and it supposedly performed really well, and nobody anywhere bought a phone using it.

When you buy a cell phone, there are are about a dozen features which are all more important than CPU performance. I don't see this going anywhere quickly. Maybe a few niche phones will use it, but it won't really be relevant.

If it's not used in the iPhone or some form of Samsung Galaxy it doesn't really matter for a real smartphone enthusiast.

 

[Arachnotronic]

Does this even matter? Intel already had a phone processor, and it supposedly performed really well, and nobody anywhere bought a phone using it.

When you buy a cell phone, there are are about a dozen features which are all more important than CPU performance. I don't see this going anywhere quickly. Maybe a few niche phones will use it, but it won't really be relevant.

If it's not used in the iPhone or Galaxy it doesn't really matter for a real smartphone enthusiast.

The Medfield and CT+ didn't come paired with an LTE modem; this is why only Qualcomm has been successful thus far in the mid to high end phone space.

Also, Intel's last chip didn't massively outperform the competition like this one does.