Atom Z3370-D 3Dmark Cinebench11.5 etc

[SiliconWars]

Hard drives are one of the main culprits for bad idle power I believe. Haswell draws so little power because every component is designed especially for that. Show me a Haswell system near the price of a Temash one then we'll talk idle power at the SoC level. You won't get it either way for that exact reason, price.

 

[Khato]

To be fair the 6.9 and 6.5 numbers are platform idle numbers for quad core chip with usb3 and higher clocks.who is to say that the a4 won't be much lower with tablet optimized components like ssds and 1.25v ram -among others.

Precisely what I'm getting at. The information from actual reviews that we have available right now isn't even up to par with AMD's marketing materials. Now sure, even the marketing claims aren't really on tablet levels, but they're at least better. Question is whether we'll ever see any such power optimized products - maybe that MSI W20 3M that SiliconWars linked will actually come to market and give us a better glimpse of what the platform is capable of?

Regardless, this has gotten to being a rather long tangent springing from the desire to know what kind of power the J1850 was consuming to provide these benchmark numbers. Hopefully we'll only have to be waiting a few more weeks now!

 

[mikk]

Score is 0.51 while a bobcat 1.6ghz is 0.61 , so essentialy
a dual core kabini 1ghz is worth a 1.33ghz bobcat in this
application at least , no doubt that it would suit a tablet.

It's not state of the art anymore, hency why no OEM is keen to build a new tablet around the A4-1250. Not when they know what Silvermont can do or a state of the art ARM Quad. Power efficiency and CPU raw speed looks pretty bad on A4-1250.

 

[Abwx]

Regardless, this has gotten to being a rather long tangent springing from the desire to know what kind of power the J1850 was consuming to provide these benchmark numbers. Hopefully we'll only have to be waiting a few more weeks now!

J1850 is 10W TDP.

http://www.cpu-world.com/news_2013/...nch_J1750_J1850_and_J2850_ULV_processors.html

 

[Abwx]

It's not state of the art anymore, hency why no OEM is keen to build a new tablet around the A4-1250. Not when they know what Silvermont can do or a state of the art ARM Quad. Power efficiency and CPU raw speed looks pretty bad on A4-1250.

Not worse than Z2760 according to the numbers
in the previous posts , so why should it be pretty
bad , exactly.?.

And not state of the art ? what is a better product
in X86..?

 

[Khato]

J1850 is 10W TDP.

http://www.cpu-world.com/news_2013/...nch_J1750_J1850_and_J2850_ULV_processors.html

... Because TDP means so much. Especially when you don't even know what Tjunction it's specified at.

 

[Abwx]

... Because TDP means so much. Especially when you don't even know what Tjunction it's specified at.

At 50 or 100°C a TDP of 5W is 5W in both case ,
Tjunction is function of the heatspreader thermal
resistance , i dont think that you really know
what you are talking about here cause any engineer
will tell that what you said is without any meaning.

 

[Abwx]

Hard drives are one of the main culprits for bad idle power I believe. Haswell draws so little power because every component is designed especially for that. Show me a Haswell system near the price of a Temash one then we'll talk idle power at the SoC level. You won't get it either way for that exact reason, price.

We ll soon know more about low power 2C Temash ,
there s already a few clues that it will be quite efficient.

http://www.sweclockers.com/nyhet/17...01-tums-dator-med-amd-processor-pa-4-watt-tdp

 

[Sweepr]

J1850 is 10W TDP.

http://www.cpu-world.com/news_2013/...nch_J1750_J1850_and_J2850_ULV_processors.html

Please stop pretending that you're talking about a tablet chip, this is Valleyview-D (desktop chip), not Valleyview-T (the chip meant to be used in tablets). Valleyview-T is a 2W SDP/~3W TDP out-of-order quad-core running at up to 2.4GHz.

 

[Khato]

At 50 or 100°C a TDP of 5W is 5W in both case ,
Tjunction is function of the heatspreader thermal
resistance , i dont think that you really know
what you are talking about here cause any engineer
will tell that what you said is without any meaning.

You might consider reading up on the subject. I'll even save you the trouble of finding a basic explanation and provide a link to a nice write up IDC did almost two years ago on the subject of how operating temperature affects CPU power consumption - http://forums.anandtech.com/showthread.php?t=2200205 .

 

[Abwx]

Please stop pretending that you're talking about a tablet chip, this is Valleyview-D (desktop chip), not Valleyview-T (the chip meant to be used in tablets). Valleyview-T is a 2W SDP/~3W TDP out-of-order quad-core running at up to 2.4GHz.

This is targeted numbers in old roadmaps , dont expect
this chip to run at 2.4ghz for only 3W , that s being naive.

The numbers already available for DT chips tell us that
it will run likely at 1.2-1.6 GhZ with a hefty turbo when
only one core is loaded.
After all they are the same chips built with the same process.

 

[Sweepr]

If these numbers are real (4 Silvermont cores @ 2GHz -> 1.35 pts CB11.5 32bit) then a 2.4GHz 4C Silvermont chip would score around ~1.78 pts in CB11.5 64-bit. That's Ivy Bridge Core i3-3217U level of MT performance and pretty good ST performance (A4 5000 quad-core Jaguar @ 1.5GHz scores 1.49 pts). Very respectable CPU performance for an entry level chip.

 

[Abwx]

You might consider reading up on the subject. I'll even save you the trouble of finding a basic explanation and provide a link to a nice write up IDC did almost two years ago on the subject of how operating temperature affects CPU power consumption - http://forums.anandtech.com/showthread.php?t=2200205 .

I dont think that i need a course in this matter , frankly ,
at least not of this level.

 

[Sweepr]

This is targeted numbers in old roadmaps , dont expect
this chip to run at 2.4ghz for only 3W , that s being naive.

The numbers already available for DT chips tell us that
it will run likely at 1.2-1.6 GhZ with a hefty turbo when
only one core is loaded.
After all they are the same chips built with the same process.

Comparing a tablet chip to a desktop chip is also naive (or ill-intendioned). Intel says up to 2.4GHz, how often this 4C chip sustains this clock is purely speculation at this point. The point is, they will soon release a ~3W x86 quad-core targeting ARM SoCs and it seems like this market is not AMD's main focus right now, their closest x86 ''competitor'' would be the A4-1200 3.9W chip with 1/3 A4-5000 CPU performance and <1/2 GPU performance.

 

[Abwx]

We ll know better in a few months but personaly i dont believe
that frequency will be high if 3W is to be the TDP , not SDP wich
is technicaly irrelevant data.

Besides the tablet chip is the same as the DT one basicaly ,
with some parts perhaps fused off and a slower GPU.

You can eventualy use the link provided by Khato
to evaluate the TDP in function of frequencies ,
you ll see that going from to 10W to 3W you ll need
to reduce frequency substancialy.

 

[Idontcare]

... Because TDP means so much. Especially when you don't even know what Tjunction it's specified at.

That's the kicker. Personally I believe that is why AMD refuses to spec TJmax for their piledriver parts, it makes the TDP spec meaningless.

At 50 or 100°C a TDP of 5W is 5W in both case ,
Tjunction is function of the heatspreader thermal
resistance , i dont think that you really know
what you are talking about here cause any engineer
will tell that what you said is without any meaning.

I dont think that i need a course in this matter , frankly ,
at least not of this level.

How can it not matter?

The power-consumption from leakage is critically dependent (exponentially) on the operating temperature.

Further, the required operating voltage (a factor in both dynamic and static power consumption) is likewise dependent on the operating temperature.

 

[Vesku]

Why are we letting the "but the quad ARM chips score high in multithreaded benchmarks" stuff slide? Notebooks and desktops are still rated on single thread performance by most on this forum and they are much better platforms for true multitasking than a tablet. I'd take a powerful dual core over a faster only with all 4 cores running quad when it comes to phones and tablets.

How have phone and tablet ARM chip makers managed to get people so hooked on the number of cores? Qualcomm actually had to fend off questions about whether they will introducean eight core SoC to "counter" Samsungs Octa configurations, rofl.

Seems to me a 4, 6, or 8 core chip should be much more interesting in desktops and notebooks than phones and tablets. I see no practical reason anyone shouldn't instead want 2 of those cores dropped off the 4 core tablet chips and that die space used for more graphics. Which is exactly what Apple has been doing with their Ax line of SoCs.

 

[monstercameron]

Why are we letting the "but the quad ARM chips score high in multithreaded benchmarks" stuff slide? Notebooks and desktops are still rated on single thread performance by most on this forum and they are much better platforms for true multitasking than a tablet. I'd take a powerful dual core over a faster only with all 4 cores running quad when it comes to phones and tablets.

How have phone and tablet ARM chip makers managed to get people so hooked on the number of cores? Qualcomm actually had to fend off questions about whether they will introducean eight core SoC to "counter" Samsungs Octa configurations, rofl.

Seems to me a 4, 6, or 8 core chip should be much more interesting in desktops and notebooks than phones and tablets. I see no practical reason anyone shouldn't instead want 2 of those cores dropped off the 4 core tablet chips and that die space used for more graphics. Which is exactly what Apple has been doing with their Ax line of SoCs.

but which is better power-wise 8 low clocked[lets say 1GHz], low voltage cores or 2 high clocked[2.4GHz], high voltage cores? especially running multiple things concurrently that are dependent on each other?

 

[Abwx]

How can it not matter?

The power-consumption from leakage is critically dependent (exponentially) on the operating temperature.

Further, the required operating voltage (a factor in both dynamic and static power consumption) is likewise dependent on the operating temperature.

Leakage is voltage dependent and rise with temp but
the chip we re discussing all are used at the minimum
of their functional voltages , hence leakage is low
and wont be much affected by temperature increasement.

Even if it grows exponentialy with temp it doesnt matter
much because the lower voltage will shift the exponential
growth at a segment of the curve where the slope is still
growing slowly and eventualy even slower than a square
or cubic law.

 

[Vesku]

but which is better power-wise 8 low clocked[lets say 1GHz], low voltage cores or 2 high clocked[2.4GHz], high voltage cores? especially running multiple things concurrently that are dependent on each other?

2 high clocked low voltage cores with ~50-100% larger GPU than the 8 low clocked cores.

"especially running multiple things concurrently that are dependent on each other?"

Are you claiming phone and tablet system usage is more demanding than notebook and desktop system usage (OS and applications)?

 

[Abwx]

but which is better power-wise 8 low clocked[lets say 1GHz], low voltage cores or 2 high clocked[2.4GHz], high voltage cores? especially running multiple things concurrently that are dependent on each other?

Grossly , once doubling the CPU frequency increase
TDP by more than what be consumed by a doubling
of the cores at constant frequency.

It occur well below doubling the frequency ,
basicaly it s better to add cores but then
you wont gain nothing in lowly threaded tasks ,
so the choice is as much a technical than a marketing
one given the emphasis given by some sites (purposely?)
to single threaded perfs.

 

[monstercameron]

2 high clocked low voltage cores with ~50-100% larger GPU than the 8 low clocked cores.

"especially running multiple things concurrently that are dependent on each other?"

Are you claiming phone and tablet system usage is more demanding than notebook and desktop system usage (OS and applications)?

maybe I should be more specific, I mean like the samsung touchwiz features such as eye tracking, multiwindow, voice command, multitasking, and all those features that are built upon other applications, features that are more power limited but require decent performance[like eye tracking]

 

[Khato]

I dont think that i need a course in this matter , frankly ,
at least not of this level.

And yet you initially claim that "At 50 or 100°C a TDP of 5W is 5W in both case" only to then backtrack and say that it's a relatively small effect once an explanation of temperature's effect upon power consumption is provided. But hey, you're welcome to your opinion.

Anyway, it'll be quite interesting to see how the TDP of these parts compare to actual power consumption under full load. Will it be like the low power Core SKUs that seem to exceed it? Or like Atom up to this date which never really gets up to the TDP.

 

[fusion238]

I disagree the dual 1GHz should be fine
http://www.3dmark.com/is/187221
my c-50 9w scores
3DMark Score 10053.0
Graphics Score 12461.0
Physics Score 5998.0

so the a4-1200 should score a little higher around 7200 for the physics score.

I've been setting up a HP Touchsmart 11z laptop (which sells in the $350-$425 range) with the AMD quad A6-1450 for a relative and even on wifi, it seems that is browses as quickly as my desktop six core FX with discrete graphics card. Will try to post a mini review with a few benchs and playing DOTA2.

 

[monstercameron]

I've been setting up a HP Touchsmart 11z laptop (which sells in the $350-$425 range) with the AMD quad A6-1450 for a relative and even on wifi, it seems that is browses as quickly as my desktop six core FX with discrete graphics card. Will try to post a mini review with a few benchs and playing DOTA2.

great, can you link to the results page?