The Intel Atom Thread

[Abwx]

So instead you want us to all look at a review that skews the results in the other direction? I do believe that's the definition of being hypocritical.

Wait, so PClab use two different set ups for Kabini and BT and is not skewed but hardware.info who use the same set up for both would be the one who are skeweing results.?..

In case you can't figure out what I mean, take a look at the idle numbers. 9.6W for the 5350 compared to 16.8W for the J1900. That's the only reason why the 5350 looks even remotely competitive on any of the load tests and it's purely due to their choice in components for the J1900 system. If you assume a level playing field of equal idle power (Baytrail is actually better, but we'll give the 5350 a handicap) then you're looking at delta numbers from that review of 12.1W for the 5350 and 2.8W for the J1900 on 3dmark and then 16.7W for the 5350 and 4.3W for the J1900 on cinebench. Which aren't too far off from what other reviews see for the delta measurements, eg http://www.tomshardware.com/reviews/athlon-5350-am1-platform-review,3801-9.html

Actualy none of THG inflated numbers contradict hardware.info numbers,
think about it a moment, i hope that hardware.fr will soon do their reviews
as they are to this day the only ones who are isolating the PSU losses
from the MB effective comsumption , btw take a look at the thread
about the 5350 i posted a graph from a site that used an oversized
400W PSU but an extremely efficient one and compare with THG s.

 

[ElFenix]

So instead you want us to all look at a review that skews the results in the other direction? I do believe that's the definition of being hypocritical.

In case you can't figure out what I mean, take a look at the idle numbers. 9.6W for the 5350 compared to 16.8W for the J1900. That's the only reason why the 5350 looks even remotely competitive on any of the load tests and it's purely due to their choice in components for the J1900 system.

do you mean the board?

 

[Khato]

Wait, so PClab use two different set ups for Kabini and BT and is not skewed but hardware.info who use the same set up for both would be the one who are skeweing results.?..

Where precisely did I say that PCLab was not skewed/any better? Oh wait, I didn't... Because you're quite correct that their methodology is flawed. Meanwhile the hardware.info review appears to be fine on the methodology, but their results aren't representative of other reviews due to the abnormally high idle power consumption on the J1900. As well, you tried to claim due to that abnormality that "hardware.info s numbers that show BT being hardly more CPU efficient and even less efficient once the GPU is used." when they do no such thing once idle power is accounted for.

Actualy none of THG inflated numbers contradict hardware.info numbers,
think about it a moment, i hope that hardware.fr will soon do their reviews
as they are to this day the only ones who are isolating the PSU losses
from the MB effective comsumption , btw take a look at the thread
about the 5350 i posted a graph from a site that used an oversized
400W PSU but an extremely efficient one and compare with THG s.

Quite correct, the delta power consumption numbers compared to idle are quite comparable between hardware.info and THG. And they show that the 5350 uses over 3x the power of the J1900 in CPU loads - how that can be considered "hardly more CPU efficient" is beyond me... unless you were taking the hardware.info numbers without accounting for the idle power discrepancy. Oh, and since you want to be so obsessed about power supply sizing in these low power reviews - that only has a marked effect on idle power. Sure the efficiency fluctuates quite a bit at low power, but that's simply due to the fixed power losses being responsible for the majority of the inefficiency. Which is to say that you don't see much of a difference in idle -> load delta power consumption numbers going from a super efficient brick + 80 watt pico psu to a 1200 watt ATX beast. Which is why even on the pclab.pl article you still see a 5W delta on CPU load and 6W delta on GPU load for the J1900 and a 15W delta for CPU load and 16W delta for GPU load on the 5350... which, by the way, are actually the worst numbers for the J1900 out of all three reviews while they're also the worst for the 5350 on graphics and then between the other two for CPU.

 

[Khato]

do you mean the board?

Motherboard or other configuration details that aren't noted. It's pretty obvious from other reviews that J1900 based systems can demonstrate equal/lower idle power compared to the 5350... and we know from motherboard roundups that there can easily be 10+ watt differences in idle power due to the motherboard being used.

 

[Abwx]

Khato, i posted numbers from two different reviews using a same efficient
400W PSU , oversized though but apparently quite efficient below 20% load,
the german review include a J1800 wich should have close iddle comsumption
than a J1900 and numbers seems to correlate within errors margins, please check here, posts 102 and 103 :

http://forums.anandtech.com/showthread.php?t=2374795&page=5

 

[blackened23]

Seeing these conspiracy theories about "rigged" reviews from you is just tiring. Shall we just call your hardware.fr reviews "rigged" AMD biased reviews? Give it a rest man. Clearly you discard any reviews that has information you don't like. Clearly you think AMD is being victimized and you feel that everyone is out to get AMD. Whatever.

Like I said. Maybe the reviews YOU like are the rigged ones. The biased ones. We'll just go with that.

Bottom line is both BT and Kabini embedded chips basically suck compared to the Celeron 1820 and Pentium G3220. Both of those are nearly the same price as well. So I really question why anyone would even consider those platforms when they could easily get a Celeron 1820 or G3220 pentium platform for around 100$...

 

[Abwx]

Seeing these conspiracy theories about "rigged" reviews from you is just tiring. Shall we just call your hardware.fr reviews "rigged" AMD biased reviews? Give it a rest man. Clearly you discard any reviews that has information you don't like. Clearly you think AMD is being victimized and you feel that everyone is out to get AMD. Whatever.



Like I said. Maybe the reviews YOU like are the rigged ones. The biased ones. We'll just go with that.


Bottom line is both BT and Kabini embedded chips basically suck compared to the Celeron 1820 and Pentium G3220. Both of those are nearly the same price as well. So I really question why anyone would even consider those platforms when they could easily get a Celeron 1820 or G3220 pentium platform for around 100$...

I discard reviews that are technicaly not consistent and i sustain my
point with technical analysis rather than ad hominem attacks , isnt it...

I wont insist because laws of physics are what they are and i ll just remind you that with the slightest knowledge of what is a watt one would instantly understand that THG comsumptions numbers for both Kabini and BT in the quoted reviews are physicaly impossible because the chips would reach temperatures well above 100°C, particularly for BT celeron as they are passively cooled making it more easy to estimate the cooling apparatus thermal resistance, now if you want to make a test stick a 15W thermal source to such a BT heatspreader and try touching it after 5 minutes.....

 

[blackened23]

Hah. Yeah. I discard your reviews because they're technically not consistent and biased. We'll just go with that.

I love these conspiracy theories. Maybe we should just go both ways with these things just for the hilarity of it.

 

[Abwx]

Hah. Yeah. I discard your reviews because they're technically not consistent and biased. We'll just go with that.

I love these conspiracy theories. Maybe we should just go both ways with these things just for the hilarity of it.

Come with numbers and an analysis, anything else doesnt interest me.

 

[Khato]

Khato, i posted numbers from two different reviews using a same efficient
400W PSU , oversized though but apparently quite efficient below 20% load,
the german review include a J1800 wich should have close iddle comsumption
than a J1900 and numbers seems to correlate within errors margins, please check here, posts 102 and 103 :

http://forums.anandtech.com/showthread.php?t=2374795&page=5

Post 102: Aren't these pretty much the same delta power numbers that everything else shows?

Post 103: This shows the 5350 having a 17W delta between idle and CPU load. I believe that's the highest yet? As well, the power supply they used there appears to have something around a 10W base power loss. And yes, the J1800 idle power consumption is likely quite similar to the J1900, but load characteristics/performance are going to be markedly different due to the fact that the J1800 runs its two cores at higher frequency than the quad core J1900. So there's really nothing to be learned from attempting to compare the load delta power numbers on the J1800 to the 5350 - it'd be pretty pathetic if the 5350 didn't win by a nice margin.

 

[Khato]

I wont insist because laws of physics are what they are and i ll just remind you that with the slightest knowledge of what is a watt one would instantly understand that THG comsumptions numbers for both Kabini and BT in the quoted reviews are physicaly impossible because the chips would reach temperatures well above 100°C, particularly for BT celeron as they are passively cooled making it more easy to estimate the cooling apparatus thermal resistance, now if you want to make a test stick a 15W thermal source to such a BT heatspreader and try touching it after 5 minutes.....

The only way I can think of to interpret the above simply doesn't make logical sense? You're implying that the THG consumption numbers are impossible because it's putting all that power into the SoC itself?

 

[blackened23]

Come with numbers and an analysis, anything else doesnt interest me.

What you're doing is hardly analysis. It's more like grasping at straws to discredit any review that doesn't show AMD in a favorable light in any aspect. So if seven reviews show power levels that aren't to your liking, well they're obviously RIGGED according to you. Some sort of underhanded conspiracy against AMD. Right? Meanwhile you'll link the one outlier website which according to you is the "true" test. Whatever you say. Well, both sides can play that little game of yours. Any and all reviews you link from this point forward are rigged and AMD biased. Of course you will claim 20 reviews are rigged while the one outlier at hardware.fr is the true review. Now THAT doesn't sound silly, does it? So we'll just play that game, but on the other side. I'll be sure to point how rigged and biased that one review you link is from this point forward. Just to underscore how silly you sound doing the same thing over and over and over and over and over again. And over again.

Besides which, nothing i've seen you state is even remotely accurate in terms of "technical analysis".

 

[Abwx]

The only way I can think of to interpret the above simply doesn't make logical sense? You're implying that the THG consumption numbers are impossible because it's putting all that power into the SoC itself?

Yes , about all the power is dissipated in the SoC, the SSD/HD, RAM and VRMs share is about negligible when the system is fully loaded, they should account for 5-7W at most so if a system really pump 47W then it means that the SoC dissipate about 40W and heatspreaders like Kabini s cant dissipate that much heat without the fan going crazy, even worse for BT since the passive cooling wouldnt allow to dissipate 20W or even 15W continuously.

 

[Abwx]

What you're doing is hardly analysis. It's more like grasping at straws to discredit any review that doesn't show AMD in a favorable light in any aspect. So if seven reviews show power levels that aren't to your liking, well they're obviously RIGGED according to you. Some sort of underhanded conspiracy against AMD. Right?

Not right , and i didnt say that theses are conspiracies, stop using such straws, i said that it is clulessness and thus incompetence and i clearly said why....

Meanwhile you'll link the one outlier website which according to you is the "true" test. Whatever you say. Well, both sides can play that little game of yours. Any and all reviews you link from this point forward are rigged and AMD biased. Of course you will claim 20 reviews are rigged while the one outlier at hardware.fr is the true review. Now THAT doesn't sound silly, does it? So we'll just play that game, but on the other side. I'll be sure to point how rigged and biased that one review you link is from this point forward. Just to underscore how silly you sound doing the same thing over and over and over and over and over again. And over again.

Besides which, nothing i've seen you state is even remotely accurate in terms of "technical analysis".

Meanwhile i m still all ears for technical explanations if ever you have
even a single one , for the time all i m , agonizingly, reading is a wall
of words that could well be posted on a rants dedicated forum...

 

[blackened23]

Meanwhile i m still all ears for technical explanations if ever you have
even a single one , for the time all i m , agonizingly, reading is a wall
of words that could well be posted on a rants dedicated forum...

The wording here is what's agonizing. Perhaps you should post on your wonderful AMD computer instead of your phone (i'm assuming)? Just a thought. You do, have an AMD based PC, right?

 

[Abwx]

The wording here is what's agonizing. Perhaps you should post on your wonderful AMD computer instead of your phone (i'm assuming)? Just a thought. You do, have an AMD based PC, right?

It s a core2 pentium t4400 with 4gb ddr3 1333mhz Asus with a Gt320m nvidia
17" led screen laptop.

Other than that IL is wonderfull.

 

[Sweepr]

New Bay Trail-T based 64-bit Android tablets coming soon:

Asus Transformer Pad TF103C Bay Trail Android tablet

Here’s what to expect from the Asus Transformer Pad TF103C:

10.1 inch, 1280 x 800 pixel IPS display
Intel Atom Z3745 quad-core, 64-bit, x86 processor
Android 4.3 Jelly Bean
1GB RAM
16GB storage
microSDHC card slot for up to 64GB of removeable storage
16GB of Asus WebStorage
802.11b/g/n WiFi
Bluetooth 4.0
1 USB 2.0 port
5060mAh battery

http://liliputing.com/2014/04/asus-transformer-pad-tf103c-bay-trail-windows-tablet.html

ECS TA80TA1 Bay Trail Android tablet hits the FCC

It’s expected to feature an Atom Z37353E quad-core CPU, 1GB of RAM, 16GB of storage, and a 1280 x 800 pixel display. It has a micro USB port, microSD card slot, and micro HDMI port.

Thanks to the FCC, we can see some of those components on the motherboard, along with a 4150mAh battery, front and rear cameras, and stereo speakers on the back of the tablet.

There’s no word on how much the tablet will cost, or what name it will be sold under if and when it goes on sale in the States.

http://liliputing.com/2014/03/ecs-android-tablet-is-powered-by-bay-trail.html

Acer A1-840 Bay Trail Android coming soon?

Acer appears to be testing a new 8 inch Android tablet with an Intel Atom Bay Trail processor. Specs and test results for the unannounced Acer A1-840 showed up at the GFXBench website recently, and they paint a picture of a low-cost tablet powered by Intel.

http://liliputing.com/2014/04/acer-a1-840-bay-trail-android-coming-soon.html

 

[blackened23]

Very very nice. These are the first 64 bit android devices on the market, correct? Erm. Does the latest android release support 64 bit? I don't recall. I do know the other SOC Vendors are expected to be 32 bit until next year from what I remember.

Now with full android I expect intel to do that much better with BT. Seems to be a great SOC, now that it isn't limited to Windows i'd expect it get some android based design wins. There were rumors that the next Google Nexus for 2014 would be Bay Trail based, but who knows. Would be an absolutely amazing design win for Intel if true, as it is hands down the highest volume android tablet available.

 

[coercitiv]

Yes , about all the power is dissipated in the SoC, the SSD/HD, RAM and VRMs share is about negligible when the system is fully loaded, they should account for 5-7W at most

Tom's Hardware used a conventional 500GB WD Black in their review, and that alone uses 6W while idle.

If you take that 48W power consumption under mixed load and consider PSU efficiency around 75%, you end up with 36W board power consumption. Shave 6W for the HDD, 2W for the RAM (assuming 1 stick), 2W for whatever else on that board, and you end up with 26W. Factor in VRM efficiency and you just got under 25W for the SOC power usage.

It looks like they went out of their way to make these measurements irrelevant, but they are... in the realm of possibility

 

[Abwx]

Post 102: Aren't these pretty much the same delta power numbers that everything else shows?

Post 103: This shows the 5350 having a 17W delta between idle and CPU load. I believe that's the highest yet? As well, the power supply they used there appears to have something around a 10W base power loss. And yes, the J1800 idle power consumption is likely quite similar to the J1900, but load characteristics/performance are going to be markedly different due to the fact that the J1800 runs its two cores at higher frequency than the quad core J1900. So there's really nothing to be learned from attempting to compare the load delta power numbers on the J1800 to the 5350 - it'd be pretty pathetic if the 5350 didn't win by a nice margin.

You are right for the 17W delta and this is about the delta measured by THG so they have at least one relevant number assuming the reader take notice.

Now about the german site and its efficient PSU.

Extrapolating from its review in the french site where it show close to 90% efficency at 80W the PSU seems capable of over 80% efficency at 10-30W power levels inducing losses in the 3-6W range in the PSU with the remaining , 7-24W being effectively absorbed by the plateform.

Since the german site provide numbers for BT as well we can assume confidently from thoses figures and once we substract 5W fot the rest of the plateform that a J1800 consume about 4W at iddle and 14W when loaded by luxmark , numbers being 6W and 23W for Kabini.

http://www.golem.de/news/kabini-fuer-am1-im-test-gesockelte-jaguare-fuer-jedermann-1404-105670.html

 

[Khato]

Yes , about all the power is dissipated in the SoC, the SSD/HD, RAM and VRMs share is about negligible when the system is fully loaded, they should account for 5-7W at most so if a system really pump 47W then it means that the SoC dissipate about 40W and heatspreaders like Kabini s cant dissipate that much heat without the fan going crazy, even worse for BT since the passive cooling wouldnt allow to dissipate 20W or even 15W continuously.

Last I checked pretty much every SoC out there was doing everything it could to get idle power consumption as low as possible. I'd be surprised if the SoC accounts for even 1W of system power when idle. Hence why the above sounds quite ludicrous since, oh yeah, the THG review had idle levels of 27.5W for the J1900 and 30.9W for the 5350. So you're basically saying that the rest of the system 'should account for 5-7W at most' and these SoCs are using 20W+ idling? Or can we agree that SoC power consumption is basically the delta between the load and idle power numbers, at which point the THG numbers have a maximum of 17W for the 5350 and 7W for the J1900, both of which are pretty much exactly where they should be. And 7W is quite reasonable for a passive heatsink.

 

[Abwx]

Tom's Hardware used a conventional 500GB WD Black in their review, and that alone uses 6W while idle.

If you take that 48W power consumption under mixed load and consider PSU efficiency around 75%, you end up with 36W board power consumption. Shave 6W for the HDD, 2W for the RAM (assuming 1 stick), 2W for whatever else on that board, and you end up with 26W. Factor in VRM efficiency and you just got under 25W for the SOC power usage.

It looks like they went out of their way to make these measurements irrelevant, but they are... in the realm of possibility

Thank for the precision, this does help understand that THG methodology is good for big systems but not for thoses tiny PCs.

Last I checked pretty much every SoC out there was doing everything it could to get idle power consumption as low as possible. I'd be surprised if the SoC accounts for even 1W of system power when idle. Hence why the above sounds quite ludicrous since, oh yeah, the THG review had idle levels of 27.5W for the J1900 and 30.9W for the 5350. So you're basically saying that the rest of the system 'should account for 5-7W at most' and these SoCs are using 20W+ idling? Or can we agree that SoC power consumption is basically the delta between the load and idle power numbers, at which point the THG numbers have a maximum of 17W for the 5350 and 7W for the J1900, both of which are pretty much exactly where they should be. And 7W is quite reasonable for a passive heatsink.

I m saying that THG review doesnt allow for an easy isolation of the parts respective contribution to power consumption, i m doing all my assumptions based on golem.de review wich allow to clearly extrapolate realistic numbers and for informations i ll explain why they seem to me reliable.


First their review of J1800, Athlon 5350, let see the numbers :

http://forums.anandtech.com/showpost.php?p=36250715&postcount=103

Nothing extraordinary but they specify the PSU used :

Be Quiet Straight Power E9 mit 400 Watt

Let s see what this PSU is capable of :

http://www.ginjfo.com/dossiers/test...ht-power-e9-400w-silence-et-economie-20120112

According to the site at 20% load it still has 87% efficency but it could be an exceptional sample.?..

Yes but it happens than another site used exactly the same PSU and they measured exactly the same power comsumption for Kabini wich say that this excellent efficency is reproducible on a large scale manufacturing.

http://www.eteknix.com/amd-kabini-am1-athlon-5350-apu-fs1b-review-winning-sub-60/11/


Hence golem.de numbers are a good basis for said estimation as the available technical infos are enough to extrapolate valuable numbers without brain torture.

You are right for the 17W delta and this is about the delta measured by THG so they have at least one relevant number assuming the reader take notice.

Now about the german site and its efficient PSU.

Extrapolating from its review in the french site where it show close to 90% efficency at 80W the PSU seems capable of over 80% efficency at 10-30W power levels inducing losses in the 3-6W range in the PSU with the remaining , 7-24W being effectively absorbed by the plateform.

Since the german site provide numbers for BT as well we can assume confidently from thoses figures and once we substract 5W fot the rest of the plateform that a J1800 consume about 4W at iddle and 14W when loaded by luxmark , numbers being 6W and 23W for Kabini.

http://www.golem.de/news/kabini-fuer-am1-im-test-gesockelte-jaguare-fuer-jedermann-1404-105670.html

 

[Khato]

According to the site at 20% load it still has 87% efficency but it could be an exceptional sample.?..

Yes but it happens than another site used exactly the same PSU and they measured exactly the same power comsumption for Kabini wich say that this excellent efficency is reproducible on a large scale manufacturing.

So you're surprised that the exact same power supply, motherboard, and processor used by two different review sites measured at the same idle power? Why exactly?

As well, none of the reviews give adequate information to determine actual SoC idle power versus other system components, so why do you care about the precise idle power? Other than comparisons with all other components as similar as possible there's nothing of interest to be found there. And the majority of reviews indicate that idle power is same/better for the Celeron J1800/J1900 than the 5350.

The only other item of interest when it comes to power consumption is the delta between idle and load. For that all that matters is that all other components are kept as similar as possible - doesn't matter if they use a 100W or a 1kW power supply the results in the delta power consumption will be comparable.

 

[CryNet]

PassMark

Atom z3745

http://www.passmark.com/baselines/V8/display.php?id=20177836568

z3740
http://www.passmark.com/baselines/V8/display.php?id=21968332795


think as rumored a surface 3 mini maybe could be z3745? or better you think?

I'm hoping some w8x64 tablet 8"

 

[Nothingness]

These are the first 64 bit android devices on the market, correct? Erm. Does the latest android release support 64 bit? I don't recall. I do know the other SOC Vendors are expected to be 32 bit until next year from what I remember.

The current Intel release of Android has only the kernel running in 64-bit mode, all the rest is 32-bit.
https://01.org/android-ia/q-and-a?keys=&page=2

What part of Android on Intel Architecture is 64-bit?

Android* as released from Google*, runs on a Linux* kernel. The kernel used by Android on Intel Architecture is a 64-bit kernel Until Google releases the 64-bit version of Android, Android on Intel Architecture will run 32-bit Android on a 64-bit Linux kernel.

So right now it's only an Intel thing, and only the kernel.

There were rumors that the next Google Nexus for 2014 would be Bay Trail based, but who knows. Would be an absolutely amazing design win for Intel if true, as it is hands down the highest volume android tablet available.

My understanding is that Samsung is the largest Android tablet vendor, and their recently announced tablets have no Intel chips. But I still hope we'll see a BT-based Nexus, as I'm tired to see sh*t like the above-mentioned TF103, 1280x800 on 10" yeah nice