Intel & AMD: Rated TDP vs. Actual TDP?

[OCGuy]

So you will admit then the only thing being "vastly exceeded" here is the misinformation in your other threads?

 

[AtenRa]

Techspot. Go search yourself.
12W lower power consumption under load than the 25W Athlon 5350. Probably a different MB+J1900 combo than the one used in the AMD-biased review you picked.

Funny that i dont seam to be able to find any specs of the systems tested in that Techspot review you quoted.

Edit: Not to mention his AMD-Biased review takes power consumption measurements from Cinebench for the CPU and 3D Mark for the IGPU when your unbiased review measures Prime 95 for CPU and Prime 95 + Furmark.

 

[piesquared]

Techspot. Go search yourself.
12W lower power consumption under load than Athlon 5350. Probably a different MB+J1900 combo than the one used in the AMD-biased review you picked.

Well you read it, so just tell us what the power supply was? Isn't it appropriate to link to the site you are linking to? Are you trying to hide something?

The reason I'm asking is because almost all of the reviews of the 5350 were using ridiculous power supplies that are completely over speced. And you were cross referencing charts with an appropriate pico power supply to something that we don't know and hodgepodging it all together.

 

[SPBHM]

the only useful analysis you can make with these numbers is to compare idle vs load, because TDP is not supposed to cover the entire PC,

also you are comparing different motherboards, even if you try to keep the rest unchanged, and it can make a huge difference for low TDP CPUs,

 

[AtenRa]

Well, the TDP cannot be measured from power consumption graphs. It is the same thing with the FX CPUs that people got fascinated with a few months ago.

Those graphs posted above, measured Power Consumption for the entire system, idle to full load delta also doesnt provide any meaningful data towards TDP evaluation. It only shows how power efficient the platform is to lower its power consumption when Idle.

And for the Billionth time, TDP is not power Consumption.

 

[BSim500]

the only useful analysis you can make with these numbers is to compare idle vs load, because TDP is not supposed to cover the entire PC,

also you are comparing different motherboards, even if you try to keep the rest unchanged, and it can make a huge difference for low TDP CPUs

Indeed, I mentioned the same thing on page 1. Different motherboards can make a huge difference. Example : AMD's 5350 uses less power idle than a 10w J1900 Celeron in OP's test, and yet 20% more power idle & load than a 60% faster 17w i3-3217U (NUC D33217CK) in another test:-
http://www.bit-tech.net/hardware/2014/04/24/amd-athlon-5350-kabini-review/8

Comparing entirely different platforms, motherboards, etc, based on "CPU only" TDP is a bit naive to say the least...

 

[SPBHM]

Well, the TDP cannot be measured from power consumption graphs. It is the same thing with the FX CPUs that people got fascinated with a few months ago.

Those graphs posted above, measured Power Consumption for the entire system, idle to full load delta also doesnt provide any meaningful data towards TDP evaluation. It only shows how power efficient the platform is to lower its power consumption when Idle.

And for the Billionth time, TDP is not power Consumption.

TDP and power usage are not the same but definitely related, the difference between idle and load is much closer to a relevant information for TDP than the entire PC power usage, which was how this topic started.

 

[Shivansps]

WTH, Power Consumption has nothing to do with SOC TDP.
TDP means "Thermal design power", Thermal != Power Consumption.

Its a calculated value depeding the loads, If you wanna read the correct value use Open Hardware Monitor or HWinfo, the value is read as "CPU Package Power" and its primary used by turbos, and the value limit can be changed, up and down, by OEM.

Power is related, because as lower the TDP is lower the power it uses too, less heat, less power, but there is not really a good way to figure out the exact value, funny enoght, Intel adopted SDP beucase people where calling TDP as power usage, and Z3770 SDP seems to be consistant with the power it uses.

 

[Lepton87]

Do note, though, that TDP isn't about power consumption specifically, but rather the maximum heat the cooling system needs to dissipate.

What is the difference? That's a serious question. Is that even measurable? How much of the consumed energy does not end up as heat? The chip does not do any mechanical work so I'd wager that it is something in the order of milliwatts at most.

 

[Abwx]

Idle / CPU Load difference:
- Celeron J1900: 5W
- Athlon 5350: 16W

Idle / CPU Load difference:
- Celeron J1900: 5W
- Athlon 5350: 15W

Idle / CPU Load difference:
- Celeron J1900: 4,4W
- Athlon 5350: 16,7W

Now Cinebench 11.5 performance:
Celeron J1900: 1.8
Athlon 5350: 2.0

http://us.hardware.info/reviews/533...ap-desktop-platforms-benchmarks-cinebench-115

What did you say again?



Mods have been very tollerant with him, hardly adds something useful and usually derails threads and gets involved in flame wars.

The first site doesnt detail their testbed, curious isnt it , as for PC lab you forgot to add that the Bay trails use an ISK box with a 90W PSU while the Kabini usev a 550W PSU but PClab did test the AM1 plateform with thev ISK 110, see the result , it match hardware.info numbers :

 

[Abwx]

Well you read it, so just tell us what the power supply was? Isn't it appropriate to link to the site you are linking to? Are you trying to hide something?

The reason I'm asking is because almost all of the reviews of the 5350 were using ridiculous power supplies that are completely over speced. And you were cross referencing charts with an appropriate pico power supply to something that we don't know and hodgepodging it all together.

There s another review using a same efficient pico psu for both set ups at computerbase.de and their numbers, very low, tell the same story..

idle Videoplay Videoconv. gaming

Athlon 5350 + ITX-Mainboard 8,0 Watt 11,0 Watt 21,0 Watt 20,5 Watt

Celeron J1800 + ITX-Mainboard 9,0 Watt 10,0 Watt 16,5 Watt 15,5 Watt

Celeron J1900 + ITX-Mainboard 12,0 Watt 13,0 Watt 20,5 Watt 19,5 Watt

http://www.computerbase.de/2014-04/amd-athlon-5350-kabini-sockel-fs1b-test/3/

 

[Abwx]

Could Intel be better under gaming load, because that s what the quote above is implying but neverless from the numbers above it is clear that AMDs Kabini is more efficient at idle and under load and to answer the topic question it is obvious that AMD s TDP numbers are exagerated to the same extent than Intels are understimated.

 

[AtenRa]

What is the difference? That's a serious question. Is that even measurable? How much of the consumed energy does not end up as heat? The chip does not do any mechanical work so I'd wager that it is something in the order of milliwatts at most.

The difference is that you can raise or lower the TDP depending of the Temperature you will like to operate your IC (CPU).

That means that if you like your BayTrail to never reach 50c then you should raise the TDP to lets say 20W.

Or,

If you dont have a problem with your Kabini to reach 100c then you can lower your TDP to 10W.



Now obviously a 100W TDP CPU will consume more power and Energy than a 25W TDP but TDP doesnt measure Power consumption. Especially total system power consumption.

 

[witeken]

How can it be idling at 0 when the compared chip is idling 7W lower?

If Silvermont were idling 7W higher than the competition, (1) it wouldn't be suited for tablets and soon phones, and (2) the battery life would be insanely short.

 

[AtenRa]

If Silvermont were idling 7W higher than the competition, (1) it wouldn't be suited for tablets and soon phones, and (2) the battery life would be insanely short.

The J1800 and J1900 is not the same SoC or the same platform as BayTrail-T Z3770. And yes they may have higher Idle power consumption than the competition.

 

[witeken]

The fact remains, Kabini is using significantly less power at idle and practically even on load while outperforming bay trail. The 5350 has a 25W TDP, the j1900 has a 10W TDP.

Please, learn what "fact" means. Let's quote a reliable source, AnandTech:

The single threaded performance numbers are just barely ahead of AMD's Jaguar based Kabini SoC. The big difference however is power. I had Intel measure SoC power at the board level while running a single threaded Cinebench 11.5 run on the Atom Z3770 and saw a range of 800mW - 1.2W. AMD on the other hand lists the A4-5000's SoC/APU idle power as 770mW. I don't have equivalent data for AMD, but with the A4-5000 idling at 770mW, it's safe to say that SoC level power consumption is lower on Bay Trail.

Here you have the facts.

 

[AtenRa]

Please, learn what "fact" means. Let's quote a reliable source, AnandTech:




Here you have the facts.

Well, that was for Kabini

Mullins has 20% lower leakage and better thermal and power management, so they may have close to the same single threat power consumption. But again it still depends on the application. BayTrail may be more power/energy efficient in Cinebench than Mullins. That doesnt mean it will be more efficient in general. Mullins may have better power/energy efficiency in other applications.

 

[witeken]

What is the difference? That's a serious question. Is that even measurable? How much of the consumed energy does not end up as heat? The chip does not do any mechanical work so I'd wager that it is something in the order of milliwatts at most.

Indeed, much more than 99% of the energy becomes heat. I and others already explained why TDP doesn't necessarily have something to do with specific power consumption.

 

[Abwx]

Please, learn what "fact" means. Let's quote a reliable source, AnandTech:

Here you have the facts.

That s a capped 3770 not a J1900, besides Anand did measure nothing, all he was doing was reading the numbers on Intel s measurements hardware screens inside an Intel office, that s where this "review" was performed as well as all other review when the chip was launched...

 

[TrulyUncouth]

Please, learn what "fact" means. Let's quote a reliable source, AnandTech:




Here you have the facts.

His definitely seems more factual than what you posted. Its ironic to try calling someone out on their facts then post another person's conjecture as fact... This is what we get with a flamebait thread like this one.

Seems out of all these claims, the only person to post a real link is Abwx for the computerbase review. http://www.computerbase.de/2014-04/amd-athlon-5350-kabini-sockel-fs1b-test/3/

Assuming they didn't screw up testing it looks like CERTAIN SKU's of Baytrail do in fact use more power than Kabini. I don't see why that is so hard to believe or admit when it is true- as it appears it is.

However, this doesn't necessarily prove that J1900 uses more than 10 watts- in fact it looks like its close to, or a little over 10W... so quit splitting hairs on that one.

The long and short of it is that Global Foundry's version of desktop Kabini is surprisingly efficient and desktop Baytrail is less efficient than a lot of people assumed. Either way you guys are arguing over a couple watts- I think its time for everyone to sing Kumbaya.

 

[Abwx]

Seems out of all these claims, the only person to post a real link is Abwx for the computerbase review. http://www.computerbase.de/2014-04/amd-athlon-5350-kabini-sockel-fs1b-test/3/

That is a very professional site as not only their numbers are the ones that are the closest to the plateforms real power comsumption and actualy we can even extract the power losses and exact comsumption since they also published the efficency curves of their PSU.

LC-Power LC75ITX 75 Watt

http://www.computerbase.de/2014-02/lc-power-lc75itx-test/3/

 

[witeken]

The J1800 and J1900 is not the same SoC or the same platform as BayTrail-T Z3770. And yes they may have higher Idle power consumption than the competition.

The chip itself is the same, and that's what we're comparing, right?

 

[Abwx]

The chip itself is the same, and that's what we're comparing, right?

One, a 3770 is not a J1900 and two Computerbase did the measurements themselves while Anand just repeated numbers provided by Intel in an Intel office and wich are not the total power comsumption anyway..

 

[AtenRa]

The chip itself is the same, and that's what we're comparing, right?

The Chip is not the same, the dies are different and the manufacturing process may be different. Also the platform is different, the one is a Tablet with a different power delivery system and the other is a Desktop. Not comparable.

So people should not directly compare the two or should point out the differences.

Also, comparing a Tablet Baytrail to Desktop Kabini is not apples to apples.

 

[witeken]

His definitely seems more factual than what you posted. Its ironic to try calling someone out on their facts then post another person's conjecture as fact... This is what we get with a flamebait thread like this one.

I didn't post anything as facts. I just meant that this is the best information you can get (the facts are in the quote), because using different motherboards etc. and drawing conclusions out of that doesn't really follow the scientific method. So the best thing you can do is compare the idle/load delta, or get some good measurements of the chip alone.

In the case of Kabini, it idles at 0.77W, which is about what a single Silvermont thread consumes under load, so the conclusion that Silvermont uses much less energy than Kabini/Jaguar is an extremely easy one to make.

You say that his post was more factual. Let's take a look...

The fact remains, Kabini is using significantly less power at idle and practically even on load while outperforming bay trail. The 5350 has a 25W TDP, the j1900 has a 10W TDP.

1. Kabini uses less power at idle.
A few lines above, I already showed that that is false (and it is also mentioned in the preview of Puma+).

2. Jaguar uses the same power under load (while outperforming Silvermont).
Unfortunately, we don't have precise measurement of Jaguar under load. However, we can compare the idle/load delta. And since the chips idle at less than 1W and the delta of Kabini is like 5W bigger than Silvermont, we can again safely conclude that Silvermont uses less energy under load.

However, we made it much too difficult. This is not rocket science, we can reduce the question to a much simpler version. Silvermont has 22nm Tri-Gate transistors, while Kabini is made on regular 28nm transistors. Of course Silvermont consumes less energy at idle and load.

Seems out of all these claims, the only person to post a real link is Abwx for the computerbase review. http://www.computerbase.de/2014-04/amd-athlon-5350-kabini-sockel-fs1b-test/3/

Assuming they didn't screw up testing it looks like CERTAIN SKU's of Baytrail do in fact use more power than Kabini. I don't see why that is so hard to believe or admit when it is true- as it appears it is.

However, this doesn't necessarily prove that J1900 uses more than 10 watts- in fact it looks like its close to, or a little over 10W... so quit splitting hairs on that one.

The long and short of it is that Global Foundry's version of desktop Kabini is surprisingly efficient and desktop Baytrail is less efficient than a lot of people assumed. Either way you guys are arguing over a couple watts- I think its time for everyone to sing Kumbaya.

I don't see how you come to that conclusion, but I'm very sure it's wrong.