Here's measurements of just the CPU and it's VRC:
http://ht4u.net/reviews/2011/intel_sandy_bridge_e_hexa_core/index17.php
http://ht4u.net/reviews/2011/intel_sandy_bridge_e_hexa_core/index17.php
Trinity vs Ivy Bridge
- Thread starter rpg.314
- Start date
You got the percentage calculation wrong on the 2600K. 63.6/95 = 67%
You have to add some 20W to the consumption of Nehalem and most probably Sandy Bridge, because the memory controller is not powered from the separate "ATX12V" lead. link (Also note that the memory controller and much of the IO is off-chip in case of the Core 2 CPU's.)
For two simple reasons. One: you have said here that the "almost half" of the consumption of a 35W TDP part cannot be 17W, which is wrong. Is it clear now? Then, you've said: "an Intel 95W TDP chip will consume less than 95W; an AMD chip typically consumes more." The latter one is plain wrong, again, right? Yes, as I've admitted already that Intel usually consumes less, as well, but there was and are exceptions (perhaps mainly on the Xeon line). All in all (and this is the relation between these off-topics and matter of the 17W Trinity), if they've said the part consumed "almost half" of that of a 35W TDP one, it indeed could be 17W. Could you acknowledge it already, so that we can move on? That if they were speaking about the actual part in the notebook, is another topic.
I think it's obvious that all the claims on the slide is referring to the same product. So, it's 4-core and 17W TDP. And I see no reason why couldn't it be in that notebook on the show, as well. Did you see what even a Brazos is capable of, with only two cores and a lesser IGP?
I doesn't need any discussion, regarding AMD parts. It is a known thing.
Really? All these discussions could be saved if you were more knowledgeable about AMD's TDP.
Last edited by a moderator:
If you can provide better numbers, so be it. My miscalculation for i7-2600k (mentioned above) means that, even if the measurement was off by 20W, it would still be in-line with the FX-8150. Nothing you've provided shows me that Intel nor AMD's power consumption versus TDP are far-separated.
Did you watch the video? The AMD PR person mentioned the CES Trinity unit was half the power of an undisclosed former mobile part. He did NOT mention if his claim was thermal design power, or if it was power consumption. For his claim to have any validity, he had to be using the same units of measure (comparing TDP of a former part to Actual power of a current part is an outright lie.) Thus, it is still my opinion that the new Trinity part is NOT anything near half the power of the former part.
Further, have you not watched the (thrice-linked in this very thread) video of AMD spokesperson telling us definitively that the Trinity display was using a mainstream mobile part? Their mainstream parts are TDP of >=35W, and the most likely former candidate would be the Llano which is also a >= 35W TDP part. My opinion still has not changed, I do not believe that Trinity is half the power of Llano. Sorry, you haven't convinced me. Mostly because you've either put up marketing slides, or tried to make some passing commentary on how AMD's TDP is somehow more reflective of -- something? -- than Intel's.
Interesting, because you told me I was exactly backwards and Intel always consumes more and AMD always consumes less. As it turns out, I corrected my own claim when finding real data to look at:
So, I came out and said that it looks flat after all, and yet here you are telling me I'm wrong and never admitted it... Really? I'm pretty sure you're projecting, because:Albuquerque said:
The burden of proof is on you. It is you, not I, who are making the accusations that somehow AMD's TDP is more 'relevant' than Intel. For you to make this claim, you will now go find the requisite material to prove it. If this is such a well known fact as you say, then there will be data overflowing from countless websites out there, although somehow I'm not finding it as easily as you suggest.dess said:
Fact: We now have three sites that give some attempt at metering CPU power consumption. IN all three, when comparing against their rated TDP among chips who are roughly performance equivalent, AMD and Intel are proving to be roughly equal in terms of relation between actual consumption and TDP rating.
Fact: AMD stated the CES Trinity demo was a mainstream mobile part, not ULV.
Fact: AMD and Intel only rate processor wattage in terms of TDP, so a "17W processor" in PR terms would mean a "17W TDP Processor."
Fact: The CES demo was not run on a 17W TDP processor.
Fact: There will be 17W TDP Trinity chips
Possibility: There may not be quad-core Trinity chips at 17W TDP. Your ULV slide does indeed specify that quad core will be available in the ULV space, but they do not say that it will be in the 17W TDP profile. They say ULV "starts at 17W".
Opinion, based on what we know about GloFo and AMD's ability to build processors: Trinity at CES is likely NOT half the power consumption of the prior Llano 35W TDP part. It's the same lithography process, with significantly more transistors thanks to more GPU and CPU cores. Sure, L3 cache goes missing, but is that going to save half of the power draw? Nope. Power doesn't go down in a scenario where your lithography stays the same, your computation power goes up, clocks stay flat, and you add transistors... Although, strictly speaking, it also doesn't mean TDP had to go up either.
Last edited by a moderator:
Interesting, lends credence to Dess's claim of ~20W possibly gone missing from the SB platform due to uncore neneds thatr aren't fed by the ATX12V. Also interesting, that article includes FX-8150 and old Phenom II X4 and X6 data, which also appear to be partially affected when referencing against the ATX12V data I found.
To my eye, it looks like both companies draw some amount of power that isn't fed directly by the ATX12V, so again, it's looking like a (generally) flat comparo to me. I mean, even if we say only Intel does it (which isn't correct, but whatever) we'd still end up with approximately the same percentages when calculating actual draw vs TDP. Not that this artificial benchmark has any true bearing on reality, but whatever
That table was painful to create, and then I go and muck it with a bad calcuation. Bleargh! Thanks for the correction; I've updated my post...
ronvalencia
Newcomer
AMD A8 3500 vs Intel Core i5-2520M
From
http://www.tomshardware.com/reviews/a8-3500m-llano-apu,2959-22.html
Sigh.
It doesn't matter. Each company has a different way of measuring it; an Intel 95W TDP chip will consume less than 95W; an AMD chip typically consumes more. No matter what, AMD's own people rate the chip in that demo unit as one of thier "mainstream", which is not in the 17W class.
Thus, it doesn't matter how you personally want to spin it, the only people perpetuating the 17W myth are misinformed at best. Now that you know, you can help stop the myth
From
http://www.tomshardware.com/reviews/a8-3500m-llano-apu,2959-22.html
The power use in the above graph is a result of a controlled test on an external monitor, so we repeated this metric again, this time using the laptop's own display. The A8-3500M laptop lasted two hours and 12 minutes. Assuming the Intel laptop used the exact same battery, it would run for one hour and 22 minutes.
This is very impressive. Not only does the A8-3500M get about twice as much time out of its battery, it does so while delivering far better graphics performance. The implications of this are profound: a Llano laptop user might be able to play a mainstream 3D game for an entire two-hour flight with decent frame rates, while the Intel Core i5-based platform would only last for half of the flight with choppy performance. There does, in fact, seem to be validity in AMD's excitement over its improved power story, and of course this is a real advantage when it comes to mobile devices.
This is very impressive. Not only does the A8-3500M get about twice as much time out of its battery, it does so while delivering far better graphics performance. The implications of this are profound: a Llano laptop user might be able to play a mainstream 3D game for an entire two-hour flight with decent frame rates, while the Intel Core i5-based platform would only last for half of the flight with choppy performance. There does, in fact, seem to be validity in AMD's excitement over its improved power story, and of course this is a real advantage when it comes to mobile devices.
Damnit, I thought that TH review was laptop vs. laptop. Those mini desktop systems tend to use somewhat more power than laptops because they turn off some power saving components and sometimes use desktop components.
http://www.legitreviews.com/article/1636/5/
http://techreport.com/articles.x/21099/7
The first review is interesting that they show 3DMark06 results when running on battery and AC. Other sites show the performance numbers on AC, but battery life is of course measured on DC.
Similar performance with 1.6x better battery life or 40% better performance on AC.
Albuquerque, slide down in the page says "Quad core and 17W." Of course that doesn't really tell what the CES test system used.
My main concern here was your claim that AMD typically under-rates their CPU's TDP. It's clear already it's not the case, the consumption at high loads is certainly between the value of the given TDP and the TDP on class below. Now, on Intel's side, look at the i3-2100, its real high-loaded consumption is some 26W (your source; it's unknown if it's including memory controller or not), yet Intel lists it as a 65W part. On the other hand, there is i7-880: 102W from ATX12V (your source, again) + 20W for the memory controller (it's Nehalem) = 122W, yet Intel lists it with 95W.
I don't have time to analyse them, I didn't comment on those, just corrected some claims it seemed was related to the topic at hand.
Not really, as I did not say you didn't admitted this one, I just answered your question why we're discussing it.
No. Note that it was you who "made accusations" first that AMD's TDP is "typically" under-rated... (And that of Intel never.)
Not really.
So it's 17W at most and can be even lower. (Not more as you first claimed.)
It's not a fact, more like your assumption only.
And I think the claims of "All the features of a premium 35W 'Trinity' notebook" and "The only available premium quad core, low voltage APU" is implicitly true for the whole ULV line, including the 17W TDP part.
The most important factor in max. power consumption for a given part is voltage and then clock rate. The 17W Trinity is an ULV (Ultra Low Voltage) part, most probably at a relatively low clock rate...
Also note that the max. consumption can go down even upon a respin in some cases. GloFo could falso fine-tune their processes in the meantime.
Last, AFAIK the ULV and the SHP (super high performance) processes are two distinct ones.
Last edited by a moderator:
ronvalencia
Newcomer
My post for "an AMD chip typically consumes more. No matter what" comment.
AMD A8-35x0M rated at 35 or 45 watts.
The posted TH graphs shows that the 17 watts Trinity is reachable.
D
Deleted member 13524
Guest
People, I think it's time to stop trying to correct Albuquerque
Everyone pretty much already knows his statement about AMD going above the TDP is wrong, and the 17W Trinity has a 2-module Piledriver.
It's just that he's obsessed with this "score" of his and he'll just keep doing loops of awfully random information (like calculating average power consumption on Intel CPUs, lol) until he's somehow proven right.
Now, just to make a quick review of all the info we know so far regarding the performance of the 17W part:
Hothardware took this video during CES:
http://www.youtube.com/watch?feature=player_embedded&v=lsmTDb-Mlws
At approx. 1m40s, the AMD rep. says it's a quad-core.
Regarding the TDP, the people to took the video said:
Next, there's this Dailytech interview with AMD's director of global product marketing, John Taylor where they say:
Then there's the newest leaked slides from computerbase.de:
Right here it's already pretty obvious: the 17W is a 2-module chip.
There's also the pictures: the 17W BGA APU (on the left, AMD confirmed at the CES floor that the BGA is 17W) is the same size as the socket 35W APU. There's no production of a single-module Trinity so far, at least that we know of.
The 17W Trinity could be laser-cut with only 1 module working, but besides all the other confirmations, AMD has always been making the lowest-voltage models with all the cores: the ~6W Brazos Z-01 is a 1GHz dual-core, not a single-core; the C-60 (dual-core, 1GHz) has the same TDP as C-30 (single-core, 1,2GHz), etc.
As pretty much every CPU architecture introduced during past 5 years, more cores at lower speeds seems to have better power consumption than less cores at higher speeds, so there's really no reason to assume AMD would need to cut down a module in Trinity to achieve the 17W TDP.
There's also another factor here: the demo from AMD at CES, with the Dirt3 @ Low + video conversion + video playback can be done with a 35W Llano 3500M, as it was reproduced with users in other forums using laptops with that APU. The only differentiation here is the power consumption ("almost half the power consumption", as stated by the PR).
The performance of the 35/45W Trinity parts has been gradually uncovered (+25-50% performance over Llano A series), but the 17W part hasn't been discussed that much.
A couple of pages back I did this very-wild assumption that the 17W Trinity could be bringing a 400% performance bump over the E-450 Brazos in the same power consumption class.
While it does sound way too wild, the truth is that AMD is claiming a similar performance to the 35W Llano A-series. That means it could go from an A4 3300M (dual-core 1.9-2.5GHz, 240sp GPU @ 444MHz) to an A8 3520M (quad-core 1.6-2.5GHz, 400sp GPU @ 444MHz).
The latest would best the E450's GPU power by 3 to 4x easily, while CPU power would go from 2 to 3x, depending on workload.
Maybe a fair assumption is that it should be somewhere in the A6 3400M zone. That's a quad-core 1.4-2.3GHz CPU with a 320sp 16TMU 8ROP GPU @ 400MHz.
So this 17W Trinity could be around a ~1.2/1.3GHz (Turbo up to 2.0GHz?) quad-core/2-module Piledriver along with a GPU with 256-320sp VLIW4 16TMU 8ROP @ ~400MHz.
While not a "steady" 4x performance bump over E-450, this would still be damn impressive to carry in a 11,6-12" thin form factor.
Everyone pretty much already knows his statement about AMD going above the TDP is wrong, and the 17W Trinity has a 2-module Piledriver.
It's just that he's obsessed with this "score" of his and he'll just keep doing loops of awfully random information (like calculating average power consumption on Intel CPUs, lol) until he's somehow proven right.
Now, just to make a quick review of all the info we know so far regarding the performance of the 17W part:
Hothardware took this video during CES:
http://www.youtube.com/watch?feature=player_embedded&v=lsmTDb-Mlws
At approx. 1m40s, the AMD rep. says it's a quad-core.
Regarding the TDP, the people to took the video said:
hothardware said:
Next, there's this Dailytech interview with AMD's director of global product marketing, John Taylor where they say:
DailyTech said:
Then there's the newest leaked slides from computerbase.de:
Right here it's already pretty obvious: the 17W is a 2-module chip.
There's also the pictures: the 17W BGA APU (on the left, AMD confirmed at the CES floor that the BGA is 17W) is the same size as the socket 35W APU. There's no production of a single-module Trinity so far, at least that we know of.
The 17W Trinity could be laser-cut with only 1 module working, but besides all the other confirmations, AMD has always been making the lowest-voltage models with all the cores: the ~6W Brazos Z-01 is a 1GHz dual-core, not a single-core; the C-60 (dual-core, 1GHz) has the same TDP as C-30 (single-core, 1,2GHz), etc.
As pretty much every CPU architecture introduced during past 5 years, more cores at lower speeds seems to have better power consumption than less cores at higher speeds, so there's really no reason to assume AMD would need to cut down a module in Trinity to achieve the 17W TDP.
There's also another factor here: the demo from AMD at CES, with the Dirt3 @ Low + video conversion + video playback can be done with a 35W Llano 3500M, as it was reproduced with users in other forums using laptops with that APU. The only differentiation here is the power consumption ("almost half the power consumption", as stated by the PR).
The performance of the 35/45W Trinity parts has been gradually uncovered (+25-50% performance over Llano A series), but the 17W part hasn't been discussed that much.
A couple of pages back I did this very-wild assumption that the 17W Trinity could be bringing a 400% performance bump over the E-450 Brazos in the same power consumption class.
While it does sound way too wild, the truth is that AMD is claiming a similar performance to the 35W Llano A-series. That means it could go from an A4 3300M (dual-core 1.9-2.5GHz, 240sp GPU @ 444MHz) to an A8 3520M (quad-core 1.6-2.5GHz, 400sp GPU @ 444MHz).
The latest would best the E450's GPU power by 3 to 4x easily, while CPU power would go from 2 to 3x, depending on workload.
Maybe a fair assumption is that it should be somewhere in the A6 3400M zone. That's a quad-core 1.4-2.3GHz CPU with a 320sp 16TMU 8ROP GPU @ 400MHz.
So this 17W Trinity could be around a ~1.2/1.3GHz (Turbo up to 2.0GHz?) quad-core/2-module Piledriver along with a GPU with 256-320sp VLIW4 16TMU 8ROP @ ~400MHz.
While not a "steady" 4x performance bump over E-450, this would still be damn impressive to carry in a 11,6-12" thin form factor.
french toast
Veteran
So basically 17w trinity should have the gpu power of an xbox 360! pretty cool.
This to me seems to be the area AMD was designing for all along, cpu/gpu modules together on one die, brilliant idea, now picture if AMD didn't have the finantial constraints it has and actually got some of these revolutionary ideas to the consumer on time..
As it was they stumbled over the finish line and Intel just copied their plans an threw zillions at it to catch up.
Off topic (perhaps someone in the know could PM me?) why didnt AMD patent their ideas over the years instead of letting Intel blatenly copy them?
This to me seems to be the area AMD was designing for all along, cpu/gpu modules together on one die, brilliant idea, now picture if AMD didn't have the finantial constraints it has and actually got some of these revolutionary ideas to the consumer on time..
As it was they stumbled over the finish line and Intel just copied their plans an threw zillions at it to catch up.
Off topic (perhaps someone in the know could PM me?) why didnt AMD patent their ideas over the years instead of letting Intel blatenly copy them?
He said 50% more compute capability at almost half the power. Think twice about it. AMD announced on CES 50% more compute power for the 35W Trinity. 17W and +50% power doesn't match AMDs claim. You can be sure it is a marketing trick.
Something more serious:
http://www.youtube.com/watch?v=agJxehoSBmY
He clearly stated the demo system used a Mainstream APU. Ultra Thin APU was separately mentioned.
The concept of integrating a CPU and GPU is not unique. Only the details of the integration. See the following link to see why Intel is not copying anything though AMD might have prompted Intel to rekindle the concept.
http://en.wikipedia.org/wiki/Intel_Timna
Yes, we have seen this video. He never says anything about 17 watts, he only says it's coming from a laptop. A separate video was presented, at the same booth, where the AMD person specifically mentions the laptop is using the mainstream Trinity part, which is not 17W.
There have been a lot of people claiming 17W because of the video you posted, but that video never mentions 17W. I assume that various sites have latched onto the 17W "slide" that was presented, and made the conclusion that they were the same parts. Unfortunately, while a lot of sites claimed 17W, that doesn't make it correct.
Again: AMD"s own spokesperson called it a mainstream part, and mainstream parts aren't 17W -- those are ULV parts. Case closed.
ronvalencia
Newcomer
In time stamp: 1:06, He talks about "thin" notebooks and He opens the desktop PC case and shows the notebook i.e. "what we are showing you" section. He then claims claims it fits into "ultra thin designs".
ronvalencia
Newcomer