Richland & Kabini rumours

[Phynaz]

I don't think it's really important.

TDP is the parameter used to build thermal solutions for a processor, it's an average of how much power it can dissipate for a significant period of time.

Two problems.

1. It has been shown by multiple users that an 8350 can exceed it's stated TDP running regular applications. Therefore using TDP as stated to design a cooling solution for an 8350 is insufficient.

2. As noted by MSI in one of the linked threads, motherboard manufacturers use TDP as a design spec for power delivery. Right or wrong, they are doing it, and at least MSI is blaming AMD for under specing their chips.

Edit: I miss read your post...I should really be replying to inf64.

 

[AtenRa]

Two problems.

1. It has been shown by multiple users that an 8350 can exceed it's stated TDP running regular applications. Therefore using TDP as stated to design a cooling solution for an 8350 is insufficient.

links ??? because i havent seen that happening with my setup. Even at 4.4GHz(turbo off) running x264 the system used 240W.

http://forums.anandtech.com/showpost.php?p=34481789&postcount=441

 

[Abwx]

See post 11 of the first link. You can agree or disagree with IDC, but he did say that the cpu was using 190 plus watts.

I dont disagree with him but with you since he doesnt say nothing
of the sort and you are the one in disagreement with him since
you are distorting what he said.

What he said is that he measured 194W more consumption
from iddle state to full load state FOR THE WHOLE SYSTEM but
he didnt measure what was consumed by the CPU and the rest
of the system SEPARATLY , period.

 

[mrmt]

Ok, I'm gonna bait - to my understanding default conditions would exclude power virii by - well - default.

Prime 95 isn't a virus.

Also, the P4 Prescott already did that and hardly anyone cried foul and while SB and IB won't exceed their power limit, they will throttle quite badly in some Ultrabooks.

Could you provide link for that?

And don't forget the new 10W Chips which turned out to be 17W TDP.

Sorry?

There is nothing wrong in consuming more than stated TDP, not even way more your TDP, because as long as your average thermal dissipation for the given period doesn't exceed TDP you are within safe limits. Those 10W chips are ok as long as you can design thermal solutions and components around those 10W values, and so far nobody has discredited that.

This is the crux of the issue with that MSI board. The processor is generating too much heat during an extended period, beyond levels in the thermal specifications which the board was built around. Simply put, the average thermal dissipation of the 8350 is higher than every single other 125W AMD part out there.

If the FX 8350 is built around 140W specifications then everything is fine, it's just misleading market. But the FX 8350 may not be built for 140W, it may be a 125W part, in this case it's not only misleading market, as this factory overclock will be translate in higher failure rate than other AMD processors.

I just want to bring the issue here. Are the other AMD parts suffering of the same problem or is this a one-of-a-kind issue?

 

[Atreidin]

I believe you are confusing heat and power draw. The CPU is pulling power from the board and the board is overheating. The processor is pulling more than 125W, which is causing the inadequately protected VRM to overheat.

 

[krumme]

Certainly dramatically better APU performance are always welcome, especially if they have identical TDP as the previous Bobcat APU models.

When it comes to Bobcat APU E-350/E-450/E2-1800/E1-1200/C-60, practice has shown that these models do not have enough CPU performance in many situations therefore Kabini / Temash APU models are very welcome.

Broadwell models can be compared with Kabini / Temash APU successors in the year 2014, and maybe with future ULV Kaveri APU Mobile models.Kabini / Temash APU models from 2013,you cant compare with Intel Broadwell that's what is most important.

I dont know what practice you are referring too. Bobcat is still selling extremely well, so a lot of people think different from you. That what matter in my world.

Yes - Broadwell will be another price league and far out, but probably still beeing able to take some market share from the most expensive kabini/temash. The positioning of the new Atoms is the interesting from AMD perspective.

But the performance of the new kabinis seems to be not only impressive but the scalability is what makes the huge difference. 4-10w bobcat is not usefull today. Temash will be a gamechanger here.

 

[krumme]

If i was AMD i would have made a strategic alliance with MS years back. There is only one thing that can take the stupid OS win 8 off the ground and thats tablets with temash.

It took a team of 80 in India to assemble the IP for e350. If 100 could do it for jaguar APU, it would be the cheapest way to get this turd of the ground. In hinsight MS would have paid for that 3 times.

 

[Asterox]

I dont know what practice you are referring too. Bobcat is still selling extremely well, so a lot of people think different from you. That what matter in my world.

Yes - Broadwell will be another price league and far out, but probably still beeing able to take some market share from the most expensive kabini/temash. The positioning of the new Atoms is the interesting from AMD perspective.

Bobcat APU is AMD's best-selling product, it is a very good little APU with modern graphics core but the CPU part is not strong enough and that's a fact.His biggest problem was a old 40nm manufacturing process,if had it been produced at 32nm it would have looked much better in terms of much beter CPU / GPU performance.

Now that AMD has moved this little APU products on 28nm, and additionally very much improved architecture that is now a great product if you look at the performance on 15W TDP in particular Quad Core APU Jaguar model. Actually Jaguar APU/SoC models are this year's best AMD products, there is no doubt until Kaveri APU arrives on the market.

Bobcat APU - Small and good

Jaguar APU/SoC - Small but mighty

 

[frozentundra123456]

I dont disagree with him but with you since he doesnt say nothing
of the sort and you are the one in disagreement with him since
you are distorting what he said.

What he said is that he measured 194W more consumption
from iddle state to full load state FOR THE WHOLE SYSTEM but
he didnt measure what was consumed by the CPU and the rest
of the system SEPARATLY , period.

He said the only thing that could be using the extra power is the CPU.

Edit: here is a direct quote copy and pasted from his post "The only difference between the 281W at load and the 87W at idle is the CPU which is loaded at 281W, versus the CPU being idle at 87W.

So to be sure a portion of the 87W at idle is attributable to the CPU, but even if we (for now) make the assumption the attribution is zero, the difference between idle and load is 194W which has no place to come from but the CPU."

I dont even know if he is correct. Point is you have no right to accuse me and other posters of lying or making things up. He did say that.

 

[NTMBK]

I dont know what practice you are referring too. Bobcat is still selling extremely well, so a lot of people think different from you. That what matter in my world..

I can vouch firsthand that a C-60 struggles on certain media consumption tasks- I have a C-60 netbook, and it can't really handle Netflix well. (Mainly because Netflix can't figure out how to get hardware acceleration working, but there you go.) But Temash will hopefully make a real difference.

EDIT: ...not that it looks like anyone will ever make a netbook again.

 

[Abwx]

He said the only thing that could be using the extra power is the CPU.

The CPU is powered through a switch mode power supply (the vrms+inductance+rectification+capacitors) that will hardly have
100% efficency so saying that it s the CPU that extract all the
power delta is just wrong.

Anyway , using the same MB Techreport get 20W lower iddle
compsumption than IDC while using a 7950 GFX , so from the
start the numbers are questionnable since they have no more
than 125 delta using X264 with all cores loaded.

Of course a 125W TDP CPU can consume more on short bursts
but it s just a bad faith to not point that Intel say exactly
the same thing for its CPUs , that is that power can exceed
the TDP on peaks but anyway , what is allowed for Intel
is just dishonnesty when coming from AMD , that tell a lot
about your systematic biaising.

 

[Abwx]

If the FX 8350 is built around 140W specifications then everything is fine, it's just misleading market. But the FX 8350 may not be built for 140W, it may be a 125W part, in this case it's not only misleading market, as this factory overclock will be translate in higher failure rate than other AMD processors.

I just want to bring the issue here. Are the other AMD parts suffering of the same problem or is this a one-of-a-kind issue?

I call this spreading FUD under the cover of exposing a non existent
issue.

Didnt you said that AMD sent drones by there ?...

Given their dire financial situation they are less likely to ressort
to such practices than a said competitor that has a full truck of money
to throw at discutable "marketing" fuds...

 

[ShintaiDK]

http://forums.anandtech.com/showpost.php?p=34589334&postcount=31
http://forums.anandtech.com/showpost.php?p=34591826&postcount=47

But we are all dumb and shills, only sat in this world to bash AMD with our biased attitudes, right?

 

[Atreidin]

The entire industry, except for MSI because they got caught making crappy motherboards, seems to want to define TDP as not a maximum power draw, but as an average amount of thermal power that needs to be dissipated.

I don't like the way any company plays with thermal power numbers but I thought Intel was historically worse about dishonest thermal design specs, publishing "average" numbers way, way before AMD started doing it. Maybe everyone is shocked that AMD is following Intel's lead more now?

 

[mrmt]

The entire industry, except for MSI because they got caught making crappy motherboards, seems to want to define TDP as not a maximum power draw, but as an average amount of thermal power that needs to be dissipated.

The entire industry never defined TDP as the maximum power draw, as you can check in AMD, Intel, Sun and IBM technical documentation.

 

[AtenRa]

This is the crux of the issue with that MSI board. The processor is generating too much heat during an extended period, beyond levels in the thermal specifications which the board was built around. Simply put, the average thermal dissipation of the 8350 is higher than every single other 125W AMD part out there.

Really man, you dont have a clue of what the MSI said. They clearly said that the VRM on the board cannot sustain more than 125W, the VRMs are operating at a higher temperature and the BIOS safety mechanism throttles down the VRMs. It has nothing to do with the FX8350 temperatures and its TDP.

I just want to bring the issue here. Are the other AMD parts suffering of the same problem or is this a one-of-a-kind issue?

The only part that has an issue is the MSI board, not the AMD products. MSI cheap out on the VRM design/implementation on the board, not AMD.

 

[mrmt]

Really man, you dont have a clue of what the MSI said. They clearly said that the VRM on the board cannot sustain more than 125W, the VRMs are operating at a higher temperature and the BIOS safety mechanism throttles down the VRMs. It has nothing to do with the FX8350 temperatures and its TDP.

Really, Sherlock? If the VRMs are providing more power than needed for every other 125W processor out there, how is the 8350 dissipating the same amount of heat of the others 125W processors?

 

[inf64]

What any other 125W out there? Who knows how many users run their X4 and X6 chips in this board model and have no clue their board throttles their CPUs all the time when max. load is applied on the cores. Just by slightest OCing on any 125W X4 Phenom II you will force this crappy board to throttle down the clock.

 

[Abwx]

The CPU supply (read VRMs) will mandate about 150W input power
to deliver 125W to the CPU assuming it has 84% efficency , that is
only for the CPU part supply without counting all the rest of the MB
wich will consume more when there is heavy load , even memory will
consume more.

Those who want to discuss the matter should better study some
basic electronics before throwing all the way their lack of knowledge
on this matter.

 

[Ajay]

power = volts * amps. I read it in an undergrad text book so it must be true :awe:

 

[mrmt]

What any other 125W out there? Who knows how many users run their X4 and X6 chips in this board model and have no clue their board throttles their CPUs all the time when max. load is applied on the cores. Just by slightest OCing on any 125W X4 Phenom II you will force this crappy board to throttle down the clock.

According to MSI, only the 8350 throttles down at stock speed. You can apply full load in every single other AMD processor on that board and it won't throttle down.

With an overclock, yes, you may be correct and it will throttle down, but by overclocking you are exceeding the specification parameters, exactly what AMD did with the 8350.

 

[AtenRa]

Really, Sherlock? If the VRMs are providing more power than needed for every other 125W processor out there, how is the 8350 dissipating the same amount of heat of the others 125W processors?

The problem is that the VRM implementation of that board cannot provide the necessary power to the CPU without overcoming the VRM design limits.

The CPU may need 140W and the VRM implementation is designed to give up to 125W. When the CPU needs 140W the VRM exceeds its limits and overheats and thus the BIOS throttles down the CPU in order to keep the power needed within the 125W limit the VRM of the board is designed to give.

It has nothing to do with the TDP of the CPU, it has to do with MSI and the design of the VRM in that board.

 

[Abwx]

power = volts * amps. I read it in an undergrad text book so it must be true :awe:

That s true only for DC as well as for AC if there is zero degree
phase shift between voltage and current , wich is not exactly
the case for PC PSUs.

 

[krumme]

I can vouch firsthand that a C-60 struggles on certain media consumption tasks- I have a C-60 netbook, and it can't really handle Netflix well. (Mainly because Netflix can't figure out how to get hardware acceleration working, but there you go.) But Temash will hopefully make a real difference.

EDIT: ...not that it looks like anyone will ever make a netbook again.

I think netflix is exactly what is the problem for bobcat apu, and what could be a problem for temash, unless power can be directed to cpu part.

I have an e450 in the house for one of the kids. She have a twice as powerfull notebook also, but prefers the small netbook some HP DM1z that was intented for travelling, and have never complained about the speed. In the start flash was not so good accelerated, but that problem seem to be solved now, so she can watch all the online TV she wants without probs. Its clearly about HW acceleration, and its going the right way.

I dont understand there is not market for 11.6 netbooks. I think the Atom killed this market. We had a 10 inch atom we used for travelling and it was such an extreme pain even the wife complained about speed - and then its really bad - , but also because of small screen and resolution and not fullsize keyboard. The 11.6 HP have fullsize keyboard, ok resolution for the size, and the speed is acceptable for light office use.

I think there is a market for it in a few years. Its very practical. Those heavy tablets where people bring keyboards and protective covers is not going to work.

 

[piesquared]

The problem is that the VRM implementation of that board cannot provide the necessary power to the CPU without overcoming the VRM design limits.

The CPU may need 140W and the VRM implementation is designed to give up to 125W. When the CPU needs 140W the VRM exceeds its limits and overheats and thus the BIOS throttles down the CPU in order to keep the power needed within the 125W limit the VRM of the board is designed to give.

It has nothing to do with the TDP of the CPU, it has to do with MSI and the design of the VRM in that board.

Don't waste your time with him, he's just a banker.