AMD FX-8770 and 9000 - 220W TDP, 5 GHz

[SlowSpyder]

Is that the same as running an Intel chip at like 100C?

Isn't max for the 1090T 62C?

I couldn't tell you for sure, but I have seen the 62C number mentioned before.

I bought this chip on October 13th 2011. My AM2+ motherboard was purchased on March 25th 2008. This board has only ever seen overclocked and power-hungry Phenoms. A PhI 9850 (oc + voltage), a PhII 940BE (oc + voltage), and now my 1090T (oc + voltage). I've also been using the same CPU cooler for about four years now, it is hardly bleeding edge air cooling technology. I'm sure I could get a better cooler and drop temps. I also am using Arctic Silver 5... the same stuff since I got my 1090T.

But if my CPU is pulling more than 200 watts (I think 200 may be a bit low, in fact! ) on my setup, then I have zero doubt that this chip can be cooled without issue. And a quality motherboard can give you years of trouble free operation, even when you beat the hell out of it off and on for over five years.


*edit - One more of my CPU heatsink. No reason, its just pretty.

 

[aigomorla]

We all know how accurate Intel official temps and TDPs can be

Conspiracy theory at its best.

I assume you would know that TJmax is rather irrelevant, because the main factors are the safety margins and the set of real applications considered. There is no a universal formula such as F=ma for computing TDPs.

Instead posting this kind of stuff, it would be much more interesting if you did not avoid by second time the question I did to you regarding the 100W delta claim.

The claim is now debunked in the same review site that you liked to cite to us in FX vs i7 comparisons. Weird that you have no comment, but don't worry I will not insist more.

Intel is very accurate in the reports of there chip.
If you have a 65W TDP chip, then its a true 65W TDP chip.

Also Tjunction MAX values are very important in Intel Chips.
They also list Tjunction MAX values in CPUID on each chip.
This is kinda important because Tjunction MAX determines if the chip should throttle to prevent the chip from causing serious damage to the board and the chip itself.

Finally i only know of 2 people who can argue with IDC on a fair ground.
Cho, and Ruby... his experience is what gave him that elite title.

 

[Cookie Monster]

How does a 5GHz piledriver chip stack up against its intel counterparts?

If memory serves me correctly, achieving that kind of an overclock on the 8350 is pretty difficult isn't it?

 

[Abwx]

We all know how accurate Intel official temps and TDPs can be

They can be but you ll have to dig in PDFs ,

Actualy , Haswell higher grades CPUs are specified at 84W
or so long term TDP but peak TDP can be 153W for 10ms.....

 

[toyota]

How does a 5GHz piledriver chip stack up against its intel counterparts?

If memory serves me correctly, achieving that kind of an overclock on the 8350 is pretty difficult isn't it?

an ivy at 4.2 and a sandy at 4.3 would still be faster for things like gaming while using half the power.

 

[guskline]

How does a 5GHz piledriver chip stack up against its intel counterparts?

If memory serves me correctly, achieving that kind of an overclock on the 8350 is pretty difficult isn't it?

You can achieve the 5Ghz OC to boot into windows but usually at the expense of a very high vcore and an unstable chip with large power draw.

 

[Bubbleawsome]

Why don't they make 300w cpu's? Like maybe a i7 4990x 6 or 8 core @ 5 ghz? I mean, people are fine with 300w+ GPU's, why not cpus? Just only offer it with watercooling.

 

[inf64]

How does a 5GHz piledriver chip stack up against its intel counterparts?

If memory serves me correctly, achieving that kind of an overclock on the 8350 is pretty difficult isn't it?

It's fairly hard to achieve stable 5Ghz OC on 8350. Usually people settle with 4.7-4.8Ghz as maximum for every day.
As for how much performance one gains and how does it stack up vs i5/i7, take any review, multiply stock 8350's "points" by 1.2 in benchmarks that are not measuring time and multiply 8350's time to completion by 0.83 in those that do measure time.
4.8Ghz/5Ghz(turbo) Vishera is straight 20% clock increase over 4/4.2Ghz(turbo) 8350. Thus 20% more performance in most benchmarks (most games won't see that much as they are usually GPU bound).

 

[galego]

Intel is very accurate in the reports of there chip.
If you have a 65W TDP chip, then its a true 65W TDP chip.

I was not thinking in that chip precisely.

 

[Idontcare]

Why don't they make 300w cpu's? Like maybe a i7 4990x 6 or 8 core @ 5 ghz? I mean, people are fine with 300w+ GPU's, why not cpus? Just only offer it with watercooling.

It is not chic, so there is no market for them.

My OC'ed QX6700 @ 4GHz (vapophase cooling) consumed around 270W. The chip cost me $1.5k and the cooler cost me $1k.

Not a lot of volume at that pricepoint. Why would Intel or AMD go after it? The vaporphase cooler guys all went out of business because the demand for that kind of cooling was so low.

That said, the GPU products show us you don't need vaporphase cooling to keep a 300W product functioning. The CPU platform though is the problem, we need a reconfiguration of today's mobo and case layout so the heat from the CPU HSF is directly ejected outside the case.

Only then would it make sense.

 

[Ajay]

Eh, if this leak is correct: http://wccftech.com/gigabyte-confirms-upcoming-5ghz-amd-fx-processors-computex-2013/

Then it's still just plain ole Vishera FX. Too bad, I thought AMD did something really cool there for a minute.

 

[galego]

Eh, if this leak is correct: http://wccftech.com/gigabyte-confirms-upcoming-5ghz-amd-fx-processors-computex-2013/

Then it's still just plain ole Vishera FX. Too bad, I thought AMD did something really cool there for a minute.

That link was already given in this thread

http://forums.anandtech.com/showpost.php?p=35098795&postcount=83

 

[lagokc]

That said, the GPU products show us you don't need vaporphase cooling to keep a 300W product functioning. The CPU platform though is the problem, we need a reconfiguration of today's mobo and case layout so the heat from the CPU HSF is directly ejected outside the case.

Only then would it make sense.

Well that and the GPUs are using a lot more transistors at a less advanced manufacturing process which means that heat is dissipated over a much larger area. Square mm for square mm a 300w GK110 is the equivalent of something like a 900w Haswell. Now if Intel wanted to manufacture a 7-billion transistor CPU at 28nm things would be closer to apples to apples...

 

[NTMBK]

That said, the GPU products show us you don't need vaporphase cooling to keep a 300W product functioning. The CPU platform though is the problem, we need a reconfiguration of today's mobo and case layout so the heat from the CPU HSF is directly ejected outside the case.

Only then would it make sense.

Don't give them ideas, they'll bring BTX back. D:

 

[R0H1T]

Well that and the GPUs are using a lot more transistors at a less advanced manufacturing process which means that heat is dissipated over a much larger area. Square mm for square mm a 300w GK110 is the equivalent of something like a 900w Haswell.

That doesn't make any sense, with the larger die a GPU is better placed at heat dissipation & the one thing you can extrapolate is that overall its beneficial for cooling, the placement of transistors & other logic circuits is also important but secondary in this case all else being equal.

Now if Intel wanted to manufacture a 7-billion transistor CPU at 28nm things would be closer to apples to apples...

I bet something like that would cost you ten grands at the very least

 

[Bubbleawsome]

I bet something like that would cost you ten grands at the very least

I would start a "buy me a epic $10,000 CPU" fundraiser. I bet they would bet people buying it. Like 1 or 2 people a year.

 

[lagokc]

That doesn't make any sense, with the larger die a GPU is better placed at heat dissipation & the one thing you can extrapolate is that overall its beneficial for cooling, the placement of transistors & other logic circuits is also important but secondary in this case all else being equal. I bet something like that would cost you ten grands at the very least

Reread what I wrote. GPUs can dissipate 300w in part because of their large size. It doesn't matter how many fans you have, dissipating 300w from a CPU that only has 1/3 the area is going to be very challenging.

 

[Abwx]

Reread what I wrote. GPUs can dissipate 300w in part because of their large size. It doesn't matter how many fans you have, dissipating 300w from a CPU that only has 1/3 the area is going to be very challenging.

FX has a large area so power density/mm2 is low
with its current TDPs.

7970 GPU is 365mm2 and 250W TDP = 0.685 W/mm2 ,
this while using an air cooling , extrapolating this number
for the FX 315mm2 yield 215W.

 

[Idontcare]

Reread what I wrote. GPUs can dissipate 300w in part because of their large size. It doesn't matter how many fans you have, dissipating 300w from a CPU that only has 1/3 the area is going to be very challenging.

No one is disagreeing with the spirit of your position, because of course it is correct. Thermal density matters.

Where people are going to take you to task is that you are implying that it is some binary type of immutable boundary, that creating higher TJmax parts or better thermal conductivity coolers (water cooling) isn't going to matter (the impression that I perceive you to be creating amongst those responding to your posts).

I think you'll agree that (1) it is a challenge, nothing is a given, but (2) it is simply a matter of money and the expense that AMD or Intel is willing to go to in packaging a stock cooler with their "stock" 220W TDP CPU.

IMO the engineering solutions already exist. The only question is will enough people be willing to pay for them at retail MSRP?

 

[AtenRa]

AMD can sell the CPUs alone without any Heat-Sink Fan, just like Intel sells its Socket 2011 6-cores. People buying those things usually install aftermarket oversized Coolers or Water.

 

[boxleitnerb]

PCGH posted it a bit early, about an hour ago but took it offline again. It's true, just the speculated names are not.

FX-9590: 4.7/5.0 GHz, 220 W TDP, $800
FX-9370: 4.4/4.7 GHz, 220 W TDP

Both CPUs are OEM only. Silly, just silly, especially the price.
I'll post the link here when the article goes back up again.

 

[guskline]

PCGH posted it a bit early, about an hour ago but took it offline again. It's true, just the speculated names are not.

FX-9590: 4.7/5.0 GHz, 220 W TDP, $800
FX-9370: 4.4/4.7 GHz, 220 W TDP

Both CPUs are OEM only. Silly, just silly, especially the price.
I'll post the link here when the article goes back up again.

What motherboards will support these cpus?

 

[boxleitnerb]

The article doesn't say. Just that several motherboard manufacturers are worried they cannot certify the cheaper boards for these CPUs due to the immensely high currents. As this is OEM only, AMD probably has selected a handful of highend boards for validation only.

Online:
http://translate.google.de/translat...X-9590-FX-9370-AMD-Centurion-1073412/&act=url

 

[Idontcare]

I wonder how much a 220W "TDP" certified AM3+ mobo is going to cost.

 

[boxleitnerb]

We'll never know -> OEM, just the price for the complete system