Sign in with
Sign up | Sign in
Your question

Want to OC your FX 8150? I think you're going to need a bigger PSU

Last response: in CPUs
Share
a b à CPUs
October 17, 2011 6:47:35 PM

an FYI, interesting read for those who have not read about the power demands

http://www.kitguru.net/components/cpu/zardon/power-cons...

More about : 8150 bigger psu

October 17, 2011 8:39:19 PM

That...is....just....ridiculous!
October 18, 2011 3:15:29 AM

*sigh* power consumption is actually where the 8150 looks the absolute worst. In a lot of gaming benchmarks, an 8150 loses to a stock 3.3GHz 2500k, even with the 8150 overclocked to 4.6GHz. In power consumption terms, that puts an i5 2500k at 128 watts still outperforming an 8150 at 406 watts.

It's pathetic.
Related resources
a b à CPUs
October 18, 2011 5:31:10 AM

fulle said:
*sigh* power consumption is actually where

the 8150 looks the absolute worst. In a lot of gaming benchmarks, an 8150 loses to a stock 3.3GHz 2500k, even with the 8150 overclocked to 4.6GHz. In power consumption terms, that puts an i5 2500k at 128 watts still outperforming an 8150 at 406 watts.

It's pathetic.


I think you need to look at the chart again. 2500k is 206 watts not 128.

And a lot of bulldozers power usage is likely the GF yield issues. Poor yields usually mean more power hungry chips. I expect a significant drop with die revisions.
a c 126 à CPUs
October 18, 2011 6:01:47 AM

FALC0N said:
I think you need to look at the chart again. 2500k is 206 watts not 128.

And a lot of bulldozers power usage is likely the GF yield issues. Poor yields usually mean more power hungry chips. I expect a significant drop with die revisions.


He said a stock 2500K. And It might not only be yeild issues. I remember that NetBurst (aka HeatBurst) was a very power hungry uarch in the 130nm and 90nm flavors. It wasn't until the 65nm process did we see decent speeds at decent TDPs. So there is a possibility it could be the uarch leaking and not the process.

The whole point to HK/MG with SOI was to lower power consumption even more. Instead its gone way way up.
a c 82 à CPUs
October 18, 2011 6:05:26 AM

FALC0N said:
I think you need to look at the chart again. 2500k is 206 watts not 128.

And a lot of bulldozers power usage is likely the GF yield issues. Poor yields usually mean more power hungry chips. I expect a significant drop with die revisions.

a die revision will only be a very minor help, not a significant one as you believe. they need to redesign major parts of the core to get a SIGNIFICANT power reduction. I honestly see this as the end for AMD in the CPU business. Intel are getting too far ahead.
a b à CPUs
October 18, 2011 7:15:56 AM

jimmysmitty said:
He said a stock 2500K.


I misread his post entirely. I was more tired than I thought.

His analysis was still garbage though, but for different reasons. The 8150 is a lot faster overall OCed than the 2500K is stock. Just not in gaming. You don't get to pick and choose your bench when you give a categorical assessment, which he did. Its fallacious.

Quote:
And It might not only be yeild issues. I remember that NetBurst (aka HeatBurst) was a very power hungry uarch in the 130nm and 90nm flavors. It wasn't until the 65nm process did we see decent speeds at decent TDPs. So there is a possibility it could be the uarch leaking and not the process.

The whole point to HK/MG with SOI was to lower power consumption even more. Instead its gone way way up.


I don't think its all yield. But I am sure a good chunk of it is. The yields were reportedly awful.

a b à CPUs
October 18, 2011 7:22:04 AM

iam2thecrowe said:
a die revision will only be a very minor help, not a significant one as you believe. they need to redesign major parts of the core to get a SIGNIFICANT power reduction. I honestly see this as the end for AMD in the CPU business. Intel are getting too far ahead.


Intel is no further ahead than they were 4 years ago. Its been the same story for a while. In October of 2007, AMD fastest part was selling for about $170. After the Barcellona launch, the 9600 B2 was selling for $250 to $300 and it was awful.

Sound familiar? Exactly.

There are a lot more similarities, but I am just too tired to go into them now. Bed time. For me.
October 18, 2011 8:09:54 AM

FALC0N said:
I think you need to look at the chart again. 2500k is 206 watts not 128.

And a lot of bulldozers power usage is likely the GF yield issues. Poor yields usually mean more power hungry chips. I expect a significant drop with die revisions.



Actually, no. Yield has nothing to do with power consumption. The processor design and the process node do.
a c 82 à CPUs
October 18, 2011 10:01:24 AM

FALC0N said:
Intel is no further ahead than they were 4 years ago. Its been the same story for a while. In October of 2007, AMD fastest part was selling for about $170. After the Barcellona launch, the 9600 B2 was selling for $250 to $300 and it was awful.

Sound familiar? Exactly.

There are a lot more similarities, but I am just too tired to go into them now. Bed time. For me.

Another similarity is most people back then bought a core 2 CPU and never looked back, same goes now, most people will get a core i3, i5, i7, and they wont sell many of these chips which they arent making enough money off to be profitable. Enough time of unprofitable business means an end to the business. Just lucky they still make good GPU's, although that could change now they have taken a risk with RAMBUS.
October 18, 2011 10:27:01 AM

OMFG!!!!!!!!!!!!!! Let the insanity begin.
a b à CPUs
October 18, 2011 6:13:55 PM

iam2thecrowe said:
a die revision will only be a very minor help, not a significant one as you believe. they need to redesign major parts of the core to get a SIGNIFICANT power reduction. I honestly see this as the end for AMD in the CPU business. Intel are getting too far ahead.


I agree somewhat but I don't believe AMD will leave the CPU business but rather the performance desktop business, if they don't get their act together very fast. Their early fusion processors fits the mobile market and basic desktop ($299-599) just fine but most people don't spend $1000+ on their PC (hence why laptops now equal PC sales). Gamers do, people who work with video and graphics (most GA's use macs) but we often over estimate those numbers when compared to the general public. My girlfriend does not need an i5, nor do my parents, her parents, and most of my friends. Frankly an athlon II x3 would be overkill for what they do, surf the web, pay bills, watch youtube, post pics on facebook, look at dwarf porn, all normal everyday stuff. So AMD developed a CPU to kill two birds with one stone, the server market & the performance market while keeping development cost down as much as possible. What they delivered and we got was a CPU that fails to be specific for the performance segment (shocking). Their slim market share in that segment will only get worse as AMD users switch to Intel and will not come back for slower processors just because the name on the box says AMD.



a b à CPUs
October 18, 2011 6:22:45 PM

yomamafor1 said:
Actually, no. Yield has nothing to do with power consumption. The processor design and the process node do.


Yes, yields do have an impact on power.

Bad yields are not just non-functional parts but parts that use too much power. You can have parts of the same revision produce vastly different power profiles. And thats doubly true with an immature process node. And the extremely poor yields reported suggest the node is very immature.

Like I said, it isn't the whole issue, but it a part of the problem.
a c 83 à CPUs
October 18, 2011 6:29:47 PM

yomamafor1 said:
Actually, no. Yield has nothing to do with power consumption. The processor design and the process node do.


Better yields would give more chips capable of either operating at higher frequencies with the same voltage, or at the same frequency with lower voltages. Either of the 2 would improve performance/watt.
a b à CPUs
October 18, 2011 6:42:09 PM

loneninja said:
Better yields would give more chips capable of either operating at higher frequencies with the same voltage, or at the same frequency with lower voltages. Either of the 2 would improve performance/watt.


true but it seems like you would be re-arranging deck chairs on the titanic
October 18, 2011 6:50:39 PM

I just wanted to point out that as far as I could make out of it, those are full load tests for all cores. No game is ever going to load all cores like that, so comparing this particular power test with gaming benchmarks is a bad comparison.

I would like to know how BD performs under partial/moderate loads. I doubt it paints such an ugly picture, although I am sure it is far from pretty. I thought Tom's review pointed out the improvement of their Cool & Quiet?
a b à CPUs
October 18, 2011 11:41:53 PM

dirtyferret said:
true but it seems like you would be re-arranging deck chairs on the titanic


An incredibly detailed an thought provoking analysis. Thank you for sharing.
October 18, 2011 11:51:44 PM

FALC0N said:
Yes, yields do have an impact on power.

Bad yields are not just non-functional parts but parts that use too much power. You can have parts of the same revision produce vastly different power profiles. And thats doubly true with an immature process node. And the extremely poor yields reported suggest the node is very immature.

Like I said, it isn't the whole issue, but it a part of the problem.


Eh no. Non-functioning, or parts that don't pass AMD's requirement would be disabled, so those parts no longer use power.

And what you're saying seems to be that different CPU uses different amount of power. This is true, but it is also true that Bulldozer use more power than its predecessor, as well as the Intel counterparts, universally in every reviews.

That, is no yield issue.


loneninja said:
Better yields would give more chips capable of either operating at higher frequencies with the same voltage, or at the same frequency with lower voltages. Either of the 2 would improve performance/watt.


Please look up.
a b à CPUs
October 19, 2011 5:53:06 AM

yomamafor1 said:
Eh no. Non-functioning, or parts that don't pass AMD's requirement would be disabled, so those parts no longer use power.

And what you're saying seems to be that different CPU uses different amount of power. This is true, but it is also true that Bulldozer use more power than its predecessor, as well as the Intel counterparts, universally in every reviews.

That, is no yield issue.

Please look up.

Poor yields suggest issues with the process. Issues with the process can cause higher power usage.

And if chips that don't meet AMD standard are being disabled, why are the power useage numbers so high?

The X6 1100 and the x4 955 through 980 are 125 watt parts, but the 8150 has been using consistantly 30 watts more than other AMD 125 watt parts in almost every test. So either the 1100 and the 980 are really 95 watt parts OR the 8150 is out of spec.

October 19, 2011 7:15:54 AM

FALC0N said:
Poor yields suggest issues with the process. Issues with the process can cause higher power usage.

And if chips that don't meet AMD standard are being disabled, why are the power useage numbers so high?


Then again, its the process node that's the issue, not the yield. The inconsistent yield may give varying power consumption, but it will not cause higher power consumption across the board.



The X6 1100 and the x4 955 through 980 are 125 watt parts, but the 8150 has been using consistantly 30 watts more than other AMD 125 watt parts in almost every test. So either the 1100 and the 980 are really 95 watt parts OR the 8150 is out of spec. said:
The X6 1100 and the x4 955 through 980 are 125 watt parts, but the 8150 has been using consistantly 30 watts more than other AMD 125 watt parts in almost every test. So either the 1100 and the 980 are really 95 watt parts OR the 8150 is out of spec.


Linkage please?
a c 202 à CPUs
October 19, 2011 7:50:30 AM

dirtyferret said:
I agree somewhat but I don't believe AMD will leave the CPU business but rather the performance desktop business, if they don't get their act together very fast. Their early fusion processors fits the mobile market and basic desktop ($299-599) just fine but most people don't spend $1000+ on their PC (hence why laptops now equal PC sales). Gamers do, people who work with video and graphics (most GA's use macs) but we often over estimate those numbers when compared to the general public. My girlfriend does not need an i5, nor do my parents, her parents, and most of my friends. Frankly an athlon II x3 would be overkill for what they do, surf the web, pay bills, watch youtube, post pics on facebook, look at dwarf porn, all normal everyday stuff. So AMD developed a CPU to kill two birds with one stone, the server market & the performance market while keeping development cost down as much as possible. What they delivered and we got was a CPU that fails to be specific for the performance segment (shocking). Their slim market share in that segment will only get worse as AMD users switch to Intel and will not come back for slower processors just because the name on the box says AMD.


Is watching dwarf porn everyday stuff for you? lol

I agree that AMD will fill the budget gaming PC and home use PC IF and thats a big if, the consumer realizes that AMD is A viable cheaper option. Most people would be happy with an APU but do not know enough nor are confident enough in AMD to actually get it.

So yes. I think AMD has alot of market potential if they did a massive marketing campaign of their APU's.

They also have not pushed the use of the APU's in the mobile space enough. I barely see any laptops with APU's. Considering that an APU equiped laptop can game well enough for most(whereas a normal I3 cannot compete) this is a very big shame!
a b à CPUs
October 19, 2011 9:05:54 PM

it isn't everyday stuff for everyone?

I have seen plenty of $299 and less laptops featuring the AMD fusion APUs so there does seem to be a market for the chips
a b à CPUs
October 20, 2011 4:10:34 AM

yomamafor1 said:
Then again, its the process node that's the issue, not the yield. The inconsistent yield may give varying power consumption, but it will not cause higher power consumption across the board.


Fair enough. Yields was a poor choice of words on my part.

Quote:
Linkage please?

Lets look at anandtech:

http://www.anandtech.com/show/4955/the-bulldozer-review...

The 8150 is using 30 watts more than the X6 1100 and 50 watts more than the X4 975. Both 125 chips. Something is off there. I think AMD is letting chips through that exceed the 125 watts. That or the older chips were better than rated, which is also possible.
!