• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD not adhering to 125W spec with FX8350?

Caporegime
Joined
26 Dec 2003
Posts
25,692
I found this interesting although not surprising given my experience with a Piledriver.

http://forums.anandtech.com/showpost.php?p=34588633&postcount=24

MSI said:
According to RD engineers from our HQ,

1. AMD has claimed that the FX-8350 to be 125W. However, during our internal testing, when the usage of CPU reaches 100% through Prime 95 for a while, the power consumption can exceed 125W and reaches 140W. With such condition on the 970A-G46, the high amount of power draw also causes the MOSFET to exceed its spec and will overheat.

Thus, to prevent such any potential hazard to the MOSFET, 970A-G46’s BIOS will trick the processor that its temperature is 225 degrees which will then allow the CPU to throttle. This is aimed to make the system stable in this condition.

2. We had tested using 3DMark and it did not cause throttling. Throttling only happens when the loading on a 125W CPU usage is heavy by the use of heavy burn-in tools such as Prime 95/OCCT, and such testing methods are not standard usage scenario or practical.

Please also keep in mind that this throttling behavior keeps system stable under such heavy-loading condition.

This is more of AMD’s issue on the FX-8350 because the TDP was rated lower than its actual value.

So AMD are shipping FX8350's with an actual TDP of up to (and possibly above) 140W and thus board manufacturers (on the cheaper boards at least) are having to implement throttling mechanisms to prevent crashes even at stock, this is because their VRM's are only rated at the official 125W spec as defined by AMD themselves.

Aside from being false advertising this is quite underhand by AMD.
 
naughty naughty.

why do they put out so much heat compared to the intel equivalent?

my 2500k at 4.7 woth 1.39v does about 135watts under intel burn test very high. and the FX8350 is putting out 5 watts more at stock! madness!

what wattage do they go to when overclocked?!
 
This is not really new, I am sure my 1090T at 4.1+GHz exceeded 125W in fact probably 140W. I expect an overclocked intel hexcore would also come close to that.

I have always advocated the use of 990x and 990fx mobos, in fact get the latest chipset you can afford. But skinning the mosfet down to a limit of the TDP seems a bit cheap IMO. Poor MSI.

naughty naughty.

why do they put out so much heat compared to the intel equivalent?

my 2500k at 4.7 woth 1.39v does about 135watts under intel burn test very high. and the FX8350 is putting out 5 watts more at stock! madness!

what wattage do they go to when overclocked?!

How is that measured on the 2500K?

I have seen 142w or thereabouts on HWmonitor but that is at 4.6GHz prime on eight cores.

Measured at the wall plug I get 250w full system wattage priming at 4.6 on eight cores 1.48V, this could be reduced by the 80% efficiency of the PSU to about 200w internally and includes the drives, ram, motherboard and 2D graphics so is probably about right.
 
Last edited:
Intel had a reputation for that issue with Prescott, Smithfield, Presler etc and nobody cared back then. :rolleyes:

I think its because now that you cant really bash the performance of an AMD cpu anymore with newer games and apps performing much better with proper optimization, heat and TDP are the only thing left.
 
Last edited:
This is not really new, I am sure my 1090T at 4.1+GHz exceeded 125W in fact probably 140W. I expect an overclocked intel hexcore would also come close to that.

What TDP they run at overclocked is not AMD's responsibility, being honest about the TDP at stock speeds is.

What this boils down to is AMD are lying about the FX8350's TDP, presumably because if they were honest and called it a 140W chip then it would no longer be seen as an upgrade path for the majority of older 125W motherboards already out there which can't run it (at least not without throttling being added as a normal every day feature).

It affects many of the new boards too (though not the high end overclocking ones):

1> AMD set the official spec for FX8350 as 125W.
2> Motherboard vendors build boards for 125W CPU's as per the official spec AMD gave them.
3> AMD ship 140W FX8350's labelled as 125W.
4> Motherboard vendors find that their motherboards can't cope, add throttling to protect VRM's from damage & prevent crashing.

Pretty shoddy really.
 
Last edited:
I think its because now that you cant really bash the performance of an AMD cpu anymore with newer games and apps perferming much better with proper optimization, heat and TDP is the only thing left.

Yeah, heat can be an issue, but I fail to see that MSI should enable support for a CPU that does not work at stock on their mobo VRM's. If the computers primary job was math or encoding etc. it could process data similar to prime95.

Need to spend a little more on their components methinks because as an encoding tool the 8350 is very cost effective. Basing your testing on 3dmark alone is a little eccentric. ASUS does not seem to suffer likewise
 
Last edited:
I think its because now that you cant really bash the performance of an AMD cpu anymore with newer games and apps performing much better with proper optimization, heat and TDP are the only thing left.
Of course you can still bash the performance, bet there's games released in the past 2 years that your CPU bottlenecks your GPU set up (Even if you don't play them)

:p

Tongue in cheek comment, but it's not wrong.
 
Last edited:
I am not at all surprised having owned several 8 core fx chips and tested them on a variety of motherboards. Every week there is a few people saying that their extreme 3 coupled with an 8350 is not using the multiplier set in the bios and instead a much lower one. Then you see a string of replies to tell them to update bios, check CPU temperature or change the stock heat sink but every time its actually just the VRMs throttling the clock.

I don't believe that motherboard manufacturers have implemented the throttling to combat AMD's FX chips power drain but rather most of the boards where people are finding their stuff throttled were originally made for am3 CPU's and VRMs on boards which were released pre-FX still could throttle the clock but rarely did since their power drain was more modest. Extreme 3 only has 4+1 power phasing and though outdated, it was a good motherboard. It was however made to take phenom CPU's and was only just about capable of taking a 6 core phenom with a moderate overclock, without a huge CPU air cooler nearby to help the airflow around the VRMs. People buy the extreme 3 now because it is cheap and was previously known as a good board and also Asrocks flagship amd board but find that it cant even take the high power fx chips.

Moral of the story: Dont go budget motherboard if you don't go budget CPU!

I have avoided Asrock motherboards in combination of high power fx chips purely because of their previous issues of sketchy VRM quality with some of their lower end boards. I wouldn't buy the higher end boards from them because much better boards can be had for the same price from Gigabyte or Asus. The Fatal1ty may have 12+2 power phasing but i'm doubtful its VRMs are as resilient as Asus' price equivalent board The Sabertooth. Apparently quite some effort did go into improving the power management on Asrock boards so it'd be nice to see a comparison between a Crosshair formula Z and an Extreme 9.
 
Last edited:
excuse the stupidity, but does this mean I've been wrong in thinking the TDP is the heat chucked out by the chip, not the power put in.

if i'm not wrong [face it I probably am] 140w in -125w out=35w of the energy being used in the actual computation.
 
I misspoke before, the ASUS M5A99X does not give wattage in HWmonitor, but neither does coretemp. Coretemp only gives the TDP as 125.1W.

Must have been my Giga 890 that showed watts in HWmonitor.

I fully agree with Avenged7Fold quoted below.

Moral of the story: Dont go budget motherboard if you don't go budget CPU!
 
MSI have been known to underspec their VRMs especially on their cheaper motherboards which has lead to VRM failures,and ASRock has done the same with some of their cheaper motherboards too,which has lead to throttling.

The decent Asus and Gigabyte AMD 970 motherboards have similar VRM arrangements when compared to the 990X equivlalents,and are usually fine.

The Biostar 970 and 990X motherboards seem OK too,ie,I have not heard of any throttling issues so far.
 
Last edited:
I dont know how correct HWmonitor is when it comes to watt usage, but at stock speeds my fx8350 reported 55watt at peak when running prime85.

Of course you can still bash the performance, bet there's games released in the past 2 years that your CPU bottlenecks your GPU set up (Even if you don't play them)

:p

Tongue in cheek comment, but it's not wrong.

Its funny how people will bash 1 product but forget about the competition whenever they are in the same situation, because you know Intel isnt exactly bottleneck free :)
 
Last edited:
Firstly TDP is rarely from most companies stated as max power draw, and no reviews, which all test for power, highlighted it. AS with ANY product, you can get a faulty chip that has a faulty part that for some reason is running higher voltage.

Multiple mobo's have been found to slightly over or under volt cpu's, this affects power draws, lastly, its MSI, who says MSI haven't made a dodgy board that overvolts all cpu's and is covering their ass by blaming someone else?

Intel is due to release a bunch of new chips with a 7W SDP(s, or something, they "made" a new tdp and called it something else) when infact, they are simply 13W cpu's pretending to be something else. Actually I forget their numbering scheme, it may be 17W tdp chips pretending to be 13W, maybe its both.

TDP is whatever AMD list it as, max power usage with X setup under Y conditions using Z loading program. Maybe it uses more under prime, and recommends not to use it.

Intel list TDP which ISN'T the max power draw.... welcome to life, its all marketing buzzwords to make PR people happy to spout off facts. Nvidia for donkeys years lied about products "a 480gtx uses 250W" while everyone found it used a lot more, etc, etc, etc. AMD are one of the last, afaik, people who list TDP as maximum powerdraw(unless they've changed it with piledriver release which they may have).

But considering MSI haven't publically accused AMD of making their boards dangerous and alerting customers with a public recall I'll go with, either complete excuse for their own dodgy board/bios or infact we have no idea if that e-mail is real, I'll file this under the "who gives a damn" heading.

EDIT:- also worth highlighting a 2500-3700k is a quad core chip, half the size done on either 32 or 22nm, at 32nm is way smaller, at 22nm its even smaller. Intel has eleventy billion to spend on process R&D, chip R&D and keeping ahead of the rest of the industry on processes, AMD doesn't. Also worth noting is the response from the guy who posted the e-mail, his take is that MSI are selling a board that can't support what they say it can, not blaming AMD. Mobo's undespeccing their cheapest boards to save money, and getting caught cutting to close, then coming up with how it isn't their fault, isn't new. Why aren't they telling him to get a replacement from AMD are ask AMD for a refund, but offering to help him out by getting a more expensive(properly specced) mobo from them? Blame AMD, give them a working product, hope no one notices they screwed up.

99% of people who use computers never come close to using max load or pushing the spec which is why the cheapest mobo's tend to get way way to close to the limit or cut underneath it in hope of making a little more profit.
 
Last edited:
Back
Top Bottom