FX cpu's bad reputation

[Bobnova]

For every one guy who posts on these forums with a Sabertooth Rev.2 there seems to be six who have an MSI board just waiting to explode.

To be fair, I've seen an awful lot of problems with Sabertooth boards that haven't existed on other, less expensive, boards.

To also be fair, if you're not overclocking you don't necessarily need a bunch of power phases and/or a MOSFET heatsink.
The normal user doesn't run at 100% CPU much, certainly not on a 8 core thing, and they don't overclock at all either.
Certain Gigabyte implementations of AMD Ultra Boost Turbo or whatever it is aside, stock doesn't generally stress MOSFETs tremendously.
Those Gigabyte boards however, did. FX at 1.45v stock is a problem, hopefully gigabyte fixed it in one of the half dozen revisions to that series.


What I think most people forget most of the time here is that we (overclockers.com forum members) are an extremely small part of the desktop market, most people don't need or want what we need and want.
The other "we" I'm part of, the benching world, is even smaller. Amusingly I think we might be listened to more by manufacturers, but that doesn't help the low end or mid range boards much (LGA1156 being the exception to that statement).

I very much liked the PhII chips. The first gen FX didn't excite me much, too Netbursty for my tastes. I haven't played with the more modern stuff, except for a little bit with a Trinity and building a Trinity out of ex-benching/LN2-cooled parts for my brother. As a note and a nod to AMD, that motherboard/CPU spend some hours on liquid nitrogen at -190°c or lower, at voltages I'd rather not admit to, and still work beautifully 24/7 at stock in a fairly hot environment.

 

[trents]

Looks like Deneb would have just as much trouble on cheap boards, because it looks like the same higher TDP. (With 125 W likely typical.)

Sadly, it looks like the 955 uses the same amount of watts as a hex core version. (Phenom II)

This is technically true but only relevant when the 955 and the FX CPUs are both run at stock settings. The FX CPUs begin to quickly draw much more power and produce much more heat than the Denebs when you begin to overclock.

Are there even 95 W TDP versions of the 955?

I think there were some locked OEM 95W versions of the 955 that were produced for a short time.

 

[Bobnova]

I've seen 95w 955s, non-be C3 chips, normal retail. They had a purple box as I recall.
Weren't around all that long.

 

[Woomack]

TDP doesn't mean that CPU is using that power. It's max power in theory but there is lot of CPUs with the same TDP but different clock, VID or even core count. Like hard to believe that Pentium , Core i3 and Core i5 have the same wattage or that AM1 Sempron 2 core and Athlon 4 core between 1.3 and 2.05GHz have all the same, 25W TDP.
FX max wattage is actually higher than AMD specifies. APU look much better from what I see for both wattage and mobo requirements.

 

[Waza]

My understanding is that the variance in wattage comes from leak current and required VID voltage. TDP is kind of loose term that basically means the worst case scenario. If chip fails to run under the specified TDP they just bin it to lower clock tier or close cores and sell it under different model name.
The voltage that motherboard supplies to CPU is often little bit offset from the actual specified VID value which translates to slightly higher wattage, especially in boards that struggle to deliver steady load voltage due to poor power delivery design.
And I wouldn't be surprised if they have a little bit of +- wiggle room when binning the chips
But that is just my own speculation so don't lynch/quote me.

 

[RGone]

Maybe or sort of...

Thermal Design Power (TDP) @ http://www.cpu-world.com/Glossary/T/Thermal_Design_Power_(TDP).html
The Thermal Design Power (TDP) is the average maximum power a processor can dissipate while running commercially available software. TDP is primarily used as a guideline for manufacturers of thermal solutions (heatsinks/fans, etc) which tells them how much heat their solution should dissipate. TDP is not the maximum power the CPU may generate - there may be periods of time when the CPU dissipates more power than designed, in which case either the CPU temperature will rise closer to the maximum, or special CPU circuitry will activate and add idle cycles or reduce CPU frequency with the intent of reducing the amount of generated power.

TDP is usually 20% - 30% lower than the CPU maximum power dissipation.

Average Maximum ?? I thought Max was Max. How can one average a max? Every cpu would have to be tested in order to get average of a Max. Probably some sort of word playing.

So TDP is n0t a theoretical max power. Nor by those definitions is TDP the worst case scenario.

Both AMD and Intel have sort of played a little loose and free with how they define TDP. Get the white papers out and you can read for yourselves.

Now for power consumption pointers; AMD is trying to use ACP (Average CPU Power) which none of the big guys use. Instead, the big industry players like Intel, Sun, HP and IBM have settled on the suite of SPECpower benchmarks run by a committee of industry players hailing from all those firms and even AMD. This, for the most part, promotes TDP (Thermal Design Power), a measurement Intel favours and which AMD considers inherently biased.

If you have not spent nearly 2.5 years with the 8 core FX processors thru Bulldozer and Piledriver and run them on something less than DICE, LN2 or LHE, then it is hard to really have experienced the power the 8 core processors (actually all FX cpus) can consume and throw off the resultant heat at higher constant frequencies. 4.8Ghz and beyond is a load on the cooling system and motherboard VRMs.

In truth, I doubt many people benching for points that are not in the top 10 around the world, give a hoot what AMD says about power consumed or heat generated anymore. At least not in the discrete cpu market place. Gamers are interested in IPC and that falls to Intel to lead currently. AMD has stated time and then time again that they have no interest in competing in the upper tier of discrete cpu performance. They completely stamped this as TRUE when they did not bring a "steamroller" variation to us discrete cpu users.

AMD gave us "consumers" nearly 11.5 months of HYPE before releasing the Bulldozer onto the consumer. When they sent the first Bulldozer Review packs out to the review sites they had probably the best mobo packed with the cpu (CHV with correct bios) and the AMD water cooler and a special review package for monitoring from CPUID.

By this manner of bet hedging, the average consumer had no idea what they would face when they turned ALL the cores on and began to overclock as most of us do with BIG WATER on a "hefty" VRM circuited mother board. The review sites wanting to continue to receive test parts and pieces; fell right into step and only said the FX cpus did not bench equal to Intel at relative speeds. Again a gap in real information to the amd buyers.

Not the first bit of this has any real meaning to 98% of the computer buying market. There are only a small segment of users that hinge their next drawn breath on the performance of their computer in benches and how far can they overclock their cpu. Just a small segment of the computer market place.

Now AMD execs are leaking more hypeage about something for x86 again in 2016. Hold your breath.

Just an FYI. I have a good FX-8350 used for home video editting. It literally lifts the front end and totes the mail down the track for me in that usage. I like it. But I do not appreciate the hypeage that makes most of us that do the helping in the AMD cpu forum section 95% of the time; feel like mobo police from users coming in wanting to overclock with entry level mobos that are not up to the task of ALL cpu cores turned on while overclocking, which is necessary if you want an FX to finish just behind an Intel cpu clocked at a lesser speed.

Long live those whom can put up with the stress of "mobo policing".

RGone...ster.

 

[ShrimpBrime]

Reading the various forums it is pretty obvious lots of people buy boards that are only marginally "FX cpu compatible". Lots of these boards are Micro ATX, some are full ATX, 4+1 phase power, no mosfet cooling, small chipset heatsinks. The advertising makes them sound great though. The common denominator is that these boards are cheap. Human nature being what it is people buy them. The user has high expectations. Then they find they can't overclock. They may not even be able to run at stock speeds without throttling. The easiest thing to blame is the AMD cpu. Then the forums are flooded with negative posts from these end users. They tell people they know "get Intel, AMD is terrible. This perception has to be damaging to AMD.

Many Moons ago, AMD stated no competition for Intel Enthusiast market.

Nothing wrong with 760G. Just be sure and buy the right motherboard. Not recommended for daily with 8 cores. Very recommended for LN2 as it's a cheapy board and killing 45$ worth with LN2 doesn't hurt the bank.

AMD is great and always has been. First "true quad" produced. Yea it was slower than Core 2 quads, but it was innovation that drove AMD further into the future.

AMD is seeking revenue, and it's not in the enthusiast market, and never was......

my 2 pennies.

 

[chrisjames61]

Seeking revenue would be in big OEM cpu contracts I presume. I bet the margins are pretty low but on a large scale. Enthusiasts are pretty much what make up the discrete video card market buyers. I wonder what the breakdown on AMD branded cards vs licensee sales revenue is?

 

[Mr.Scott]

Server market is AMD's big market.
APU's will be the big ticket in the OEM's very soon.

 

[chrisjames61]

Server market is AMD's big market.

I thought that they had basically no real presence in the enterprise server market at this point?

 

[Mr.Scott]

As I recall, around 15%. ARM should increase that a little. APU's will be the big thing next, along with the game console chips and tablet stuff.

 

[Woomack]

Most manufacturers are not even offering servers on the mass market with AMD cpus. I can't find any AMD based servers when I look at the distribution offer in Poland. I'm not saying that AMD is not existing on the server market but % is really low.

AMD is moving to SoC and mobile chips. Nothing else is new in their offer. Even console chips are generally older APUs with some changes that's why they're so slow.

I'm selling business hardware and believe me I haven't sold even 1 AMD based PC, server or laptop in last 5 years. Simply noone wants it. When you compare Intel and AMD prices then even in cheap series Intel is winning when you look at performance/price.

APU seem good but those who play games don't need IGP and those who won't play games don't need anything better than standard GPU. Also APU's graphics looks good only in highest series. Every step lower and you get worse GPU so you won't play anyway.

We hear about server ARMs from AMD for about 2 years. So far we haven't seen anything on the market. On the other hand it's not a product which will be popular so won't bring AMD big profit.

I wish to see something like AM1 series but with 2 channel+ memory controller and 4GHz clock. Looking at these low clock chips performance it could be something good.