AMD preparing 5GHZ CPU?

[Greg83]

lets see what this "rumor" really is.
we have
an essay starting with. AMD is gonna release a new processor. Including 3 essay points (references to proof)
Name
Speed
Price

Sound and logically. its elementary

Doesn't mean they're gonna release it if all their "market" was not interested in it. Especially to the point of , claim fud and wouldn't buy it if it were true.

But at the same time. Prob learned alot from the work they've done about can they of the same identical product minus the cpu string. 8320 and 8350. release the 8370 at 4.2Ghz 4.3-4.5 turbo for $210

and obvious amd wouldn't cost itself sales announcing a product as "marketing" they'd never release. and if it was just fud it would be an intel fanboy trying to cause amd harm, by stopping someone from buying the 8350 now in anticipation of this. which is the only other option with the near identical name of Intel Centrino? Centurion? the price of $795 , luck nein (german for no) 7 being luck and 5 being $. and then $.0 being the speed :P or completely dutch T=7, to, nein , no, 5 $ money. 5.0 $0 .

Cause maybe amd is gonna release the 8370 and would never announce it before they wanted to cost sales of 8350's and this will do the same. forcing amd to announce what is actually coming.

[The Stilt]

This is getting ridiculous.

Unless there is a major node overhaul (which reduces the power consumption by 30-50%) there is absolutely no way AMD can release a 8-core FX clocked to 5.0GHz. Nothing indicates that there would be a new node version coming and releasing one would not make any sense anyway. There is no point in using the limited resources to anything else but on the Gen. 3.

A nasty thing to say but using the limited resources on Piledriver would be like force feeding hay to a dead horse.

Currently the fastest 8-core FX is clocked to 4.0GHz.
At this frequency the 125W TDP envelope is already fully used in well optimized multithreaded applications (such as rendering or video encoding), there is no headroom really.

A proper binning of the dies can make around 200MHz of difference on current Piledriver dies.
The additional scalability comes purely from the lower leakage properties of the binned parts.

The current Piledriver cores do not scale well beyond 4.6GHz, which seem to be the critical frequency really.
They can be pushed beyond that but the increase in the power consumption & emitted thermal power can no longer be justified by the resulting performance increase. Beyond this point the performance-power curve is anything but linear.

We might see a 8-core FX (Piledriver) clocked to 4.3 - 4.5GHz (base frequency) at some point, but anything higher than that is very unlikely.

Even if AMD could produce a 8-core FX clocked to 5.0GHz, there would not be any motherboards for the chip.
A 5.0GHz 8-core Piledriver FX consumes 180W - 200W of power during Cinebench R11.5 for example (DCR Pmax, from CPU VRM).

In some applications the power consumption can be even 30% higher than that so the motherboard design has to have enough of headroom to accomodate the difference + the design margin. The motherboard would need to be able to deliver 235W - 260W of constant power. After the average conversion (VRM efficiency) loss the total VRM input power would be 294W - 325W (24.5 - 27A from the 12V rail).

The power dissipation of the CPU VRM alone would be 36W - 46W.

There are only couple of motherboards which can temporarily manage such of power requirements.
Even these motherboards cannot support such power requirements reliably for extended periods of time.

Having a 5.0GHz Piledriver based system in daily use is far from impossible.
It just requires conditions of which 99.95% of the users cannot attain.
A high performance custom watercooling system for the CPU and the motherboard (CPU VRM).

AMD and an enthusiast also have a slightly different definition for the term "long term reliability and stability".

[AMDforME]

Clearly with some overclocked FX-8350 CPUs already running at 5.0 GHz. there is no issue with AMD binning and selling 5.0 GHz. Vishera CPUs should they desire. This ain't rocket science. I can run P95 on my FX-8350 @ 4.7 GHZ. for 25 hours without overheating @ 1.462 Vcore with an Xigmatek Aegir HSF. In normal operation this system rarely gets above 50C even under heavy video load.

If AMD were to offer a 5.0 GHz. Vishera - which they have never said they would, then they could if they so desired, include an H2O cooler as they have offered with the FX-8150. Remember people that the reports of AMD selling a 5 GHz. CPU are at the moment just a RUMOR that has not been confirmed.

[STaRGaZeR]

This is getting ridiculous.

Unless there is a major node overhaul (which reduces the power consumption by 30-50%) there is absolutely no way AMD can release a 8-core FX clocked to 5.0GHz. Nothing indicates that there would be a new node version coming and releasing one would not make any sense anyway. There is no point in using the limited resources to anything else but on the Gen. 3.

A nasty thing to say but using the limited resources on Piledriver would be like force feeding hay to a dead horse.

Currently the fastest 8-core FX is clocked to 4.0GHz.
At this frequency the 125W TDP envelope is already fully used in well optimized multithreaded applications (such as rendering or video encoding), there is no headroom really.

A proper binning of the dies can make around 200MHz of difference on current Piledriver dies.
The additional scalability comes purely from the lower leakage properties of the binned parts.

The current Piledriver cores do not scale well beyond 4.6GHz, which seem to be the critical frequency really.
They can be pushed beyond that but the increase in the power consumption & emitted thermal power can no longer be justified by the resulting performance increase. Beyond this point the performance-power curve is anything but linear.

We might see a 8-core FX (Piledriver) clocked to 4.3 - 4.5GHz (base frequency) at some point, but anything higher than that is very unlikely.

Even if AMD could produce a 8-core FX clocked to 5.0GHz, there would not be any motherboards for the chip.
A 5.0GHz 8-core Piledriver FX consumes 180W - 200W of power during Cinebench R11.5 for example (DCR Pmax, from CPU VRM).

In some applications the power consumption can be even 30% higher than that so the motherboard design has to have enough of headroom to accomodate the difference + the design margin. The motherboard would need to be able to deliver 235W - 260W of constant power. After the average conversion (VRM efficiency) loss the total VRM input power would be 294W - 325W (24.5 - 27A from the 12V rail).

The power dissipation of the CPU VRM alone would be 36W - 46W.

There are only couple of motherboards which can temporarily manage such of power requirements.
Even these motherboards cannot support such power requirements reliably for extended periods of time.

Having a 5.0GHz Piledriver based system in daily use is far from impossible.
It just requires conditions of which 99.95% of the users cannot attain.
A high performance custom watercooling system for the CPU and the motherboard (CPU VRM).

AMD and an enthusiast also have a slightly different definition for the term "long term reliability and stability".

Wait, you taking this seriously?

[tajoh111]

Clearly with some overclocked FX-8350 CPUs already running at 5.0 GHz. there is no issue with AMD binning and selling 5.0 GHz. Vishera CPUs should they desire. This ain't rocket science. I can run P95 on my FX-8350 @ 4.7 GHZ. for 25 hours without overheating @ 1.462 Vcore with an Xigmatek Aegir HSF. In normal operation this system rarely gets above 50C even under heavy video load.

If AMD were to offer a 5.0 GHz. Vishera - which they have never said they would, then they could if they so desired, include an H2O cooler as they have offered with the FX-8150. Remember people that the reports of AMD selling a 5 GHz. CPU are at the moment just a RUMOR that has not been confirmed.

Your ignoring the power envelope. Yes some GPU can get that high, but not without consuming a huge amount of power.

AMD killed their chance to sell 800 dollar processors when they made the black series FX highend lineup and gave them unlocked multipliers. At this point, AMD not only needs clocks to sell an expensive processor, they need moar cores(more than their regular fx line), similar to what Intel does with sandybridge E. A high clock alone isn't enough to sell a AMD CPU, especially when their brand is in the gutter as far as CPU's go.

[Rollo]

. A high clock alone isn't enough to sell a AMD CPU, especially when their brand is in the gutter as far as CPU's go.

Only to the untrained eye. They purposely have let their market share slip to 15% to give their cpu's an "edgy, rebel" cachet.

After they've established themselves as THE cpu for "in the know" elite system builders, they will pounce with $600 Centurions!

[AMDforME]

tajoh111-

I'm not ignoring anything. As I said some folks are already running 5.0 GHz. FX-8350s on existing high end AM3+ mobos. You're assuming that these 5.0 GHz. CPUs which no one knows if they will even be released or how they originate - will consume more power than current FX-8350's running 4.7-4.8 GHz. without issue. The 5.0 GHz. chips may not draw any significant increase in power, we just don't know. I have run my current FX-8350 @ 5.1 GHz. but it was not P95 stable so I throttled it back.

BTW you're hi-jacking this thread with nonsense about AMD sales... Their brand ain't in the gutter with them winning the contract for the next Xbox, PS4 and Wii consoles. The AMD brand is doing just fine, thank you. AMD is having no problem selling Vishera based FX chips as they are excellent performing CPUs at a great price.

AMD doesn't need more cores at all, but Intel could use a few more. AMD needs better front end schedulers and Steamroller will bring that before you know it. Hating on AMD is wasted energy as they just continue to crank out better laptop APUs, desktop APUs, low power consumption server APUs/CPUs with more goodies on the way. That discussion however is for another thread, not this one.

[Rollo]

.

BTW you're hi-jacking this thread with nonsense about AMD sales.. The AMD brand is doing just fine, thank you. AMD is having no problem selling Vishera based FX chips as they are excellent performing CPUs at a great price.

.

http://online.barrons.com/article/SB...888614202.html

Intel's (ticker: INTC) total fourth-quarter unit share increased 150 basis points sequentially to 84.8%, its highest since third-quarter 2002, with gains across the server, desktop and notebook segments.

Yeah, they're selling tons of those Visheras.

AMD has a BIG problem- their main market is the dwindling group of AMD motherboard owners who don't feel like swapping motherboards.

[STEvil]

Assuming they're gunning for the desktop market. /ot.

[PedantOne]

haha, 5GHz stock clock only in dreams! maybe TWKR 2 with 5 GHz guaranteed.

[Stukov]

I thought the rumor said 5ghz turbo speed?

[FlanK3r]

haha, 5GHz stock clock only in dreams! maybe TWKR 2 with 5 GHz guaranteed.

after The Stilt comment, I think it will be this (5 GHz quaranteed at air, low VID chips, ideal for LN2). But maybe we will see in future FX-8370 with 4.2GHz clock...

[Mungri]

*Le wild Intel I7 920 from 5 years ago enters the thread and asks:

'AMD, Why u no beat me yet????'.

[AMDforME]

http://online.barrons.com/article/SB...888614202.html

Yeah, they're selling tons of those Visheras.

AMD has a BIG problem- their main market is the dwindling group of AMD motherboard owners who don't feel like swapping motherboards.

You're completely cornfused Rollo or in denial.

Actually AMD are selling a lot of Visheras and also laptop APUs. In addition AMD's contract wins for the next PS, Xbox and Wii are estimated to represent sales of 5 million more APUs - which are premium priced unique APUs for the specific consoles. AMD's Steamroller is reported to drop into AM3+ mobos with just a BIOS update. That delivers greater value to current AMD customers and mobo mfgs. Seeing as though Vishera performs quite well and is an excellent value, AMD owners are not sacrificing anything other than being exploited by inflated Intel CPU prices.

Yes AMD is doing far better than the critics would like people to believe. AMD still has a long road to full recovery ahead to be at the top of their game in all market segments but Steamroller is on the way and it will improve FX CPU performance considerably. AMD's transition to 28nm node size for APU/CPU will also lower power consumption even more than the current APUs are now, which is excellent already.

AMD's eggs ain't all in one basket as their GPU sales are good and their ULV products are now going into tablets and other portables. AMD is expanding it's revenue streams and market base to continue their growth. Intel was shut out of all three popular consoles because their APUs are simply inferior to AMDs. AMD GPUs are doing well also. Nvidia was shut out on consoles sales because they don't have an APU - which is what the majority of all PCs and portable devices will be using in the not too distant future. AMD current leads in APU products and tech.

Those in denial need to take the blinders off and get in touch with reality. All of the above info. has been published by PC hardware sites and industry analyst so it's not speculation at all.

At this point the AMD 5.0 GHz. Centurion CPUs are a RUMOR but it would be cool if AMD decides to sell a few of these, though I think the $795. price tag is absurd and for those folks with more money than good judgement.

This thread is suppose to be about the AMD 5.0 GHz. CPU until it was hi-jacked with off-topic bogus rants about AMD's sales...

[Rollo]

You're completely cornfused Rollo or in denial.

Actually AMD are selling a lot of Visheras and also laptop APUs.

Those in denial need to take the blinders off and get in touch with reality. All of the above info. has been published by PC hardware sites and industry analyst so it's not speculation at all.

Denial is right-

http://www.pcworld.com/article/20337...free-fall.html

A “perfect storm” of struggling PC companies, aversion to Windows 8 and wider mobile-device adoption plunged the already struggling PC market into a free fall during the first quarter this year, IDC said.

That Q1 2013- sales in the whole industry way down. Yet according to you, AMD is kicking ass! Yay!

http://www.cpubenchmark.net/market_share.html

People that use Passmark have AMD at about 25% and intel at 75% in Q1 2013.

http://beta.fool.com/leokornsun/2013...her-dog/28650/

In the first quarter of 2013, AMD’s graphics card market share declined from 36.3% to 35.7% year-over-year, while Nvidia’s share rose from 63.4% to 65.7%.

Meanwhile, AMD’s share of the CPU market plunged to 16.70%, down from 27.70% a year earlier. Meanwhile, Intel’s CPU market share rose from 72.30% to 83.30% over the same period.

Between the third quarters of 2011 and 2012, AMD’s share of the notebook graphics business plunged from 52% to 34%, while Nvidia’s share rose from 49% to 66%.

With only $1 billion in cash, AMD is shouldering $2.04 billion in debt with a negative cash flow of $471 million, down 180.7% over the past three years. Meanwhile, expenses have risen 31.97% to a whopping $6.60 billion.

AMD has to post some strong top and bottom line growth to get out of this rut. However, those trends have been overwhelmingly negative over the past three years.

AMDforMe, some people back up what they say with links to sources. Some people post PR-speak and wishful thinking.

I'm the former.

[c22]

Rollo,
you pretty owned amd fanboy so to speak.

As to AMD 5 GHz CPU - even if they do it, even if they maintain stable clocks, even if it will be available widely.... for 800$ it would have to be faster than 3930K and I can already say that: it won't be.

[paulbagz]

As someone who just recently started caring about energy usage (paying 25c/kWh) I don't think I could ever go for a 200+w CPU ever again.

-PB

[Aberration]

With a handle like AMDforME I wonder why anyone would bother?

[Aberration]

As someone who just recently started caring about energy usage (paying 25c/kWh) I don't think I could ever go for a 200+w CPU ever again.

-PB

You still do the same calculations regardless. The newer CPU's are more power efficient than your QX. Given the same behavior and same workloads you should notice a decrease in cost.

[The Stilt]

after The Stilt comment, I think it will be this (5 GHz quaranteed at air, low VID chips, ideal for LN2). But maybe we will see in future FX-8370 with 4.2GHz clock...

Like said before chips which are ideal for air do not scale well under LN2 and vice versa.
There might be some exceptions, but this is the pattern in general.

Even the 5.0GHz would only be a boosted frequency it still would not be doable.
Some of data is based on actual measurements, some on estimation based on experience.

At the same voltage and cooling capacity an average W/MHz (watt per MHz) ratio is 0.016925, for an 8-core OR-C0 (Node 2) die with very low leakage properties. The value is far from absolute and it is not fixed either, but it is more than accurate enough for this purpose.

The official TDP for an FX-8350 operating at 4.0GHz is 125W, of which 118.25W is reserved for the compute units and 6.75W for the rest (NB, L3-cache, etc) of the CPU. Even if AMD could magically pull a rabbit out of a hat and make the die scale 1GHz higher at the same voltage levels as the FX-8350s currently operate: 118.25 + (1000 * 0.016925) + 6.75 = 141.925W. The 140W TDP envelope (with does not even exist for this infrastructure group) would be exceeded. The Thermal Design Power is not the absolute maximum amount of power the CPU can consume (or emitt). The actual maximum power can be 15-30% higher than the design power in some rare cases.

Case 125W TDP: Core Pmax = 135.9875W (1.15 * 118.25) to 153.725W (1.30 * 118.25).
Case 141.925W TDP: Core Pmax = 155.45125W (1.15 * 135.175) to 175.7275W (1.30 * 135.175).

The core voltage levels on FX-8350 range from 1.3000 to 1.3750V for the 4.0GHz base frequency clock profile.
The rated voltage depends on the leakage level of the die. The lowest voltage being used on the parts with the highest leakage properties and vice versa. TDC (Thermal Design Current) would be 91A (118.25 / 1.30) to 86A (118.25 / 1.375).

Under conventional cooling (tCTL <65 celsius under the load) even the best dies with the lowest leakage properties require at least 1.4500V+ to be even remotely stable at 5.0GHz, in a well controlled optimal enviroment. High leaking dies could reach this frequency at lower voltage, however they usually will overheat with anything higher than 1.4000V.

The difference in voltage appears to be small, in reality it is anything but.
Increasing the voltage of a semiconductor does not result a linear increase in power consumption.
Increasing the voltage will also increase the temperature, unless the cooling capacity is increased in the same proportion.
As the temperature increases so does the leakage. The temperature coefficient cannot be easily and accurately measured so I only have some crude estimations available.

The estimated W/mV (watt per millivolt) ratio is 0.350625W for a 8-core die, operating at <65 degree celsius.

Lets dig up the TDP estimation for a 5.0GHz clocked part, operating at the same voltage levels as current FX-8350s do.

141.925W - 6.75W = 135.175W

Required voltage increase: 75mV (1.4500 - 1.3750) to 150mV (1.4500 - 1.3000).

Increase in power: 26.296875W (75 * 0.350625) to 52.59375W (150 * 0.350625).

Total core power: 168.221875W (141.925 + 26.296875) to 194.51875W (141.925 + 52.59375).

New TDP: 174.971875W (168.221875 + 6.75) to 201.26875W (194.51875 + 6.75).

None of the current OR-C0 CPUs available to public come even close to the values displayed above.
Only some of the specially binned dies do.

Regarding the 5.0GHz being a boosted profile rather than a base frequency:

The power plane of the core is shared between the whole node (die).

For example;

PState Pb0 = 5.0GHz / 1.4500V, PState Pb1 = 4.8GHz / 1.4125V, PState PS0 = 4.4GHz / 1.3750V

During a light load CU0 - CU1 would operate at the boosted 5.0GHz frequency, while CU2 - CU3 would operate at the base 4.4GHz frequency. The effective core voltage would be 1.4500V as specified by the Pb0 PState. As the example above show, the frequency itself has only a very minor effect to the power consumption, while the voltage makes the actual difference.

In this scenario the TDP would be:

168.221875 / 2 = 84.1109375
5000 - 4400 = 600
600 * 0.016925 = 10.155
84.1109375 - 10.115 = 73.9959375
84.1109375 + 73.9959375 = 158.106875
158.106875 + 6.75 = 164.856875W

or

194.51875 / 2 = 97.259375
5000 - 4400 = 600
600 * 0.016925 = 10.155
97.259375 - 10.115 = 87.104375
97.259375 + 87.104375 = 184.36375
184.36375 + 6.75 = 191.11375W

In case where the 5.0GHz frequency would happen only during Turbo, the reduction in TDP would be only just over 10W.

AMD could only market the chip as being capable for 5.0GHz, however they cannot guarantee ANYTHING at those clocks.
Not with OR-C0 stepping and probably not with anything made on the current 32nm process.

[AMDforME]

Guys the insults prove you are immature and unable to have an intelligent coversation. Hating on reality don't change reality. If some of you socially and technically challenged folks spent more time reading and less time posting tripe in threads, then you'd already know that the information I posted was public knowledge. Of course if all you do is believe the foolishness posted by Intel fanbois or those in denial, then you're bound to be uninformed.


Stilt-

People are already running FX-8350s @ 5.0 GHz. The high end AM3+ mobo VRM circuits are able to supply the power as long as they are properly designed and cooled. With Asrocks F-Stream you can even monitor the VRM phases and Vcore - real time. On the Asrock 990FX Fatality Pro you need to add some airflow over the VRM heatsink if you want to run P95 @ 4.7+ GHz. for 24/7 operation but it can provide sufficient power for 5.0+ GHz. with cooling. Been there, done that.

[Gener_AL (UK)]

Currently running a 8120 using a h80 @4.8CPU 2.6NB and DDR 1600@1866.
pretty much 24/7 for all types of work and shock horror gaming.

* at stock it will run at 4.4cpu 2.4nb 24/7 (manually set NB or vid changes to 1.4v wtfkebab)

[highoctane]

Whats the saying, pot calling the kettle black...

[Rollo]

Guys the insults prove you are immature and unable to have an intelligent coversation. Hating on reality don't change reality. If some of you socially and technically challenged folks spent more time reading and less time posting tripe in threads, then you'd already know that the information I posted was public knowledge. Of course if all you do is believe the foolishness posted by Intel fanbois or those in denial, then you're bound to be uninformed.
.

Why don't you post some links to sources of this "information" then?

I think it would be great to see AMD is making a bunch of money and selling lots of parts.

Not like the lack of competition for Titan did me any favors. I was planning on buying two at the rumored $850 price, only to get smacked down by the $1000 price.

If AMD takes a dirt nap, I don't see computer gaming getting cheaper. So do tell, enlightened one- where are these great sales you speak of referenced and public knowledge?

[FlanK3r]

Stilt: I think it could be C1 revision with better power consumption. But it is only theory. I know, i have one FX-6300 low VID, he reached over 4950 MHz at 1.44V in load, but it is very hot chip after OC, because Δ voltage offset is the same as some middle VID chip with 1.5V in load. But if is it price around 750 dollars after that I hope for some bundle with AMD FX liquid cooler or something else.