AMD unleashes first ever commercial “5GHz” CPU, the FX-9590

[SlowSpyder]

It'd be awesome if Intel responded with a Haswell that turbo'd to 5GHz.

 

[BallaTheFeared]

There is no point in debating it Gus. Just accept the cold hard reality that the FX-8350 uses a scant 10W while the motherboard uses 350W and you'll feel better about the world.

Makes perfect sense to me, when my Haswell system is drawing 210w at the wall at 4.7GHz in a linpack it's because my video cards are using most of it :thumbsup:

 

[sushiwarrior]

Makes perfect sense to me, when my Haswell system is drawing 210w at the wall at 4.7GHz in a linpack it's because my video cards are using most of it :thumbsup:

Could be that pesky DDR3 high power consumption too. You know the rule, 1W per 10MHz! My new 2400MHz sticks are killing my power bill.

 

[Tsavo]

I'll bet you chewy chocolate cookie that no a single person who is in the market for these would ever even consider using a stock cooler.

I'd consider using a stock cooler for a bag of brownies.

 

[PliotronX]

It'd be awesome if Intel responded with a Haswell that turbo'd to 5GHz.

I surely would hope they'd use a better TIM for the IHS in that case though.

 

[BallaTheFeared]

Haswell - Extremely Actually Soldered Edition ™

 

[ShintaiDK]

I surely would hope they'd use a better TIM for the IHS in that case though.

The TIM is quite good. Its the gap thats the issue.

 

[ViRGE]

I can't wait to see a review of the AMD FX 9590 at Anandtech

I wouldn't hold my breath. They didn't sample this chip ahead of time. And since it's OEM-only they likely never will.

 

[UaVaj]

could care less about the TDP. just buy a bigger psu and bigger hsf. problem solved.

as for IPC. as long as it is better than IVB-E.

will wait for the BF3/BF4, Crysis3, FarCry3, Metro, SC2, WOW-MoP benchmarks.

until then. intel wins.

 

[Imouto]

The TIM is quite good. Its the gap thats the issue.

The problem is Intel doing it on purpose.

 

[Magic Carpet]

The TIM is quite good. Its the gap thats the issue.

Wouldn't the gap act as a safety system against too much pressure when fitting heatsinks?

 

[Remobz]

could care less about the TDP. just buy a bigger psu and bigger hsf. problem solved.

as for IPC. as long as it is better than IVB-E.

will wait for the BF3/BF4, Crysis3, FarCry3, Metro, SC2, WOW-MoP benchmarks.

until then. intel wins.

True dat.

 

[ShintaiDK]

The problem is Intel doing it on purpose.

Do you have any edvidence on that?

 

[ShintaiDK]

Wouldn't the gap act as a safety system against too much pressure when fitting heatsinks?

There needs to be a gap obviously. But the current gap is, usually, a little too much when OCing.

They obviously tuned the assembly systems so it complied with all stock requirements. But they didnt take the time to fine adjust it further.

That might also explain the overclocking differences between if they come from Costa Rica or Malaysia.

 

[Abwx]

So you have one do you? Post screenshots of this pls

It s , well , an opinion , i ve no FX and indeed
i m not an overclocker at all.

So what is it again, a cherry-picked 8350?

A cherry picked Vishera , the namings only comes after due binnings..

 

[Imouto]

Do you have any edvidence on that?

You must be really naive if you think that going from a reliable and performing soldered solution to a crappy TIM one for two generations in a row isn't on purpose.

Wait, we are talking about your worshiped Intel here. My bad.

 

[Abwx]

Amazingly , there s quite a lot of animals by there ,
two little felines , a big one as well as a cow ,
but after all that s a thread about a beast....

 

[Magic Carpet]

There needs to be a gap obviously. But the current gap is, usually, a little too much when OCing.

They obviously tuned the assembly systems so it complied with all stock requirements. But they didnt take the time to fine adjust it further.

That might also explain the overclocking differences between if they come from Costa Rica or Malaysia.

It would be interesting to bench/delid cpus from different factories and see if there is any difference. I wonder, if the later batches will be more mature and OC friendly. Maybe IDC will get around to it, at some point

On topic, if the upcoming AMD cpus are within 150W TDP, I might assemble a few systems, if the price is right, of course.

 

[Vesku]

There needs to be a gap obviously. But the current gap is, usually, a little too much when OCing.

They obviously tuned the assembly systems so it complied with all stock requirements. But they didnt take the time to fine adjust it further.

That might also explain the overclocking differences between if they come from Costa Rica or Malaysia.

More like having an adhesive assembly with that little variance would eat up whatever they saved ditching solder.

 

[Tsavo]

Do you have any edvidence on that?

NSA surveillance gap.

 

[Imouto]

More like having an adhesive assembly with that little variance would eat up whatever they saved ditching solder.

There's a lot of workarounds if heatsink pressure is a problem. A bracket around the die resting on the PCB seems like a good idea.

 

[galego]

It will work at 8350 frequencies with a lower voltage ,
so no , it s not an overclocked 8350.

Exact.

 

[guskline]

NSA surveillance gap.

Sad but FUNNY! Getting back to the topic, I agree with Virge on the fact that it is highly unlikely we will see a test on the 9590 unless one of us is lucky enough to snag one.

 

[Idontcare]

You must be really naive if you think that going from a reliable and performing soldered solution to a crappy TIM one for two generations in a row isn't on purpose.

Wait, we are talking about your worshiped Intel here. My bad.

Imouto, I don't know your background so I have no idea if you are aware of what goes into the engineering trade-offs that a company might make in the pursuit of developing a product that performs consistently.

Consistency is key when trying to lower your cost of production per part. Engineering is rarely tasked with the goal of producing the highest performing product "regardless of cost". Outside of the NASA program to get to the moon, and a few other aerospace programs, there are very few engineering projects in this world where engineers got to do what they wanted so long as performance was improved.

Intel's situation with the gap is very much on purpose, no question, but you seem a bit confused as to what the purpose is. The purpose is not for "milking", it is all about "tolerance" (as in the engineering kind, machinery being involved and all) and the cost invoked when increasing engineering tolerances for mating multiple parts.

The engine in your car would be silly expensive if engineered to the tolerances of the gap between the IHS and CPU. Intel is already designing their products with a spendy budget in mind.

Could it be better? Sure. But at what cost? And with what tradeoff?

We all know there are legitimate concerns regarding stress-strain curves and the performance of finfet devices, that isn't rocket science nor is it a national secret. So it is of no surprise that Intel has engineered in a sacrificial slip layer to accommodate the unavoidable mismatch in coefficients of thermal expansion.

What none of us know in the public domain is how much of a performance vs cost tradeoff did Intel take in the change? Was it pennies or was it dollars (or tens of dollars)?

Arguing about how much better Intel could have made the 22nm IHS situation, in the absence of a balanced discussion regarding cost-adders, is no different than arguing that their stock HSF is going to be silly if one tries to overclock to 5GHz...nothing about that thermal conduction has been optimized for 200W power dissipation - not the IHS, not the CPU TIM, not the stock HSF.

It is all optimized for 4GHz operations and with a bevy of engineering related thermal stress concerns in mind. None of us know enough about the specific decisions made at Intel to second-guess the validity of those decisions. Project management triangle FTW

 

[galego]

The engine in your car would be silly expensive if engineered to the tolerances of the gap between the IHS and CPU.

This is one of the worst car analogies that I have read.

What none of us know in the public domain is how much of a performance vs cost tradeoff did Intel take in the change? Was it pennies or was it dollars (or tens of dollars)?

We know Intel would not go to bankruptcy with a better design.