News - AMD set to release 5GHz-clocked FX-9590 CPU

[jigger]

Don't fall for the PR. What true desktop enthusiast gives a [y'know] about efficiency?

Frankly, we should be encouraging AMD to do this sort of thing. We, as enthusiasts, will be regretting it if we don't - when we're stuck with no-upgradeable laptop-grade parts in our gaming rigs

Very true. Ive got a few hundred quids worth of water cooling and a 1100 watt PSU that cost me a fortune that I would mind getting my moneys worth out of.

[keithwalton]

I doubt any current motherboards could sustain a 220w tdp cpu for long.
Even heavily overlocked chips dont run that hot. Would have to be decent water cooling, air wouldn't cut the mustard.

As for the overboost speed hope it is 4 cores, as a single core only at 5ghz is about the speed of a 2.5ghz haswell if not slower.

What has happened to there naming convention, is this a 9 core chip, as that's what there previous guide would of suggested.
8350
8 = 8 core
3 = bulldozer
50 = speed

From the leaks I thought someone at amd was wanting intel to bring forward the ivy-e release!

[watercooled]

The second digit as 3 = Piledriver, 1 was Bulldozer.

Again, clockspeed is a useless metric for comparison.

And even if the 220W TDP is true (as DanceswithUnix said, something doesn't sit right about it), then (also) again, you may be surprised by how much overclocked CPUs do use - it's not linear with clock speed, and there's a reason you can't OC very far with a stock heatsink, which is usually designed for roughly the TDP of the chip.

[flearider]

part I don't like is it's 5ghz at turbo .. does anyone use that ? it's the first thing you turn off for overclocking ..bet it can't do more than my 8350 stable which will go to 5.4 but like to keep it at 5 and under 1.45v

[miniyazz]

I doubt any current motherboards could sustain a 220w tdp cpu for long.
Even heavily overlocked chips dont run that hot. Would have to be decent water cooling, air wouldn't cut the mustard.

Unfortunately all the CPU power calculators online seem to be down at the moment, but I disagree. Power consumption rapidly rises with volts and clock speeds (I think my i7 920 was estimated at >300W at one point) and 220W is certainly achievable with decent air cooling.

[watercooled]

TBH it's generally possible to get some CPUs to draw over their TDP even at stock, with certain loads.

[keithwalton]

The second digit as 3 = Piledriver, 1 was Bulldozer.

Again, clockspeed is a useless metric for comparison.

And even if the 220W TDP is true (as DanceswithUnix said, something doesn't sit right about it), then (also) again, you may be surprised by how much overclocked CPUs do use - it's not linear with clock speed, and there's a reason you can't OC very far with a stock heatsink, which is usually designed for roughly the TDP of the chip.

My bad on mixing up bulldozer and piledriver codes, does seem though that a naming convention lasted 2 generation before they've thrown it away.

Power consumption is linear with clockspeed, and square of voltage. (so a 10% voltage increase is equiv to 21% clockspeed gain)
Which suggests they've had to significantly turn up the vcore to achieve the turbo speed.

I have a Q6600 (hot running chip by intel's standards at 130w) which has been overclocked by 33% from day one. But I run stock vcore
so tdp comes out at around 175w it's watercooled in series with my gpu with a substantial radiator that's external to my house which keeps it cool and quiet.

I suppose amd now have experience in designing chips with such a high tdp as just look at the current crop of graphics cards, they're 200w+ monsters but they run extremely hot.

[watercooled]

No, it's not even as simple as that. You seem to be referring to an *approximate* calculation (it only truly applies to a simplistic model) (and, unless I'm misunderstanding what you wrote, a 10% voltage increase does not guarantee you'll be able to increase clock by 21%) which may give ballpark figures but is usually miles off nowadays. It gets far more complex when you consider granular power gating of modern CPUs, leakage, etc.

But let me rephrase/clarify my original statement to stand up to pedants: power consumption is not linear with clock speed, or accurately predictable, when considering the fact voltage often also requires increasing to ensure stability at the higher clock. My point being, you must expect a fair jump in power consumption when overclocking a CPU, more so than a simple linear estimation based on clock speed. 220W, if it's even true, is nothing unusual in the world of overclocking.

CPUs cannot be compared to GPUs - they're vastly different, made using different lithography processes, and it's generally acceptable for GPUs to run far hotter than CPUs, 100C load isn't unheard of especially for older models *cough*fermi*cough*.

[hero1]

I wonder how hot those two chips will run. My i5 3570K @4.4 runs at 47 degrees Celsius with custom loop when it is fully,100%, stressed. It's time they keep on improving their CPUs to catch up to Intel.

[Jimbo75]

I wonder how hot those two chips will run. My i5 3570K @4.4 runs at 47 degrees Celsius with custom loop when it is fully,100%, stressed. It's time they keep on improving their CPUs to catch up to Intel.

What's the weather like on Pluto these days anyway?

[DanceswithUnix]

Got that earlier calculation wrong didn't I, from basing it on 8350.

Richland can do 4.4GHz in a 100W envelope, with some of that used by graphics.

So, that's a 14% boost over Richland, with at least 40W (40%) thermal budget to do it and stay within the 140W AM3 limit.

The thing is, it doesn't matter whether AMD can manage a massive TDP in a CPU. They couldn't get motherboard companies to make proper Bulldozer socket motherboards, they had to cut it back into AM3+ losing access to the PCIe lanes built into the CPU. If they can't get boards made for features they need, what chance have they got to get big VRM changes made to boards that are going to sell very few?

So, if there is an increased TDP, then I am guessing they went to ASUS and Gigabyte and asked what their top of the range motherboards are actually capable of, and tuned for that. Then you have at least two qualified motherboards right off the bat. But then what happens if you plug it into a 95W capable uATX board? Because some muppet will try it.

The more I think about it, the more I think people are just guessing.

[flearider]

I wonder how hot those two chips will run. My i5 3570K @4.4 runs at 47 degrees Celsius with custom loop when it is fully,100%, stressed. It's time they keep on improving their CPUs to catch up to Intel.

and my 8350 only does 44-46c @5ghz with 1000rpm fans whats your point ? its down to cooling

[DanceswithUnix]

and my 8350 only does 44-46c @5ghz with 1000rpm fans whats your point ? its down to cooling

Yeah 47 on water does sound a bit high doesn't it, but I find it is usually best not to reply to such posts.

You certainly don't want to point out to such people that their CPU is running the AMD64 instruction set

[AETAaAS]

AMD showed off the chip at E3. Looks like they may package it with a water cooler as they did some of the previous chips... (skip to around 2 minutes if you don't want to see the laptops)
http://www.youtube.com/watch?v=RCyfk4y50ak

I am not familiar with water cooling (air cooling guy myself) but could anyone identify the pump/block? If it's bespoke, it might be what comes along later. It looks to have some kind of chromed piping which is unique AFAIK for any AIO solution. Can't see the rad setup as well but I suspect it will probably be a single thick 120 for compatibility. However this IS an enthusiast chip, so more extreme solutions (eg. 240) could still be on the table.

[Captmario]

What's the point of having 5 GHz when its slower then intel in most of the stuff? also not to forget such high TDP? won't be a good overclocker.

[iranu]

What's the point of having 5 GHz when its slower then intel in most of the stuff? also not to forget such high TDP? won't be a good overclocker.

Because there are still loads of people out there who are still living in 2004 who think Ghz equals speed. I was reading the comments on a couple of tech sites and I was surprised by just how many people simply don't have a clue.

Eg:

Either AMD just threw a knock out punch to Intel, Or this is the beginning of round 2. Looks like it might be time to upgrade

They do this every now and then. The AMD K6/III and AMD64 kinda caught Intel with their pants down. AMD then decided to do an Intel and hike the prices, Intel released the Core series and stomped AMD back into second place again.

If this is another AMD blinder being pulled, I'd suggest grab it now before they hike the prices and give Intel some wiggle room to stomp on them again.