AMD "Piledriver" refresh of Zambezi - info, speculations, test, fans

[DedEmbryonicCe1]

with a nice air cooler and water cooling setup, enthusiasts can easily break the 5 – 5.5 GHz barrier with the new FX-Vishera processors.

What are they basing that on? Are there legitimate reports of 24/7 overclocks floating around or are they just assuming it's better than BD?

[informal]

No reports as far as I know,they are just assuming better OCing...

[EniGmA1987]

It is probably based on the new clock mesh info, people are just extrapolating current average overclocks and then adding an extra 15% clock speed on because the clock mesh is supposed to either allow 10-20% lower power consumption at the same clock speed, or 10-20% higher core speed with the same power consumption as before. Since average OC used to be about 4.6GHz, if we take that and add 10-20% then we get a range of 5-5.5GHz. The problem is that this has no basis on actual truth at all and we dont know if the clock mesh system even lives up to its promise. We also dont know if those speeds increases/power decreases are at any frequency or just a certain range.

[Olivon]

Only thing we know is that AMD demoed a 4.8GHz 1.45v specimen during IDF :



http://www.hardware.fr/news/12617/id...o-vishera.html

[tbone8ty]

yah but 4.8ghz at looks like 1.475 cpu volts depending on LLC

i dunno

we will see next week

regardless i think a majority of people here who have a 8150/8120 bulldozer r going to upgrade to the 8350/8320 vishera

and judging by the rumored prices leaks $199 for the 8350 seems pretty good

and besides this will probably be the last enthusiast 8-core chips from Amd for awhile now.

so pick one up for the nostalgia of it all lol

[BeepBeep2]

Only thing we know is that AMD demoed a 4.8GHz 1.45v specimen during IDF :



http://www.hardware.fr/news/12617/id...o-vishera.html

Well, if not cherry, then good.
If cherry, no better than BD...they could have easily set up a demo running wPrime 1024/Cinebench all day on a PC that wasn't Prime stable, or stable enough for say a folding/crunching program...
C0 stepping, but this has higher (Piledriver) IPC, right?

[sequoia464]

Well, if not cherry, then good.
If cherry, no better than BD...they could have easily set up a demo running wPrime 1024/Cinebench all day on a PC that wasn't Prime stable, or stable enough for say a folding/crunching program...
C0 stepping, but this has higher (Piledriver) IPC, right?

The one thing that I still haven't seen anywhere is power draw, I know the voltage is still close to BD, but the overall power usage is supposed to be less. Anyone seen anything on this??

On that CO stepping, CPUZ is still reporting this chip to be a Zambezi, rather than a Vishera so it looks like CPUZ isn't reporting everything accurately yet, even though it shows it as a FX-8350.
EDIT: or would that even affect the stepping report?

[tbone8ty]

http://hardforum.com/showthread.php?t=1722207

this guy said his whole system was 160w using kill-a-watt

[leo5111]

http://hardforum.com/showthread.php?t=1722207

this guy said his whole system was 160w using kill-a-watt

nice if true, not saying its not, but bulldozers numbers were way higher wernt they? so amd would realy have had to reduce power consumption to get it that low

[sdlvx]

Dig deeper in that thread. It pulls like 240 to 290w I believe, I forget the exact number, under load. It's when he plays xcom. 160w is idle.

[Evantaur]

http://hardforum.com/showthread.php?t=1722207

this guy said his whole system was 160w using kill-a-watt

but it's not overclocked or under load :P
so still a mystery

[cezar]

The countdown has begun



Source

[sdlvx]

http://hardforum.com/showpost.php?p=...&postcount=104

Found it. 295w stock, at load while playing a game. Unknown graphics card, but assuming he just threw the FX 8350 in his FX 8150 rig and is using an overclocked GTX 570 fermi. If that is the case, Piledriver might be not too bad.

test system with i7 965 EE quad and GTX 570:
http://www.bit-tech.net/hardware/201...1-3gb-review/4

Power usage on test system with stock i7 965 EE quad and GTX 570:
http://www.bit-tech.net/hardware/201...1-3gb-review/9

i7 Nehalem quad pulls 330w from the wall with that card and Piledriver 4 module pulls 295w, unless the difference in load between 3dmark06 canyon run and xcom is big, Piledriver looks good.

But all of this is with the assumption that he's still using the GTX 570. If he's not, I'm totally off.

[leo5111]

well his sig lists the 8350 with it, so hopefully thats what he is using

[Oese]

after reading the thread i think so 99,99%

[hiepgia]

Less than 13 hours beforce NDA lift....
Price as we know is true. And performance is improved too. It will beat Core i5 at the same price

[Alex-Ro]

Can't wait power and overclocking tests

[informal]

I wouldn't expect any miracles in power draw department. Maybe a bit better or on par with 8150. You still get better performance out of the box,even at the same clock, so it's an improvement.

[Evantaur]

now if only my vendor would let me buy one

i asked "is there any way to preorder one"

and they were like

[Oese]

:p

[cezar]

AMD to present Piledriver Opterons at SC12 conference:

http://www.cpu-world.com/news_2012/2...onference.html

[tbone8ty]

so whens official launch date?

[cezar]

Less than 13 hours beforce NDA lift....

So, tomorrow at 7 AM ???

[Alex-Ro]

No leaks till now....except obr but that doesnt count.

[cezar]

No leaks till now...

Are you replying to me? If yes, then look that hiepgia explicitly says it will be tomorrow.
But AMD in Japan says that the premiere is scheduled to October 23 - see my post #1087.
That's why I asked (him)