AMD - Piledriver chitchat

[CountBartok]

This is interesting:

http://translate.google.com/translat...den%2F&act=url

An AMD document talking of a native 8 module CPU.

Great find CAT-THE-FIFTH.

More here: http://www.techpowerup.com/196952/am...ll-uncore.html
"[I]AMD released developer documentation for a new processor it's working on, and the way it's worded describes a chip with 8 modules, working out to 16 cores, on a single piece of silicon, referred to as Family 15h Models 30h - 3fh. This is not to be confused with the company's Opteron 6300-series "Abu Dhabi" chips, which are multi-chip modules of two 8-core dies, in the G34 package. "

And I would love this to make to desktop boards please

[Noxvayl]

Me too, would love to see a 16 core beast available from AMD. I expect it to arrive once DDR4 is commercialised for desktops though which could be a year or two away. Would be perfect timing for me because by then I'd be finished uni and hopefully getting my first full time pay-check which will provide some upgrade funds

If it takes that long to come to market and we are able to get wind of it then you can expect Intel to have heard about it and will likely have something planned for their 2011 platform, or what ever replaces that. It is still good news, if true, that AMD are working on a high end chip like that though because they have been vacant in the high end desktop market for a while now.

[DeArmittE]

I think I should probably start catching up on all of the CPU news. Still content with my good old phenom 965 BE though

[CountBartok]

Had a 965be for quite some time before I upgraded. Was great chip and overclocked well too. I used to love my x3 720be as well.
But my favorite AMD chip was a K6-III, that thing used to fly back in dayhttp://en.wikipedia.org/wiki/AMD_K6-III

[CAT-THE-FIFTH]

Some Kaveri mini-ITX build work logs:

http://www.overclock.net/t/1448140/b...ect-kaveri-itx

[CountBartok]

Socket FS1b will become socket AM1 and will get some Jaguar core based chips.

http://vr-zone.com/articles/amds-fs1...and/69950.html

[CAT-THE-FIFTH]

A couple of FM2+ motherboards tested with Kaveri:

http://uk.hardware.info/reviews/5138...gy-consumption

It seems there is a massive variation in power consumption.

[mikeo01]

A couple of FM2+ motherboards tested with Kaveri:

http://uk.hardware.info/reviews/5138...gy-consumption

It seems there is a massive variation in power consumption.

That is a massive difference. Especially the maximum power consumption graph.

Gigabyte's flagship G1.Sniper surprisingly eats up quite (168W) a bit whilst ASUS which are known for their great VRM efficiency comes out at 114W.

Gigabyte G1.Sniper only uses a 6 phase whilst ASUS uses a 8 phase and topping out with Asrock with 10 phases. Wonder if this is simply an efficiency issue (as they have to provide for the iGPU too) which can be fixed easily or if Kaveri is somewhat more taxing on the mosfets?


I am by no means knowledgeable in power design but either way that's already put me off some boards.

[Biscuit]

It would be interesting to see how much difference you could make by twiddling with bios settings. e.g. turning all the uneeded features off.

[CAT-THE-FIFTH]

AMD Warsaw is on pre-order now:

http://www.kitguru.net/components/cp...lable-for-663/

16 cores at only a 99W TDP.

[scaryjim]

A couple of FM2+ motherboards tested with Kaveri:

It seems there is a massive variation in power consumption.

Not just a massive variation, but a significant one. AMD's own test platform uses a Gigabyte F2A88XN-wifi, so any reviews of the reference system are likely reporting high power draw and blaming the chip, when it's actually the mobo adding 15% - 20% on top of the best case scenario. Makes a mockery of the whole thing (IMNSHO).

[DanceswithUnix]

It would be interesting to see how much difference you could make by twiddling with bios settings. e.g. turning all the uneeded features off.

Probably not. Most motherboards seem to have nasty cheap linear regulators feeding parts of the board. People have managed to get motherboard overheads down quite low, but only using a soldering iron to remove or replace components.

[CAT-THE-FIFTH]

Not just a massive variation, but a significant one. AMD's own test platform uses a Gigabyte F2A88XN-wifi, so any reviews of the reference system are likely reporting high power draw and blaming the chip, when it's actually the mobo adding 15% - 20% on top of the best case scenario. Makes a mockery of the whole thing (IMNSHO).

Its kind of a fail!

AMD marketing needs to vet motherboards - I suspect they look at overclocking more than power consumption. The standard AM3+ test motherboard was the Asus Crossfire V which is poor for power consumption.

The Gigabyte is known to have highish power consumption too IIRC.

Edit!!

Hexus even mentioned their test motherboard consumed more power too.

[DanceswithUnix]

Perhaps the good power consumption boards are bad for overclocking or similar?

[watercooled]

A couple of FM2+ motherboards tested with Kaveri:

http://uk.hardware.info/reviews/5138...gy-consumption

It seems there is a massive variation in power consumption.

Just as I predicted. Although TBH those deltas are bigger than even I expected!

[CAT-THE-FIFTH]

OcUK are selling an 8GB 2400MHZ kit for £56. High speed RAM is not expensive.