AMD - Piledriver chitchat

[Bagnaj97]

There used to be plenty of microATX boards but they seem to have disappeared from the market lately.

I'm not sure about mITX but unlike more modern platforms AM3+ has a Northbridge and Southbridge to fit on the board.

The matx am3+ boards were all using older chipsets (8x0), and presumably they also had both NB and SB.

[scaryjim]

They mostly actually used the ancient 760G chipset, which was launched in january 2009. ISTR finding one ASRock board which used the upgraded 785G. Nothing based on the 790GX/890GX chipsets though, which are the ones they really should've gone with (although tbf there were never many 790GX/890GX mATX boards anyway).

mATX board choice for socket AM3+ has always been horrific.

EDIT: just checked that ASRock board: http://www.ebuyer.com/616248-asrock-...x-985gm-gs3-fx For all it uses a half decent chipset the VRMs look pretty weedy and have no cooling, and it's only got 2 RAM slots. So not really much of an improvement on the 760G boards...

[CAT-THE-FIFTH]

There was an HP OEM one.However, ASRock have recently released this mATX 970 motherboard:http://www.asrock.com/mb/AMD/970M%20Pro3/

Its available in the US.

[CAT-THE-FIFTH]

Dabs lists it for around £25:

http://www.dabs.com/products/asrock-...7DHBoCU3zw_wcB

[shaithis]

Dabs lists it for around £25:

http://www.dabs.com/products/asrock-...7DHBoCU3zw_wcB

That one is full ATX

[CAT-THE-FIFTH]

That one is full ATX

Oops,wrong link:

http://www.dabs.com/products/asrock-...970%20m&src=16

Also meant £52 too.

[watercooled]

Just spotted there's a Techreport podcast with David Kanter covering Zen on Twitch: http://www.twitch.tv/thetechreport/b/657091841

Not watched it yet but being Twitch, it will only be up for a short while if you're interested.

[CAT-THE-FIFTH]

Carrizo L is released:

http://www.techradar.com/news/comput...-boost-1293290

http://www.pcworld.com/article/29201...ors-first.html

[DanceswithUnix]

Hopefully Carrizo-L will turn up here faster than E350 did, which I think went to China first as well.

I notice Ebuyer have the AM1 chips at around £5 off each, which is quite a margin on something that cheap to start with.

[DanceswithUnix]

Fudzilla are saying Godovari is announced: http://www.fudzilla.com/news/process...n-godavari-out

GCN 1.2? Now ISTR AMD don't talk about version numbers of their GCN parts, so that is sort of meaningless, but I would usually take that to mean R9 285 shader based. If it can use make more efficient use of bus bandwidth like the 285 can, that might be interesting. Was expecting the graphics to be the same as Kaveri, but then SA point out that the part numbers imply there are silicon changes

http://semiaccurate.com/2015/05/13/b...godavari-apus/

[watercooled]

Yeah I'm really not sure what Godavari is supposed to be, I was expecting a refresh with maybe different binning and clock speed bumps but a couple of places are suggesting otherwise.

[shaithis]

So much conflicting information about the web over these

http://www.fudzilla.com/news/process...n-godavari-out

[CAT-THE-FIFTH]

So much conflicting information about the web over these

http://www.fudzilla.com/news/process...n-godavari-out

AMD's marketing(lack off) strikes again.

[CAT-THE-FIFTH]

[watercooled]

I've just noticed something on the HBM slides: http://images.anandtech.com/doci/926...ctor_575px.jpg

The core die is smaller on the (possibly hypothetical) HBM GPU vs the 290X? It could just be that the images aren't to scale or are plain mocked-up, but if true maybe it implies either a second, smaller HBM GPU or that HBM saves a large amount of die space on the flagship?

The latter seems less likely though; memory IO does take up a large amount of die space but considering the rumoured increase in shader count, an overall reduction in die size seems unlikely.

[kalniel]

I've just noticed something on the HBM slides: http://images.anandtech.com/doci/926...ctor_575px.jpg

The core die is smaller on the (possibly hypothetical) HBM GPU vs the 290X? It could just be that the images aren't to scale or are plain mocked-up, but if true maybe it implies either a second, smaller HBM GPU or that HBM saves a large amount of die space on the flagship?

The latter seems less likely though; memory IO does take up a large amount of die space but considering the rumoured increase in shader count, an overall reduction in die size seems unlikely.

edit: I see what you mean, but it's a very small difference - almost certainly just an illustration to make a point rather than real die size (though there should be some die saving from the mem IO as you say - in some years they were effectively limited by packing size due to pin outs for memory, not sure if that's the case recently, possibly.)