AMD - Piledriver chitchat

[DanceswithUnix]

Some more rumours about next consoles. http://www.extremetech.com/gaming/14...4-and-xbox-720

Edit: That link says GDDR3 on PS4 but linked article says GDDR5, could be a typo? Either way, rumours, especially on finer details, still seem to vary a lot between sources ATM.

I see that article draws from http://www.nowgamer.com/news/1769561...y_insider.html, according to which:

"According to this insider, however, Oban is actually the name of the blitter inside the next-gen machine.

A blitter is a system built into hardware designed to improve the speed at which data is transferred within the hardware's memory. It is a separate system from the CPU and can run parallel to it, moving large amounts of data relatively quickly."

A Blitter? Blimey, should give an Amiga a run for its money then. 1980's called, they want their graphics technology back!

And XBox is a 356 bit system to make it better than the PS4 as a 256 bit system???

This sound to me like a wind up, and not a very good one at that.

[watercooled]

I remember SA lambasting some sites for referring to the console as Oban, saying it was actually the name of something else, the CPU SoC IIRC.

The blitter sounds like the 'data move engine' brought up earlier, so seems likely. About the separate system/application SoCs, that's in line with that pdf leaked a while back so could be plausible but still, 3 separate SoCs and a total of 12 x86 cores? I wonder if they'll be on the same package/interposer and have some sort of cache coherency? Maybe the system SoC will be completely hidden from applications though, like how baseband CPUs are essentially transparent to smartphone apps (probably not the best analogy as the baseband CPU is also independent from the phone OS AFAIK). If true, it could hopefully mean better power management and granularity so you don't have the issue you have with current consoles e.g. a blu-ray player pulling >150W.

On memory, there seem to be a few conflicting rumours, some claiming different technologies but few talk about bus width. Although, I wonder why they'd go with DDR3 over GDDR5 for such a wide bus? The DDR3 + GDDR5 sounds plausible, considering it's what PCs essentially use ATM, but two presumably separate memory pools could cause problems for moving data between them, and more problems if CPU+GPU share a package? Maybe that's where the blitter will help?

All very interesting IMO.

[CAT-THE-FIFTH]

According to CD,Intel has been strong-arming OEMs regarding AMD laptops:

http://semiaccurate.com/forums/showp...7&postcount=92

[watercooled]

Anyone surprised?

[Biscuit]

According to CD,Intel has been strong-arming OEMs regarding AMD laptops:

http://semiaccurate.com/forums/showp...7&postcount=92

Links down? Didn't SA get into trouble for putting up stuff like this a short while ago?

[DanceswithUnix]

Links down? Didn't SA get into trouble for putting up stuff like this a short while ago?

Intel have been repeatedly prosecuted for this sort of behaviour, so whilst I am sure they will try and bully Charlie into being quiet I would have thought he would be ok.

[CAT-THE-FIFTH]

Link is fine??

[Biscuit]

Link is fine??

Yeah seems okay now, dunno why it didn't work before

[CAT-THE-FIFTH]

TH tests a whole load of CPUs:

http://www.tomshardware.co.uk/gaming...w-32628-4.html

Whats weird,is that they seem to ignore their own latency graphs.

[CAT-THE-FIFTH]

Some A10-6800K results:

http://wccftech.com/amd-apu-performa...CCFtech.com%29

However,they are from OBR.

[DanceswithUnix]

Have seen a few reports now that there won't be an AMD 8000 series graphics until the end of year, which if true would make naming such as:

A10-6800K Radeon HD 8670D 844 MHz

seem unlikely.

http://www.fudzilla.com/home/item/30...graphics-cards

[Biscuit]

doesnt the mobile naming scheme tend to use the previous generations technology with the next generation names?

[scaryjim]

Kabini is rumoured to have HD8xxx graphics too, and (as Hexus mentioned a few weeks ago) OEMs like a yearly product refresh cycle, which doesn't naturally fit with technology refresh cycles. Wouldn't surprise me at all to see Richland and Kabini use 8xxx naming - particularly if they're going to bump up to Ax-6x00 namking: wouldn't make a lot of sense to bump the CPU number and not the graphics component number...

[CAT-THE-FIFTH]

A few FX4130 reviews:

http://www.vor tez.net/articles_page..._review,1.html

http://www.tweak.dk/review/AMD_FX-4130_CPU/1449/1/1

It appears to have 4MB of L3 cache like the FX4300,and it is still hard to say whether this is Bulldozer or Piledriver based.

OTH,the 125W TDP seems too high,as power consumption is the same as FX4100 and the stock cooler is even bigger than the the FX8350 one!!

AMD are a strange company!! They should fitting the FX4130 stock cooler to the FX6300.

Edit!!

Thread detailing AMD stock coolers:

http://forums.anandtech.com/showthread.php?t=2296689

It appears a new CM made cooler is coming with the FX8350.

[CAT-THE-FIFTH]

Some A10-6800K results:

http://wccftech.com/amd-apu-performa...CCFtech.com%29

However,they are from OBR.

The A6-5200 and A4-5000 are meant to be Jaguar based SOCs!!

They keep up with the desktop A4-6300!!

[kompukare]

Wiki seems to think the FX4130 is Zambezi (Bulldozer) not Vishera (Piledriver) and it was this as a reference. But that's dated August 2012 so not sure what's going on.

Thread detailing AMD stock coolers:

http://forums.anandtech.com/showthread.php?t=2296689

It appears a new CM made cooler is coming with the FX8350.

There was also the AT forum thread from Idontcare where he lapped his FX8350 and he was very impressed with the new cooler - for a stock cooler of course.

http://forums.anandtech.com/showthread.php?t=2300800