The official AMD Piledriver Discussion Thread

[Blitzvogel]

I've been browsing Trinity benchmarks, and judging from those, the Piledriver cores/modules seem to be quite an improvement in efficiency, even coming close to full 4 module/8 thread Bulldozers and beating out Llano. Considering Llano's wider cores to Trinity's, this is a pretty impressive increase even if it came at higher clock speeds than Llano to make it happen. In the end it comes down to performance per watt and cost to purchase the product more than anything.

I would consider Trinity before I would consider a dual-module Piledriver simply because of the great value of OpenCL and decent IGP graphics in Trinity and suspecting that like with the FX-41xx, dual module Piledriver products will probably be 4 moduled CPUs with a couple of defective modules. I don't care for the power leakage and high TDW that is going to come with that, unless the price is substantially lower than Trinity. The only other argument there is the upgrade path, of which AMD botched with first generation Fusion.

I really wonder if it's too late for AMD to turn things around.

 

[sequoia464]

I'm keeping it stock for awhile. I have a CoolerMaster Hyper 212+ with 2 fans(push/pull) and ArticSilver5. So far the 8150 is ryunning very cool, I'm still installing win7-64 SP1 so it will be a long night. I don't intend on OC like crazy. Just want to see how thios chip performs at stock. It is cool to see 8 CPU temps in my licensed version of Aida64😀

They are fun to play with - I was using a 6100 for a while.

I don't think that raising the FSB is as important on these as the Phenoms, I have seen quite a few overclock with just a multiplier increase - when you do overclock that is.

 

[Dresdenboy]

There is that wrong assumption on IT tech forums that somehow 2500K is better chip than 8150 ,in everything. That is wrong since unlike in games, 8150 is faster in many today's desktop workloads since they are more and more MTed. Without SMT,SB is not able to beat 8150 in MT workloads.

I think the perception derives from strong single/low threaded performance. It's the same reason the 2500K comes close to the 2600K in some benchmarks - with the advantage that no background task could disturb a running thread on a core.

 

[Rickyyy369]

This was posted by a guy over at extremesystems:

http://666kb.com/i/c6nq04mym74m5zulu.png

Just shows us the base clocks and features, but it also says that the expected release date is "late Q3 2012", so that gives me some hope that we wont have to wait until Q4 or next year to see these chips floating around.

 

[Puppies04]

It's a server processor first with the desktop as an afterthought, much like every CPU made in the past 5 years

If you are talking about FX and 2011 chips then this is correct, of course that completly ignores 1155 (SB+IB) which blatanly aren't server chips. They are infact mobile chips with their power drinking appetites and therefore performance greatly increased.