Piledriver vs. Bulldozer - Page 3  

lmao thats what its there for actually

Edited by pwnzilla61 - 5/17/12 at 11:04pm

^ uhm not sure why that is in the quote box.

The CPU cores in Trinity already look like they're better, clock for clock, than a Bulldozer, and the Trinity cores aren't even full-on Piledriver. The desktop versions will probably be better optimized for desktop use, clock much higher, and have a lot more cache. I was just hoping they could match the IPC of a Thuban, while being able to clock it far above what a Thuban could get to. It looks like to me that they're on target to do that.

You're not going to get the IPC of an Intel chip from this architecture, but maybe you can get enough clock speed to catch up to them. It's not unrealistic, seeing as how you can get an 8120/8150 to the level of a 2500K now.

It will also be adding FMA3 which will probably get used since Intel will support it as well.
FMA4 was a big killer for bulldozer since intel was meant to support it and add it to there compiler but they dropped it at the last moment and went to FMA3 instead.

But ill be waiting for some real tests of pile driver before I descide to drop the money on it.
5ghz on bulldozer is no slouch, So for me to get pile driver it would atleast have to be able to high that, But I can always hope that it will be capable of more and a 10-15% performance increase on top of that.

I don't know why everyone keeps saying BD is a failure or that the FMA3/FMA4 debacle was a "big killer". Even with that if you remove the obviously titled benchmarks coded towards intel cpu's meanwhile handicapping non intel chips, BD is still VERY competitive. Anyone who thinks that buying a CPU that beats all others in single threaded apps and uses that as a justification is ignoring how the landscape of computing has changed. Sure Intel has some very nice high end processors but switch down to middle to low end CPU's and the results are not so clear cut. I have yet to see how in that realm any of the CPU's from either maker are "failures". Unless your going to use your CPU to run only benchmarks and brag about it, I fail to see the point. It really comes down to value for the money. Even gaming will see more benefit from GPU increases then CPU (with very few exceptions). Where BD shines and PD looks to improve on even more is in the things we do with our PC's all the time. Multi-threaded apps, video conversion etc. Honestly looking at the everyday app and even gaming performance if I had to choose between the 8120/8150 or the 2500K/2600K it would really come down to which is the better buy. Most people won't notice much difference at all. I almost went with the i5 series but for me I had a set amount of money I wanted to spend on the CPU (I don't feel for me personally spending even more on a CPU is going to benefit the uses I have for it and will be a waste). So the choice came down to a 6100, or locked i5. I chose the one that allows me to have a little fun overclocking while still keeping within the range that I wanted to spend. And I couldn't be happier with this chip. So again, I fail to see the "failure" or the big "killer" that everyone keeps talking about. Seems to me it's more perceived than actual.

I expect a 15% IPC increase with similar overclocking headroom. Not enough to make anyone change their mind about the BD arch in general but enough that it's a good step in the right direction for AMD. Steamroller will be what makes or breaks AMD in the desktop space not Piledriver.

You got the time line wrong though.

AMD first used FMA3, Intel switched from FMA4 to FMA3, then AMD switched from FMA3 to FMA4.

That said FMA is not what "killed" BD.