How will 'Piledriver' stack up against 'Sandy bridge' ???

[nomad_1911]

Hello everyone,since my pc is built on an am3+ platform with phenom II procesor,i'm kind of excited about the upcoming piledriver chips.Now i've heard that 'Piledriver' will be 10% faster than Bulldozer,but how much of an improvement those 10% actually are compared to 'Sandy bridge'.Will 'Piledriver' be able to compete with Intels offerings(performance per clock),cheers !!!

 

[Lucem Ferre]

Bulldozer is currently about 35% slower then Ivy bridge so piledriver should be 25% slower then ivy bridge.

Also amd cannot compete with intel's far superior IPC. The only place piledriver will compete in is multithreaded apps. In gaming it should be slaughtered.

 

[popatim]

AMD had announced last year that were no longer competing with Intel to create powerful cpu's and instead concentrating on 'user experience'

 

[jaguarskx]

If Piledriver can manage a 10% performance increase over Phenom II / FX, then it should be equivalent to the 1st generation Core i3/i5/i7 CPUs. It would need to boost performance by about 23% - 25% to get to Sandy Bridge. To equal Ivy Bridge, it would have to be an increase of 30%.

 

[abitoms]

If Piledriver can manage a 10% performance increase over Phenom II / FX, then it should be equivalent to the 1st generation Core i3/i5/i7 CPUs. It would need to boost performance by about 23% - 25% to get to Sandy Bridge. To equal Ivy Bridge, it would have to be an increase of 30%.

from what i heard / read, Ivy is about 6% faster than sandy. Sandy is about 25% faster than BD and Ivy is 6% faster than Sandy so Ivy is 25%(100%+6%) = 25(1+0.06) = 26.5% faster.

So, Ivy is 26.5% faster than BD.

I think you took the percent that Ivy is faster than Sandy (around 6%) and just added it to the lead Sandy has over BD (25%) to arrive at the lead Ivy has over BD(31 or 30%).
Not to worry, it is a common error, and actually works in some scenarios

 

[Anonymous_26]

Even if Piledriver does turn out to be half decent it will only perform as good as the first generation I Core procssors so it will still be behind Intel by two generations.

 

[$hawn]

No one knows, until the PD is released

However, if u want some guestimation for the FX 8350 performance increase over FX 8120, here it goes,

+15% due to improved IPC (See desktop Trinty review)
+ 2% due to 8MB L3 cache (absent on Trinity)
+ 10% due to improved clocks (Cyclos technology)
+ 2% better clocks due to reduced leakage ( GF process matured quite a bit over the past year)

Totally, that adds up to, 1.15*1.02*1.1*1.02 = 1.31, ie around 31% faster on average

So then, you tell me now, how does PD stack up against SB/IB ?? :lol:

 

[nomad_1911]

Thanks for the info guys,but i guess we will have to wait for 'Piledriver' to see the actual performance,but i dont expect anything radical,cheers !!!

 

[jaguarskx]

from what i heard / read, Ivy is about 6% faster than sandy. Sandy is about 25% faster than BD and Ivy is 6% faster than Sandy so Ivy is 25%(100%+6%) = 25(1+0.06) = 26.5% faster.

So, Ivy is 26.5% faster than BD.

I think you took the percent that Ivy is faster than Sandy (around 6%) and just added it to the lead Sandy has over BD (25%) to arrive at the lead Ivy has over BD(31 or 30%).
Not to worry, it is a common error, and actually works in some scenarios

It works like this:

Clarkdale / Nehalem = 10% faster than Phenom II / FX
Sandy Bridge = 12% faster than Clarkdale / Nehalem
Ivy Bridge = 6% faster than Sandy Bridge

Therefore, Ivy Bridge = 1.1 * 1.12 * 1.06 = 1.306 the speed of Phenom II / FX = An estimated 30.6% faster than Phenom II / FX.

 

[MU_Engineer]

Two things will be certain about Piledriver vs. IB performance:

1. Piledriver CPUs will continue to offer more cores for the dollar than IB CPUs, and more cores on the desktop, period.
2. PD Performance on Linux relative to IB will be much better than PD performance relative to IB on Windows, due to better OS and especially compiler support for AMD Family 20 CPUs on Linux/GCC.

So in other words, no different from Phenom II or Bulldozer versus Intel CPUs.

Now for the guesstimation:

1a. Single-threaded performance will improve but the top-clocked PD will still lag at least 10% behind the top-clocked LGA1155 IB on Windows. Intel had quite the lead going with single-threaded performance and I doubt PD will completely close it.
1b. I'd wager a guess that the top AM3+ PD will be very close to the single-threaded performance of the top LGA1155 IB on Linux as long as you use an up-to-date GCC.
2. Heavily multithreaded performance will overall be better for PD than any IB at a similar price range. I also have a hunch that the top AM3+ PD will beat the LGA1150 i7s in multithreaded tasks too, despite likely being priced at least 10% less.

We'll see how it turns out. I would like to see #2 happen in a big way so that Intel finally relents and releases 6-core CPUs w/HyperThreading enabled (which should beat an 8-module PD without too much trouble provided they are clocked reasonably high) on desktop sockets with prices less than $400. If PD happens to beat IB in single-threaded tasks, which I highly doubt will happen on Windows, it will be interesting to see what Intel does. Resume soldering their IB heatspreaders on and then bump up clock speeds 300-500 MHz?