AMD Bulldozer Finally!

ChrisGTL · 12 Oct 2011 at 17:11

One thing that I'm struggling to get my head around on these BD benchmarks;

The benchmarks I have seen are using a OC BD, and benched against other AMD/Intel's - but what are the other CPU's benched at? I presume stock or an equivalent OC to the BD?

If I was to have a 2500k and a BD system with similar RAM/GPU/SSD/Monitor how would I bench them equally as possible?

I understand a 2500k @ 4.5Ghz would not be equivalent to a FX 8150 @ 4.5Ghz - because clock for clock is different. Is there a formula to work out equivalent speeds to bench 'as fair as possible'?

Martini1991 · 12 Oct 2011 at 17:12

Is Gareth going to be the only FX8150 owner

?

Matterez · 12 Oct 2011 at 17:20

Guys... Look on the bright side..

http://www.anandtech.com/bench/Product/79?vs=434

It's still better than an Intel Atom...

Admetos · 12 Oct 2011 at 17:21

djkav said:
As long as the unified FPUs in bulldozer don't bottleneck the modules in your simulation.

I'm having a bit a faith!

Did some more digging, and Visual Studio 2010 has had support for Bulldozer since service pack 1.

I think the main issue with benchmarks is there not optimised for BD. I expect when new software is released things will improve.

jaybee · 12 Oct 2011 at 17:21

Klo said:
Oh dear =/

I'll stick to my Q6600 then

Shame about Bulldozer. That power draw absolutely sucks. I think the systems I am about to build for friends are gonna be be i5 and i7 I suspect. I am glad I told them to hold on a few weeks more to see what Bulldozer would be like. Didn't cost anything to wait. Sometimes these things come along and just don't deliver, especially the hyped and anticipated stuff.

LeJosh · 12 Oct 2011 at 17:22

http://www.anandtech.com/bench/Product/98?vs=434

That's compared to my current CPU. Dear lord.

hedgey · 12 Oct 2011 at 17:23

hopefully amd cut the price of it by 80%. id pay £50 for one. no more.

edit - nope its slower than my e8400 unclocked. lol.

http://www.anandtech.com/bench/Product/55?vs=434

Wouldnt pay £1 for it.

chaotic_uk · 12 Oct 2011 at 17:25

compaired to the 1090T it seems on par lol

http://www.anandtech.com/bench/Product/146?vs=434

skotish| · 12 Oct 2011 at 17:27

Heads need to roll after this tbh lol

mmj_uk · 12 Oct 2011 at 17:28

skotish| said:
Heads need to roll after this tbh lol

Didn't a guy get fired a couple of months ago at AMD? related?

SheepBeast · 12 Oct 2011 at 17:32

hedgey said:
hopefully amd cut the price of it by 80%. id pay £50 for one. no more.

edit - nope its slower than my e8400 unclocked. lol.

http://www.anandtech.com/bench/Product/55?vs=434

Wouldnt pay £1 for it.

ummm.... no it's MUCH faster than your e8400, from the very benchmarks you linked to. you do realise a lot of them are "lower is better"?

come on people, BD is bad but it's not that bad. there's no reason to exaggerate

aln · 12 Oct 2011 at 17:37

mmj_uk said:
Didn't a guy get fired a couple of months ago at AMD? related?

They booted their CEO. They cited reasons like a lack of plan for the mobile market, but it's also possible that AMD knew how good bulldozer was months ago and simply thought this it wasn't good enough.

In reality, I'm not sure if the chip is as bad as everyone claims, but it's also not as well balanced as I'd hoped it would be.

RavenXXX2 · 12 Oct 2011 at 17:37

If this had rocked I was tempted to pick one up and skip ivybridge but the thing hardly beats out my 920 at stock.

http://www.anandtech.com/bench/Product/47?vs=434

DragonQ · 12 Oct 2011 at 17:38

It seems that the only real reason to go FX is if you're mainly doing multi-threaded stuff like 3DMax or x264 encoding, in which case you can get an FX-8120, overclock it and get better performance than an i5-2500K for a similar price. To be fair, the FX-8120 is a lot cheaper than the i7-2600K so it still has a target audience.

However, since I'm not paying for my imminent upgrade there's only three factors I'm bothered about:

1) Single-threaded performance (i7-2600K wins easily)
2) Multi-threaded performance (FX-8120 wins slightly)
3) Power consumption (i7-2600K wins easily)

I would like a guinea pig who buys one of these to test a program on it for me though, just to be sure. Any volunteers, send me a trust message please.

logi-cat · 12 Oct 2011 at 17:39

all that wait and it dissapoints.

Jokester · 12 Oct 2011 at 17:50

They really should just die shrink Thuban onto 32nm and see how it goes tbh.

I'm appalled at the die size and transistor count, surely that can't be right?

psychas · 12 Oct 2011 at 17:52

agreed and probably would cost less to make it then bulldozer ^

LeJosh · 12 Oct 2011 at 17:52

Jokester said:
They really should just die shrink Thuban onto 32nm and see how it goes tbh.

I'm appalled at the die size and transistor count, surely that can't be right?

You would think 2 billion would actually do some good, eh?

lay-z-boy · 12 Oct 2011 at 17:56

Jokester said:
They really should just die shrink Thuban onto 32nm and see how it goes tbh.

I'm appalled at the die size and transistor count, surely that can't be right?

Reminds me of the AMD K5, 4x the branch prediction logic vs the intel pentium and it was neck and neck.... What a waste of transistors.

ots3go · 12 Oct 2011 at 18:01

Whilst the Sabertooth was on TWO last week, I was genuinly debating getting it ready.

Thank god I didn't.

Don't get me wrong, I was well aware it wouldn't be amazing (was expecting just below an i5) but still I enjoy to tinker with new technology.

Alas, the doom in my mind is the energy usage, no one in the right mind is going to want this chip.

With regards to be saying it's for servers, data centres are really being pushed to be more 'green', I doubt they'll want servers powered by these.