AMD Bulldozer review thread.

[3dilettante]

AMD's frame rates are all over the place.
It might have been worthwhile putting the AMD chip back to stock or even downclocking it to see if performance evened out. The sawtooth effect may have been caused by periodic overheat and throttle cycling, or AMD's dynamic frequency algorithm starts to flake out when overclocked.

 

[Albuquerque]

AMD's frame rates are all over the place.
It might have been worthwhile putting the AMD chip back to stock or even downclocking it to see if performance evened out. The sawtooth effect may have been caused by periodic overheat and throttle cycling, or AMD's dynamic frequency algorithm starts to flake out when overclocked.

This is actually what I was thinking while reading that review; it seems like the overclock would be the very first thing I'd question in the face of those results. A better indicator for them to be looking at would've been actual CPU clock speed...

Also, a stock-clocked benchmark would've helped for those folks who didn't intend to overclock by default, like my stepfather who would use the box (at stock clocks) for all his Photoshop and Premier work.

 

[denev2004]

They do share their floating point units, and they contend for bandwidth, just like HT "virtual cores" do. I'd say Bulldozer would benefit from HT-friendly software, as long as there are less active threads than cores. If they're all loaded up, there should be no difference anymore.


---

Taking this opportunity to express my surprise at the weird cache hierarchy decisions AMD has made for BD, as well as for what appear to be signifcant empty spaces on the die.

Well...We know some HT-Frendly software are real muilt-task. But some are not really...

The Cache seems strange to me too. I can't understand why they need so much L2

 

[Mendel]

AMD FX-8150 Multi-GPU Gameplay Performance Review

They were able to run deus ex without bluescreening but with very low performance, apparently
Difference being the motherboard! MSI 990FXA-GD80

 

[denev2004]

Actually is there anyone discussing about GF's 32nm?
Some people do blame to it as well as AMD said it is disappointing

 

[Mendel]

I have a Sempron installed ATM, Waiting on my new RAM as well... 16Gb of RAM for £65 = Bargain!! And It would be even less once I recoup some money from my old RAM!

Here's my HWBot submission.... Link

I'm 9th fastest in the world but when you consider every overclock above mine is on DICE or LN2 I would be willing to bet a good chunk of money that mine is the fastest 1075T on the planet as unlike the other results 4.8Ghz is my 24/7 overclock ( Well it's actually 5013Mhz now after some tweaking ) and I know of no one else that runs a 1075T at that speed 24/7.

I'm so curious as to how BD scales with the cold that multiple times I've had an FX 6 core sitting in a checkout basket!

I´ve heard you can get a BD to go very near 6GHz with good phase changers.

http://www.legitreviews.com/article/1705/1/

Simon shows of the phase change system that they have an AMD FX processor running at nearly 5.9GHz at just over 1.6 Volts at -40 degrees Celcius!

That said... a 2500k @5.5GHz should still be very competitive to it consistently and 2600k should give it a fair beating.

 

[almighty]

I´ve heard you can get a BD to go very near 6GHz with good phase changers.

http://www.legitreviews.com/article/1705/1/
.

It'll still be slower then my 1075T

Bulldozer alone at 6ghz will consume more power then my 1075 AND my phase unit...

 

[Mendel]

I know this is offtopic but please do post more of benchmark results with your rig under your hwbot account if you have time. You got me curious

 

[fellix]

AMD's Bulldozer server benchmarks are here, and they're a catastrophe

 

[Miksu]

AMD's Bulldozer server benchmarks are here, and they're a catastrophe

That post is little strange. It is based on Anandtech's review which isn't nearly as negative.

 

[hkultala]

theoretical addition and multiplication throughput with SSE code is twice compared to 3dnow in K10.

Not if they run on the same execution units.

There are 128 bits SIMD execution units in K10.

When executing 3dnow instructions, the upper halves of the units are not used.
When executing SSE instructions, the upper halves of the units are in use.

Twice the throughput, using same units, as one instruction does teice the amount of work.

 

[fellix]

AMD extended only the SIMD SSE throughput in K10. Legacy MMX/x87/3DNow! code runs at the same rate as in K8.