News Scheduler updates promise Bulldozer performance boost

Don't expect a miracle, though.

http://www.bit-tech.net/news/hardware/2012/01/12/scheduler-updates-bulldozer-boost/1

 

have you ever seen a cock up as big as this?

AMD, what were you thinking? I've said this before - just die shrink the X6 1100...

 

I could not care less :/

 

Potential reasons why die shrinking the X6 wasn't plausible was they beleived that the arch was getting to old to still improve, also the IMC was struggling a bit as it was worse than the IMC CPC for deneb. If you've seen the video at CES of the new APU using the piledriver core (updated bulldozer) they had a laptop playing Dirt 3 whilst encoding a video and playing another in HD all at the same time.

As far as a worse launch, Pentium D? Also remember intel's first HT cores weren't the best, unless the code was optimised.

However hasn't the launch of bulldozer actually been quite successful for the general computer buyer (not enthusiast)?

 

They have discontinued Phenom II units - so a new buyer has little choice for AM3

 

Will be interesting to see what kinda improvements, if any, this brings.
Am potentially looking to go Bulldozer

And yassarikhan786, why bother posting then? :/

 

I think AMD will take anything they can get at the moment. I really wish a miracle would happen and give Intel a little cause for concern.

 

Erm, so treat it more like Hyperthreading then? I'm sorry but it comes across as simply a flawed design, they went after big marketing numbers such 8 core and high Ghz.

 

I'll dig through later and compare performance, see if it won't do anything to Geekbench scores and the like while I'm at it.

To be fair; it is two cores. But two cores with only one Integer unit between the two. It only leads to failures because of the sheer amount of shared Cache too, should Piledriver fix those issues; it will do very, very well.

 

Integer calculation is more efficient than FP, but you need to optimise code (e.g. PI number solvers) to use INT and a lot of programmers will just use FP to save themselves a headache. Hence why the FX-8150 will perform anywhere between a quad core Llano (A6-3850) on FP calculations or an i5-2500K in mixed calculations.

 

Hey remember when it was a good idea to go massively parallel for a mainstream-focused processor in a marketplace where heavily threaded applications are confined largely to professional applications?

Me neither.

 

Seriously, I kept the faith AMD and just Meh. Surely at the point at which you saw you were going down the same route as Netburst you should have known this was a bad idea...

 

Incredibly curious as to how well bulldozer is performing in the consumer markets...

 

Me, too - in the paper twice a week is a flier for Fry's (computer chain store, for those not in the US). I check it every week just to see what's happening. I haven't seen any BD-based prebuilts ever. Today, though, for dirt cheap, was an Athlon II system. Athlon II! I'm starting to see some A6/A8 laptops, but almost always the prebuilts are Intel - high end is SB, low end is Pentium. Granted, could be some marketing dollar BS happening, too, but what gets put up on sale from the AMD camp is hardly exciting.

 

That's like saying you're going to remove the spare tyre from a Ford Fiesta to improve the performance so it can compete against a Ferrari.

 

Bulldozer isn't actually that slow in tests that simply use all cores, so the HT core-parking advantage is negated. It's just terrible in tests that happen to use less than 8 Threads. Ie; Most common workloads. Should Microsoft ever elect to send me the links needed to download these files; I shall install and then test.

 

Would this be the two patches in question?

KB2646060

KB2645594

Just noticed you apparently get an email with download links, which is what you're already waiting for...

But this and this may be of more assistance.