2793Quote:
Originally Posted by saaya
--
http://img262.imageshack.us/img262/5...917076fvf2.jpg
Printable View
2793Quote:
Originally Posted by saaya
--
http://img262.imageshack.us/img262/5...917076fvf2.jpg
AMD now a days belongs in the budget bins, no way they should be pricing their products close to intel, when they perform like this......the amd x2 is a cpu from 2005 and the phenoms barely beat it, thats sad, the only way for amd to stay in the game is to get in the budget bin and stop trying to make high end cpu's if they are not high end.
They must be, Dell will be using them in Optiplex systems (mass produced office PCs).Quote:
Originally Posted by AngryArtichoke
exactlyQuote:
Originally Posted by saaya
"Nonsense cpus from AMD" not a bad slogan for marketing
that's what I was thinking. patch or no patch I've seen A64 better.Quote:
Originally Posted by SparkyJJO
...and got raped by AMD in every other benchmark...lolQuote:
Originally Posted by TopherTony
Intel has always been king of SuperPI and will remain as such. Too bad that even it is some 60% faster in SuperPI, it doesn't really mean it is 60% faster overall, let alone in servers where scaling is more important.
it is sad to see more pathetic stuff coming from AMD:(
re SuperPI: I wonder what would happen if a PI calculation program were written and optimized for K10 instead, and calculated PI faster by a landslide. Would people then use this as an industry standard benchmark as well?
Pathetic? Sure intel would have done Tri-Core if they could. They couldn't. Sure the K10 is a joke, but Tri-core isn't.Quote:
Originally Posted by AAbenson
TLB patch averages about 20% slower results... on quads.
http://techreport.com/articles.x/13741/4
I mentioned this earlier in the thread, but it seems to have been overlooked. I doubt that will help on super pi, but on 3dmark it should make an impact.:up:
Again, i dont understand why they tested with the patch on, when by the time these are public.. there will be no tlb errata, or very few available that are b2.
Amazing to see the amount of dumb responses, when several people have allready pointed out that the TLB fix hasn't been disabled.
This chip will be amazing value, and intel will have nothing against it.
Owning the high end is not everything.
Why would Intel need a tri-core? It's not like their current lineup isn't cluttered enough already. Their MCM approach means they don't have to 'salvage' broken quads into tri cores, since each quad is made up of two perfectly working duals.Quote:
Originally Posted by Calmatory
AMD is only doing this for financial reasons, it's much better to get something for a tri-core than to just throw the defective quads away.
How do you know the chip will be amazing value when the official prices haven't even been released? :rolleyes:Quote:
Originally Posted by Jakko
I'm sure it'll be priced competitively against C2Ds, but it doesn't outperform them either, even in multithreading. 2 faster cores = 3 slower cores.
Because they will be cheaper than the quad cores.
And the quad cores are allready priced to compete with intel.
This means intels dual cores will be priced against amd's triple cores.
And this may be a battle intel can not possibly win, depending on what apps you buy your CPU for of course.
If they're going to be cheap; i may get one for my mediacenter/gameserver rig :)
I don't think AMD will be able to sell them as cheap as they should though.Quote:
Originally Posted by Jakko
You make it sound like Phenom tri-core is superior to a C2D, when actually its not. A Phenom 8600 is roughly equal to an E6750 in multithreaded performance, and gets slaughtered in single threaded performance. Hence, C2D is the faster processor overall.Quote:
Originally Posted by Jakko
Also, by the time it is actually released, a Phenom 8600 will be competing against faster Wolfdale C2Ds, not Conroes. AMD will have to price tri-cores below $150 to compete IMO. We will see when it launches I guess.
I'd rather have a Intel Dual Core than this.
Not only that, but the wolfdales will almost certainly use much less power and will be much better overclockers on top of offering better performance on a majority of applications. AMD will have to price these below the cheapest Wolfdale, well below.Quote:
Originally Posted by Epsilon84
are you considering the clockspeed differenceat all?? The K10 core is a generous advancement of the k8, just not as much as we'd hoped.Quote:
Originally Posted by GAR
Yeah, it'll most likely end up competing against Allendales and the Wolfdale 3M refresh, I think 3GHz+ Wolfdales would be a step above it in performance.Quote:
Originally Posted by Periander6
Sort of off topic but how do apps handle 3 cores. For example if something is optimized for 4 cores, does it use all 3 or just 2?
A design that offers a 15% IPC increase but tops out at a 25% lower clock is not an advancement. Now of course phenom offers double the cores but considering the 4X4 offers equal or better performance and came out back in 2006 it's hard to consider the Phenom at this point anything but a step backward.Quote:
Originally Posted by ryboto
http://www.xtremesystems.org/forums/...d.php?t=176857
Problem is this is not the case: the Cinebench score shows that the multiprocessor speed-up is only about 2,66, considering phenoms scale very well in multiprocessor (around 3,85 for 4x cores in Cinebench) this is clearly a step backwards. I wonder if it's the TLB-patch causing this slowdown...Quote:
Originally Posted by SparkyJJO
The K10 architecture is a step forward. The clock speeds are an issue, yes, but blame that on the manufacturing, not the architecture.Quote:
Originally Posted by Periander6