Intel Core i5 Benchmarked

[wuzy]

This is solely based on 'feel' through relatively comparisons, so I have no solid data.
The next gen. top-end single GPU card e.g. G300/RV870 should not surpass the bandwidth provided by PCIe 1.0 x16/2.0 x8 (P35 is still quite capable). On the other hand 2xG300 or R800(2xRV870) on a single card will. If you're crazy enough to go for a 4-GPU setup then you'll certainly have the cash to blow on LGA1336.

[EDITED]So either way it's two GPUs max. for LGA1156 with next gen. cards.

Logically speaking if the next gen. top-end single GPU card do surpass the bandwidth provided by PCIe 1.0 x16/2.0 x8, then that would mean dual-GPU on one card would surpass bandwidth provided by PCIe 2.0 x16. Therefor even X58 without PCIe 3.0 16x would be a bottleneck to those dual-GPU, one card.

 

[BadTrip]

I got this link from Anandtech forum. It really doesn't answer these questions, but it leads me to believe that bus is very close to saturation.

http://www.tweaktown.com/articles/1472/intel_p45_vs_x48_crossfire_performance/index.html

 

[wuzy]

I got this link from Anandtech forum. It really doesn't answer these questions, but it leads me to believe that bus is very close to saturation.

http://www.tweaktown.com/articles/1472/intel_p45_vs_x48_crossfire_performance/index.html

After reading through this I'm beginning to question myself....
Bah, too much to think. I'll do more research later.

 

[BadTrip]

it is late here in Louisiana, and I have had a more than a couple of beers, I will read this in the morning, and try to formulate an opinion.

 

[JAYDEEJOHN]

Looking at your Legion link wuzy, 1 thing crosses my mind. Using 4850s instead of 4870s. Thats a 20%+ increase overall right there. Alot of that is due simply to BW , as seen in benches where the 4870 outdies the 4850 by more than clock increases alone. It all depends on game or synth used of course, and whats required to run it etc.
Now, imagine something with maybe twice the theoretical output, tho real world, possibly 70%? Im thinking itll be close, if not slightly BW compromised

 

[lashton]

Intel created the new socket to stop people buying high end motherboards, and i5 CPUs and overcloking the pants of them and getting the i7 perfomance for the price difference, but they also have maybe shot themselves in the foot, or have foolish thinking that AMD will not strike with this, there are people with AM2+ or AM3 based systems thinking about an upgrade to i5 have now been given another performance option Phenom II 945BE or 955BE, i just got mine upto 4GHZ (955BE), think about this, save yourself say around $300 for the motherboard, even the 775 owners have another option to go with to AMD also.

 

[fazers_on_stun]

2011-2012? Sandy Bridge is scheduled for 2010, after Westmere near the end of 2009.

Yep, according to the roadmap, it'll be Ivy Bridge out in 2-3 years.

According to Fuddo , Bulldozer is not being brought forward so it and AMD's 32nm SOI are apparently still scheduled for 2011. So no new architecture for AMD for another 1.5 - 2.5 years:

AMD has confirmed that it plans 32nm server parts of next generation, something we all know as Bulldozer, for 2011. There is no indication whether we are talking about early or late 2011 but this is what the company plans at this time. This confirmation come from AMD's Server division.

Bulldozer should be a brand new architecture and AMD’s chance to get the leadership crown and beat Westmere based products, that are are basically optimized Nehalems in 32nm.

Some have expected 32nm cores from AMD in second half of 2010, but it’s unlikely that this will happen. AMD looks to be doomed to 45nm until 2011 and Bulldozer.

Even AMDZone is thinking 2010 is going to be rough for AMD. Probably explains why the admin there is in a sour mood and is busy perma-banning even loyal AMD enthusiasts who dare complain about AMD not bringing new arch forward, or even advertising what they do have properly .