Intel Nehalem Preview/Benches - Slaughters Core2

[MR.B]

Anandtech has a preview on Nehalem with benches.

They have a Core 2 and a Nehalem, both at 2.66GHz. The performance increase seen with Nehalem is similar to that seen with Core 2 over Pentium 4. So it's big.

Keep in mind the benches were done on a very early board, that had it's problems:

We had access to a 2.66GHz Nehalem for the longest time, unfortunately the motherboard it was paired with had some serious issues with memory performance. Not only was there no difference between single and triple channel memory configurations, memory latency was high. We know this was a board specific issue since our second Nehalem platform didn't exhibit any issues. Unfortunately we didn't have access to the more mature platform for very long at all

..Yet the memory performance (and everything else) is still vastly better than Core 2!

 

[MR.B]

I'm calling BS on these... no way Intel can pull off this twice.

Err AnandTech isn't exactly the Inquirer or FUDzilla. They played with these CPU's and tested them thereself. I have no reason to doubt there results.
They were one of the first to test the Core 2 exactly two years ago at the same show, and those results were right aswell. You're just annoyed it kills your new Core 2 system, so you say crap like that. It's really pathetic and immature.

Not that impressive to be honest.

If you look at the graphs they imply a large difference but if you look at the figures.. 3 seconds better? Not really ground shattering is it?

Seems your've just hunted for the worst increase there and mentioned it. Nearly all the results are vastly better, even the memory access speed and bandwidth - on a very early board with poor/broken memory performance.


---------------------------------- Memory Read --- Memory Write --- Memory Copy --- Memory Latency
Nehalem (2.93GHz) --------------- 13.1 GB/s ------- 12.7 GB/s ------- 12.0 GB/s ------- 46.9 ns
Core 2 Extreme QX9650 (3.00GHz) - 7.6 GB/s ------- 7.1 GB/s -------- 6.9 GB/s -------- 66.7 ns

 

[MR.B]

The question is how well this is going to translate into real world performance.

I am pretty sure its going to slaughter everything in synthetics. Real world performance is another case. Clock for clock it wont be that much of an improvement, and Core 2 was made to cope with the memory latency, why do you think Yorkfields have 12MBs of L2 Cache?

There were P4's with a ton of cache, but Core 2 still completely outperformed them in real-world tests.

And did you read the article? Because most of the tests are not synthetic.
They test: Xmpeg Encoding, 3D Studio Max, AutoMKV, POV-Ray.... the main thing they lack for now is game tests.

 

[MR.B]

As my previous response above but I didn`t get an answer to my question - Will crysis run on nehelam? or is there possibly a incompatibility with this game on this processor, and will need a patch released to sort it?

Thank you for proper response.

Of course it will run on Nehalem. They must have meant it still wouldn't run smoothly on high settings or whatever, or were joking.

But i'm sure it will be playable at 1920x1200 with a GTX 280 + a Nehalem. I mean, i already get playable frame rates on my system @ 1650x1080 on Very High.

 

[MR.B]

Come on... this is a CPU, and everything i've heard about it is that Intel are very happy even with the egineering samples, the chances of it not running Crysis are so stupidly remote that you may aswell say it's going to work. Because if it has trouble with a game, that uses the same programming langange as many other things, then it will more than likely have trouble with a ton of other software/games/OS's... and in that case, Intel would not release it.

 

[MR.B]

Some other quick performance gains in percentage from the article:

Nehalem is just over 40% faster than Penryn, clock for clock, in 3dsmax.

Encoding performance here went through the roof with Nehalem: a clock for clock boost of 44%. (with AutoMKV encoder)

Memory bandwidth is obviously greater with three DDR3 channels, Everest measured around a 70% increase in read bandwidth

While the 2.66GHz Core 2 Quad Q9450 (Penryn) takes another 127 seconds to execute twice the workload, the 2.66GHz Nehalem only needs another 49 seconds. And if you're curious, this quad-core Nehalem running at 2.66GHz is within 20% of the performance of an eight-core 3.2GHz Skulltrail system. Equalize clock speed and we'd bet that a quad-core Nehalem would be the same speed as an 8-core Skulltrail here.

In our testing we found that Phenom's L3 cache takes a similar 43 cycles to access but at much lower clock speeds (2.0GHz). If we put these numbers into relative terms it takes 21.5 ns to get a request back from Phenom's L3 vs. 14.6 ns with Nehalem's - that's nearly 50% longer in Phenom.

Finally POV-Ray echoes what we've seen elsewhere, with a 36% performance improvement over the 2.66GHz Core 2 Q9450

And lastly...

Not that impressive to be honest.

blah blah blah ....Not really ground shattering is it?

LOL

 

[MR.B]

Luckily for us Intel seem to be trying to stick to moores law so it doesn't affect CPU progress much but they are dodgy as ****. Loosing monopoly lawsuits all over the place.

I hope nVidia and VIA kick their ass but it is wishful thinking, the way it is going intel will soon be making almost the entire computer.

CPU/GPU/Mobo with wifi, sound, ethernet etc, and soon SSHDs. What the **** is left other than RAM!

Well the FTC are now ivestigating Intel.

And also lately Intel were fined $25M in a South Korean antitrust case.

The AMD case against Intel has been postponed until 2010 though.

I dont really like Intel, but then pretty much all massive companies suck anus really... but i do love there CPU's. Will definitely get a Nehalem.
As for AMD i dont feel that sorry for them, the people who run the company have made some really stupid decisions lately. Top employees are leaving them like a sinking ship.