AMD disses Nehalem

[Rob Williams]

With a title like "AMD hits back at Intel's Nehalem", how could I resist reading? Given that Nehalem has proved not only successful, but faster than AMD's top processors, it's a little strange to see AMD talking down about it, rather than creating a product that does more than force a pricing game. Here's a good quote from the source interview, where AMD is questioned about HyperThreading:

"Real men use real cores. We’ve got real cores across our products. Hyperthreading is basically designed to act like a core except that it only gives 10 to 15 percent performance bump for real applications workload. That’s because hyperthreading requires the core logic to maintain 2 pipelines: its normal pipeline and its hyperthreaded pipeline. A management overhead that doesn’t give you a clear throughput."

I like AMD as much as the next guy, but a quote like this is a little strange. It's been proven that HyperThreading is completely useful... our own Core i7 launch article gave results of that.

Either way, the interview is worth a read, and so is the Inq's article that discusses it, as it's pretty humorous in parts.

http://techpulse360.com/2009/04/06/...ims-over-xeon-5500-nehalem-performance-price/
http://www.theinquirer.net/inquirer/news/693/1051693/amd-hits-intel-nehalem

One interesting tidbit gained from these articles is the knowledge of the fact that a Dual-Core Nehalem actually exists... I had no idea until it was mentioned here.

http://www.intel.com/products/proces...5000+tab_specs (E5502)

 

[Kougar]

Only thing I see is AMD just proved they'll say anything as much as the next guy. They said themself it offers an increase in performance, so what is the problem with it?

Real men use real cores. We’ve got real cores across our products.

If AMD wants to go down that road, it might help if they could first win any of the benchmarks where Nehalem's Hyperthreading was disabled. Intel doesn't need HT enabled on Nehalem to wipe the floor with AMD.

Then, they use DDR3 memory which is more expensive, draws more power and has higher latency. So DDR3 is not a good choice for 2009.

How... oblivious is this person? DDR3 uses more power than FB-DIMM DDR2? I think AMD is unhappy Intel is relegating DDR3 FB-DIMMs to their Beckton platform, it used to give AMD a considerable advantage in power consumption benchmarks.

AMD reckons it will wait until 2010 when the latency has been lowered, and the price drops before going the DDR3 route.

Kinda odd, since latency doesn't gain much and prices for 6GB of DDR3-1600 are averaging $70 on the low end right now. And that's a CAS 7 6GB pack (OCZ Platinum). 6GB of ECC DDR3 runs $125 on Newegg just now.

And by design Nehalem servers draw more power than Opteron servers. Which means that you can put less of them in a data center than AMD servers

Considering the performance per watt works out to be in Intel's favor by as much as 40% depending on the test used, this isn't a bad thing at all. Less servers means less overhead and less sources of heat, which helps lower datacenter costs.

I will say INQ's article definitely made for an amusing read as well.

The only point of contention I see that's valid is that AMD's best, the Opteron 2389, is priced against Intel's X5550 Xeon. The performance numbers won't be as lopsided since AMD has a 230MHz clockspeed advantage here... but I'm still pretty sure Intel takes the win. (Anand has X5570 vs 2389, TTR did W5580 versus 2389) I'm a bit miffed Anand still hasn't updated their charts with X5550 results as promised, and TechReport stated they have no intention of doing so.

 

[Rob Williams]

Woops, I meant to respond to this long ago, but I agree completely with your thoughts on this.