I am planning on buying an assembled server for a startup and was looking at options available. I was considering Opteron for value and bang for the buck. AMD Opteron 2378 is priced at Rs.15250/- in India which is roughly $300. A Xeon E5405 is priced at Rs. 12000/- [ $240]. What gives?

Two things to note:

1. The Xeon E5405 is a lower performing chip. You're comparing a 2.0 GHz quad-core (Harpertown architecture) to a 2.4 GHz quad-core (Shanghai architecture). A quick google search of comparisons shows that the Opteron 2378 is going to outperform the Xeon 5405. So you're not paying a premium, you're paying for a better chip.

2. Supply and demand is going to vary a lot by region. Note that AMD ultimately does not set retail prices. AMD has never sold directly to consumers. All AMD chips are sold through resellers/distributors/retailers, and are thus set at whatever price the retailer thinks the chip is worth to their customers. You'll note that in North America the price of the Opteron 2378 is lower than the price of a Xeon E5405. This is why even across different retailers the prices can and will be different.

Thanks for the response and explanation. Another weird pricing issue in India -> AMD Opteron 2350 is listed as $255 on AMD Pricing page is retailing at $250 in India. AMD Opteron 2378 is listed as $174 on the AMD pricelist which is retailing at $300 in India -> http://www.amd.com/us-en/Proce...l?redir=CPPR01#opteron . Why is AMD pricing a slower processor at a higher price than a faster processor? Is this a typo?

Does anyone know where can I find a list of retailers in India who sell AMD processors? I am not sure if the retailer I am talking to is an authorized AMD reseller.

Thanks!

Edited: 07/26/2009 at 12:31 PM by vinjk

Hi HaricotVert,

Regarding performance comparisons between Xeon 5405 and Opteron 2378, cpubenchmark.net gave a higher performance number to the Xeon 5405 [6,353] vs. the Opteron 2378 [5,820] -> http://www.cpubenchmark.net/multi_cpu.html

Edited: 07/26/2009 at 12:31 PM by vinjk

The Opteron 2350 is a Barcelona core and is probably no longer in active production, whereas the Shanghai Opteron 2378 and other Shanghai models are still being stamped out on wafers as we speak (and thus their retail availability is much broader). This, combined with supply and demand in your region, can account for the price difference.

And if you're going to make your performance decision based on one synthetic benchmark, that's fine - just make sure the benchmark accurately represents what your server is going to do. Using PassMark's CPU benchmark score is meaningless unless your server is an algorithmic number cruncher running equations 24/7. If your server is going to be doing much more than crunching numbers, then a synthetic test is not going to accurately represent how your system will run. If you plan on running a database, virtualizing operating systems, directing network traffic, managing file servers, etc. - how is PassMark supposed to accurately display the performance of those compared to the Xeon 5405? It simply can't, because it's just measuring a very specific aspect of the CPU.

Also - the chart you linked - the Opteron 2386 SE is ranked *below* the Opteron 2382. That doesn't really make sense, considering the CPUs are identical except the 2386 SE is 2.8 GHz (vs. 2.6 GHz for the 2382). Same thing if you look at the Phenom II 940 vs. the Phenom II 955 - the 940 scores higher. There are several other CPU inconsistencies besides those. You have to take the PassMark scores with a grain of salt - the testing suite favors some CPUs over others for inexplicable reasons.