Is the chart correct for Athlon II X4 651? and server CPU selection help

[dimm]

Hi,

There is a result in the chart that seems odd. For the Athlon II X4 651, the passmark cpu mark is 819, which seems kind of low for a modern quad core processor. Has AMD dropped the ball and this chip is really a lemon, or are there aren't many results for this chip and the results are being skewed?

I'm trying to put together a very cheap development server. It's going to replace one made from bits salvaged from dustbins! It'll be on relatively light duties, CentOS, webserver and mailserver, with very low loads. I don't think I really need to spend more than 70GBP/110USD on the processor, but I'm unsure of what to look for. Any general advice on processor selection would be really great.

Thanks

[David (PassMark)]

Yes, it is very rare CPU, so the result isn't all that accurate.

I wouldn't describe the Athlon II as modern processor. AMD released 3 generations of chips after the Athlon.

[David (PassMark)]

Other cheap low cost, low power, low performance options would be,
Intel i3-2120
AMD ATHLON X2 260
Atom D525 (low performance but very power efficient)
Intel Celeron G530

[dimm]

Thanks for the feedback. I know that the athlon II is old, but AMD released this particular variant less than three months ago. AMD also released the Athlon II X4 631 round about the same time as the 651, and that scores 4431 in your list, which is why I thought one of the numbers must be an anomaly.

I'm not really bothered about power efficiency, just the overall cost of the system. As the chip will be going in a server I don't know if the new breed of AMD A4's with their integrated graphics will be a good thing or not and advice on this would be great.

The CPU's I'm looking at within my budget are:
Athlon II X2 250
Athlon II 641
A6 3500

I can't tell if the 641 is actually offering really good performance or not.

[Aristarch]

Yes, it is very rare CPU, so the result isn't all that accurate.

I wouldn't describe the Athlon II as modern processor. AMD released 3 generations of chips after the Athlon.

I am also curious about this CPU. I agree with admin that Athlon II is not modern at all, but actually Athlon II X4 651 and 631 use the Llano core (32nm) instead of Propus core (45nm). So considering their CPU architecture, they more belong to AMD A-Series. You can also see this from their socket: they use FM1, while other Athlon IIs use socket AM3.
Around two weeks ago, AMD also announced two more Athlon IIs with Llano core: Athlon II X4 641 and 638. However, they are not available in stores yet and I am checking daily for any news or benchmarks. What I can tell about the 638 is that it is supposed to have a lower TDP of only 65W (others have 100W).

Concerning the chart for the 651, I think there is something fishy here.
It has a score of 819, while the 631 has a score of 4431. Now compare:
2,6 GHz (X4 631)
2,7 GHz (X4 638 )
2,8 GHz (X4 641)
3,0 GHz (X4 651)
This listing lets us expect it should achieve an even higher score...

Oh, one final remark: Although these CPUs use the Llano core, their APU (integrated graphics, namely Radeon HD) is deactivated (culled^^). It doesn't matter for me in any way, since I am going to buy a graphics card anyways. If you were aiming to rely on onboard graphics (since you mentioned the use as a server), maybe the Intel i3 2120 is a better choice for you.

If you get any news or have comments, please reply

[dimm]

On the site I'm thinking of buying from, http://www.dabs.com, they have a 641 listed with 4MB cache and the 651 listed as the black edition with 1MB cache. I assume that the differences in cache are a mistake.

[Aristarch]

I think they made a mistake, or at least were inaccurate.
The L2 Cache is 4 x 1024KB, according to AMDs website.
So it's 4MB Cache in total, or 1MB per core.
The only difference is the multiplier (not certain for the 638 though...)

[David (PassMark)]

Had a closer look at the single benchmark result we had for the x4 651, and agree it looks far too low. Most likely the machine was badly configured. So I have deleted the result. Unfortunately this means that once the charts are updated overnight, this CPU will disappear from the list until someone else runs a benchmark on it. (We don't have this CPU model in house).