Does it make sense to replace a C2Q Q8200 with a AMD A8-5500?

[Pedroc1999]

If you get the 5800k then yes. It has an unlocked multiplier which will let you OC it in the future to squeeze more power out of it

 

[Torn Mind]

Shawn, what is the DNA sequencing program you're using? When you tax it heavily, what happens in Task Manager?

 

[Torn Mind]

And if your not convinced look at the fx6300 and the 1090t 6core. The fx can actually outperform/match it in performance. Even thouh it has only 3 physical cores and 3 logical

Physical cores and logical cores describe chips with Hyperthreading. The logical core does not occupy any die space; it is "within" the physical core.

The "not full-coreness" of the Bulldozer chip is due to the core not having a dedicated FPU unit. The Fx-6300 chip physically has 6 integer units and 3 shared FPU units, all physically extant on the chip.

According to Anand's benches, it performs almost identically to the 1090T, but with 300 Mhz more clockspeed than the Phenom II. It's pointless to point out the hypothetical "if the FX-6300 had dedicated FPU unit" because that is not the design of Bulldozer architecture chips and hence do not exist.

 

[Shawn]

Of course the mobo went OOS, now I need to find a replacement...

 

[Torn Mind]

Won't the 5600K still consume more power, even at the same clock speed as the 5500? I was reading somewhere that the CPUs that cannot meet the 65w requirements are made into the 100w versions.

Well, it might be something about leakage or something, but the clockspeed and voltage still play big roles in CPU power consumption.

I don't know much about CPU manufacture, but I do know(much thanks to Idontcare) that they are binned and undergo validation to test how reliable they are at a certain clockspeed. Binning determines the CPU's characteristics according to certain characteristics. So, after the CPU's thermal characteristics are figured out, the manufacturer then groups them together into larger groups that get sent to retail.

Validation involves testing the CPU at a certain frequency with some special program to see when the CPU becomes too error prone and unreliable. Reliability here refers to how prone they are at making a calculation error. The A8-5600K had to undergo validation at a higher clockspeed than the A8-5500. There might be some 5600Ks that were downclocked to A8-5500 levels.

Whether or not the higher leakage chips are the ones who can pass the test at a higher clockspeed, where errors become more prone if the silicon isn't "good enough", I don't know. Maybe you can ask in the CPU and Overclocking subforum here.

 

[mfenn]

Physical cores and logical cores describe chips with Hyperthreading. The logical core does not occupy any die space; it is "within" the physical core.

Hyperthreading actually does take some extra die space, mostly in extra registers to stuff the state of the non-active thread. Granted, it's not much but it is some.

This has been argued to death. The performance of these chips are similar to their PhII quad core counterparts and they're mostly considered a quad core. Yes they have only 2 floating point units but those 2 floating point units can do as much work as 4 of PhII's floating point units hecause they are more efficient and that's why they perform like a quad core.

Intel's i3 processors perform like the old Core 2 Quads but you don't see Intel calling them quad cores. Let's call a spade a spade here. The A8 is a quad core because that's what AMD's marketing department needs it to be. They need to offer more "cores" and clock speed than Intel in order to compete in the market.

In reality, the A8 is a dual core because there are only two functional units which, on their own, would be a fully-functional CPU (the two Bulldozer modules). That's not to say it's a bad CPU for this use case, it's just not a quad core.

As to your second point, the A8 performs like a dual core in floating-point intensive benchmarks, like the video encoding that the OP was interested in. Compare for yourself in the Bench.