Can 8320 achieve almost the same overclocks as 8350?

incorrect. despite this common misconception, all evidence points to the 8320 and 8350 being binned the exact same. meaning that AMD did not take "worse" performing chip cores and put them into the 8320. Both chips routinely reach the same overclocks, both chips perform identically.

the only significant difference is their stock settings.

there is a similar rumor floating around about the fx6300 and even 6350. That they are just 83xx cpus with faulty cores and poor overclock headroom. Yet all evidence points to the fact apparently AMD didn't bin either of those chips either. the 6300 apparently reaches the same average overclocks the 6350 does... and those overclocks are pretty much identical to the clocks the 8350 and 8320 reach.

simply put, getting a 8320 you have just as likely a chance to reach a high (5.0ghz) overclock as you do with a 8350. They both will get up to somewhere around 4.4ghz on stock voltage... and both on average will clock up to 4.8ghz (average) on air.

The highest clocked 8 core i've ever personally seen was a 8320, which reached a daily use clock of 5.5ghz on a water loop (it was validated at 5.7ghz, but he didn't use it at those speeds). the highest clocked 6 core i've ever seen was a 6300 which hit a day to day clock of 5.6ghz on a corsair h80i (that chip probably could have gone higher on a custom loop, it was really insane)

Granted i've seen a lot of 8350s and 6350s only get to 4.5ghz... and i've seen a 8320 top out at 4.4ghz... so like everything with overclocking it's pure luck of the draw. all i'm saying is the deck isn't weighted toward any of those chips.

That said from the small sample size of numbers i've seen online the 9xxx supersampled amd 8 cores DO seem to reach the highest clocks on a most consistent basis. from what I've seen from overclocking validation sites, the average overclock seems to be something like 5.2ghz on air from those chips. Of course those chips are more expensive so you are paying for the binning i guess.

Evidence please!

ok. of course you're asking me to dig up websites I've only gone to a few times... i'll hunt it down for you. Just gotta make my head remember where i saw this data.

EDIT: here you go. you can search by processor and cooling method and see the average overclock attained.

http://hwbot.org/hardware/processors

fx8350 (1736 submissions)
air - 4730 (+0.3%)
water - 4961 (+0.5%)
fx8320 (1089 submissions)
air - 4683 (-1.3%)
water - 4860 (-1.5%)
fx6350 (53 submissions)
air - 4645 (-1.5%)
water - 5135 (+4.1%)
fx6300 (389 submissions)
air - 4790 (+0.2%)
water - 4774 (-3.7%)

Average clock
air - 4712
water - 4932

As you can see all chips are well within any margin of error they might be due to sample size... with the largest deviation being on the fx6300 under water at -3.7%, a number likely heavily influenced by a tiny sample size as it's lower then the results gained by air cooling. Generally speaking all the numbers are statistically insignificant in spread to draw any conclusion about binning apart from the fact that it apparently doesn't happen at all.

A conclusion backed up by the fx8350 showing up closest to the cumulative average; that chip is easily the most popular of the FX lineup, so it likely provides the largest sample size of all of the chips... since it's almost identical to the average clock speeds of all 4 chips i sampled, we can assume that piledriver chips are not binned by AMD in any significant way outside of the super-binned fx 9xxx models.

What I am meaning is that one 83xx can OC to 4.0 GHz another can OC to 4.5GHz. When they bin them which will they sell as FX8320 and which FX8350. Many will overclock the same but if you get the one that only goes 4GHz then they can sell it as FX8320 not FX8350 because of the turbo that is all. It is the same as was going on with the Athlon II X3 and Phenom II X2 some would unlock fine and some would not at all but they all met the required specs for what they were sold at.
What is missing in the statistics though is that probably 90% of sold CPU's never see OC and I know many an overclocker that has returned a chip to get a better one.

I understand your point. however statistical evidence does not back it up. Granted, only the fx8350 has anything resembling a dependable sample size from that site i cited... but considering it was so close to the un-weighted overclocking averages of those 4 chips only makes my point stronger.

Its simple statistics. granted hwbot users are a small group of people... but the group of people who bin the chips themselves for overclocking purposes is even smaller... and generally are not going to overclock with a custom water loop or air.

those types usually are looking for liquid nitrogen.

There is another angle to the selection as well and that is both CPU's are 125watt tdp and that basically means they have to pass that limit on their std clock but not OC. Same can be said about the FX9xxx series two different clocks same 220watt tdp. It would be interesting to see what happens with that number on the 8320 at 8350 clocks. (we know they can reach 220watts at least)

does 8320 come in the same metal box as 8350? (not that it could possibly cost nearly as much as the difference between them)
- sorry i just had to