this should be easy to answer.

[misanthropy]

Is there an architectural difference between say an intel e6300 and a e6400 or are they just set at a different clock rate?

 

[bobsmith1492]

- different clock rate only; otherwise, exactly the same except for some identifiers that let the BIOS set the multiplier automatically.

 

[misanthropy]

Why is it that some e6300s have greater absolute overclocking potential than the e6400 has on average? What determines each processor's model number?

 

[Born2bwire]

Originally posted by: misanthropy
Why is it that some e6300s have greater absolute overclocking potential than the e6400 has on average? What determines each processor's model number?

They generally bin processors in terms of clock speed according to the tests that they pass and their current demand. It costs Intel the same amount of money to make an E6300 as it does E6400. The difference lies in the fact that the yeilds of processors that past the higher clock speed tests for the E6400 are lower than the E6300. A processor maybe able to pass the necessary tests at 1.86 GHz and qualify for the E6300, but not at 2.13 GHz. But they were all designed to go faster than that, it is just the variations that occur during manufacturing that sometimes prevent them from achieving the ideal speeds. In addition, some of the tests that they run may not be critical to the end user. So even though a chip may not pass higher speed ratings, the errors that occur may not prevent the user from running the chip at higher speeds and successfully run their programs.

I have not bothered to keep informed on how the new cores are overclocking so I do not really know by what you mean by overclocking potential. If you mean that the E6300's seem to be able to be overclocked by X% or Y MHz, then one reason is that an E6400 is clocked closer to the limit of their design as opposed to the E6300. Thus the E6300 is able to overclock by a greater differential. Or, sometimes if the demands are very high for the lower clocked chips, they sometimes will take E6400 chips and bill them as E6300.

 

[CTho9305]

So even though a chip may not pass higher speed ratings, the errors that occur may not prevent the user from running the chip at higher speeds and successfully run their programs.

..and this can be really annoying when trying to help an overclocker troubleshoot something, when the overclocker insists that their computer is "100% stable" - sure, 99.9% of things may work properly, but you can never be certain you won't run into a case that the processor executes incorrectly.

One thing to keep in mind is that nowadays power is a major concern for processor vendors, so it seems possible that you could have a chip where, say, the leakage is too high (i.e. when the transistors are off, they don't turn off well enough and still consume a lot of power) so that at a high clock speed, the chip exceeds the maximum power the manufacturer guarantees. For example, let's assume Intel says "All E6xxx processors consume at most 100 watts", and in the normal case for an E6400, 50 watts are just from transistors leaking and 50 watts are consumed when transistors switch. Now when a chip comes out of the fab and leaks 60 watts, if they run it at 2.13GHz, it would consume 110 watts. If they run it at a lower speed, the switching power goes down, so the total power might be <= 100 watts at E6300 speeds.

Now, the fun thing is that often leaky transistors are actually faster, so if you have the ability to cool the processor and aren't worried about the heat (which is the case with overclockers), you can overclock those chips really well.

 

[Born2bwire]

Originally posted by: CTho9305

So even though a chip may not pass higher speed ratings, the errors that occur may not prevent the user from running the chip at higher speeds and successfully run their programs.

..and this can be really annoying when trying to help an overclocker troubleshoot something, when the overclocker insists that their computer is "100% stable" - sure, 99.9% of things may work properly, but you can never be certain you won't run into a case that the processor executes incorrectly.

Isn't it? When I was ramping up the overclock on my current rig, I found that I could run things just fine and do the short run computational tests. But I decided to try Super PI at 30 minutes and it would crap out around 25 minutes or so. I backed it off so that I could do the full 30 minutes and the usual Prime stuff. But it made me wonder if something like that would ever come back and bite me in regards to most programs. I run a lot of computational stuff so I can't take the risk. Heh, wouldn't that be a crappy reason for why a few days of data was bad.

 

[CTho9305]

Originally posted by: Born2bwire

Originally posted by: CTho9305

So even though a chip may not pass higher speed ratings, the errors that occur may not prevent the user from running the chip at higher speeds and successfully run their programs.

..and this can be really annoying when trying to help an overclocker troubleshoot something, when the overclocker insists that their computer is "100% stable" - sure, 99.9% of things may work properly, but you can never be certain you won't run into a case that the processor executes incorrectly.

Isn't it? When I was ramping up the overclock on my current rig, I found that I could run things just fine and do the short run computational tests. But I decided to try Super PI at 30 minutes and it would crap out around 25 minutes or so. I backed it off so that I could do the full 30 minutes and the usual Prime stuff. But it made me wonder if something like that would ever come back and bite me in regards to most programs. I run a lot of computational stuff so I can't take the risk. Heh, wouldn't that be a crappy reason for why a few days of data was bad.

Here's an interesting article about overclocking from the software developer's perspective. It's frustrating when people file crash bugs about SeaMonkey or Firefox which nobody else can reproduce, and it comes out later that the reporter was overclocking "but the system is 100% stable".

 

[pinion9]

I had a "100% stable system" that was overclocked. Games played well, OpenOffice worked well, benchmarks were great.

Until I tried launching MS Office. Word/Outlook etc. always crashed when I hit a key. Backing off on the OC fixed this.