What Intel chip is the king of overclocking?

[highflyer]

Back in the day (ca. 1997), it was the Celeron 300A. I remember the satisfation of running at 450 MHz. Pretty lame now.

 

[Markfw]

Hard to say. I had my E6300 from 1.86 to 3.5 ghz. And many are getting the 2.66 I5 750 to 4 ghz, and the I7 920 from 2.66 to 4.3 ghz.

 

[error8]

There were some E2140 taken from 1.6 ghz to 3.2-3.4 ghz. Hard to beat those.

 

[VirtualLarry]

Originally posted by: error8
There were some E2140 taken from 1.6 ghz to 3.2-3.4 ghz. Hard to beat those.

Or the E5200, going from 2.5Ghz to 3.75Ghz, a 50% overclock.

 

[Billb2]

Currrent World records:
1.) Celeron LGA775 356 ...8,182.7 mhz
2.) Pentium 4 631 ...8,179.9 mhz

 

[MarcVenice]

How about a core i7 920, from 2,66GHz to 4GHz is pretty normal, that's a 50% overclock too. I bet Intel could loosen up a little on making sure they can reach high clocks, and focus on getting powerusage down, then they could still release higher clocked cpu's, but with a normal 95/125W tdp.

 

[daw123]

I just looked at the website Billb2 linked to and the fastest Superpi 32M time was done using a Core i7 Extreme 1025. What the hell is that chip - I've never heard of 1025 before.

It is an ES though, so maybe they screwed up the chip designation.

 

[MarcVenice]

Originally posted by: daw123
I just looked at the website Billb2 linked to and the fastest Superpi 32M time was done using a Core i7 Extreme 1025. What the hell is that chip - I've never heard of 1025 before.

It is an ES though, so maybe they screwed up the chip designation.

Gulftown. Where bloomfields can do 5,8GHz 32m stable, gulftown (32nm) blows way past them, and only stop at some 6,3GHz or something.

 

[daw123]

Originally posted by: MarcVenice

Originally posted by: daw123
I just looked at the website Billb2 linked to and the fastest Superpi 32M time was done using a Core i7 Extreme 1025. What the hell is that chip - I've never heard of 1025 before.

It is an ES though, so maybe they screwed up the chip designation.

Gulftown. Where bloomfields can do 5,8GHz 32m stable, gulftown (32nm) blows way past them, and only stop at some 6,3GHz or something.

Thanks for the info, MarcVenice.

 

[lopri]

Originally posted by: error8
There were some E2140 taken from 1.6 ghz to 3.2-3.4 ghz. Hard to beat those.

Mine! It does 3.60 GHz stable, @1.45V air-cooled. I will keep it until it dies its natural death.

 

[bryanW1995]

Originally posted by: MarcVenice
How about a core i7 920, from 2,66GHz to 4GHz is pretty normal, that's a 50% overclock too. I bet Intel could loosen up a little on making sure they can reach high clocks, and focus on getting powerusage down, then they could still release higher clocked cpu's, but with a normal 95/125W tdp.

I agree, this isn't the top % that is typical, but a D0 i7 920 looks like a winner to me.

 

[imported_Lothar]

What about the Williamette P4? Sure the chip sucked ass, but I remember some holding overclocking records for a long time.
I forgot which particular Williamette it was(I think the 1.8GHz chip?).

Also, add P4 2.4C to the list.
My 2.4C is still running at 3.42GHz after all these years.

 

[Concillian]

None of these really compare to the 300A, because the 300A wasn't really all that crippled. It was almost identical to a PII 450 when overclocked to 450. All these examples being given are processors that are gutless compared with a processor 2x the price, let alone 4x.

People forget about just how much the price discrepancy was between the celeron 300A and the Pentium II 450 you could overclock and nearly match. The PII 450 was around $750. That's a lot of money for a CPU. The celeron 300A was a value at around $200.

so 300A value is the ability to match the performance of a processor almost 4x the price.

so an e5200 would be competing against something in the $300 range... which are quad cores, so it can't compete in terms of performance. Current gen celerons are so crippled with miniscule caches that they can't compete with anything on a performance basis.

So no, there isn't anything comparable to the 300A now. There are processors that can be overclocked 50% or more, but none of them have the same value in terms of price / performance ratio that the overclocked 300A did.

 

[yh125d]

Originally posted by: Concillian
None of these really compare to the 300A, because the 300A wasn't really all that crippled. It was almost identical to a PII 450 when overclocked to 450. All these examples being given are processors that are gutless compared with a processor 2x the price, let alone 4x.

People forget about just how much the price discrepancy was between the celeron 300A and the Pentium II 450 you could overclock and nearly match. The PII 450 was around $750. That's a lot of money for a CPU. The celeron 300A was a value at $200.

so 300A value is the ability to match the performance of a processor almost 4x the price.

so an e5200 would be competing against something in the $300 range... which are quad cores, so it can't compete. Current gen celerons are so crippled with miniscule caches that they can't compete with anything on a performance basis.

So no, there isn't anything comparable to the 300A now. There are processors that can be overclocked 50% or more, but none of them have the same value in terms of price / performance ratio that the overclocked 300A did.

Uh, yeah there are. i7 920 for $200 clocked to 4.2 wiping the floor with an extreme edition 975 for 1k, because they're physically the same processor. And there are definitely more examples

 

[Concillian]

extreme edition doesn't really count, because the only reason to buy an extreme edition processor is for the unlocked multiplier and that's a feature that the 920 doesn't have. So no, it's not comparable.

 

[yh125d]

Well then the thread is pointless. The most expensive HT intel is 500ish and the cheapest is 200ish. Kinda hard to perform better than a 4x price processor when a 4x price processor with the same feature set doesn't exist

 

[happy medium]

Mine! I bet I can get over 4.0

 

[LCD123]

The e5200 is a great deal at under $70 and overclocked, it can come close to an e8600 minus 4mb cache. Can e5200 run at 333x10 which is what a stock e8600 does? How much slower performance will 2mb vs. 6mb cache have? Of course e8600s are doing over 4GHz on stock voltage and around 4.5GHz with a volt bump! But at 4x cheaper, e5200 is 50-70% the speed of e8600 when both are oced!

 

[Sunrise089]

Originally posted by: Concillian
None of these really compare to the 300A, because the 300A wasn't really all that crippled. It was almost identical to a PII 450 when overclocked to 450. All these examples being given are processors that are gutless compared with a processor 2x the price, let alone 4x.

People forget about just how much the price discrepancy was between the celeron 300A and the Pentium II 450 you could overclock and nearly match. The PII 450 was around $750. That's a lot of money for a CPU. The celeron 300A was a value at around $200.

so 300A value is the ability to match the performance of a processor almost 4x the price.

so an e5200 would be competing against something in the $300 range... which are quad cores, so it can't compete in terms of performance. Current gen celerons are so crippled with miniscule caches that they can't compete with anything on a performance basis.

So no, there isn't anything comparable to the 300A now. There are processors that can be overclocked 50% or more, but none of them have the same value in terms of price / performance ratio that the overclocked 300A did.

This is all nonsense.

The 300A was an affordable chip that could overclock to match the performance of an expensive chip, clocking up 50% along the way. THE EXACT SAME THING COULD BE SAID OF THE 1.8GHZ ATHLONS IN 2005. Same 50% overclock (almost exactly - 1.8ghz -> 2.7ghz matching the top chip), similar relative prices.

As wonderful as both of those chips were however, the lowest-end Conroe+ in any given range of chips have easily surpassed the 300A and the s939 Athlons/Opterons. As a general rule, if all you care about is relative performance increase for the money, find the largest spread in clock speed between chips with an equal amount of cache. Generally, all of the chips will max out at similar levels, so the lower the original clock speed the better, at least in pure percentage terms.

Incidentally the user I'm quoting above would have you believe a $200 CPU beating a $750 CPU is unattainable today. Look up what is available at $200 (hint - lynnfield, 2.66ghz) and compare it to what you can find at $750 (hint - bloomfield between 3.06 and 3.33ghz). An overclocked lynnfield will be very competitive indeed. Just like an overclocked Peryn was, just like an overclocked Conroe was.

I admire the 300A as much as the next guy, but Concillian is smoking something if he doesn't recognize we're in the golden age of overclocking (100% overclocks on stock coolers) right now.

 

[sgrinavi]

I've had 2 Q6600's that ran at 3.6 24 x 7 - stable at 3.8 with some extra juice - (58% o/c)

My D0 i7 is a no brainer at 4.0 with stock voltage & HT on , 4.1 with a little juice (HT still on)- (54% o/c) and 4.3 if I turn off HT (62%) .... I have not pushed it further.

 

[dbcooper1]

I remember buying Celeron 300a for $99 late in the cycle; they had less cache than the P450 but it ran at full speed rather than half in the P450. Benchmarks and real life results were very close in most cases. In a few, the overclocked Celeron actually outperformed the P450. And then there was the Abit BP6 which enabled not only an easy 50% overclock, but you could run them dual with no modification..