StarDestroyer.Net BBS

Nice, massive overclock reached 6GHZ in a presscott 3.6GHZ CPU

http://www.xtremesystems.org/forums/sho ... hp?t=42655



Nitrogen cooling and all, I wonder, how much could they OC the rest of the system?

Sheesh, and I thought they had actually made a commercialy viable 6 GHz CPU. Hmph.

I wonder how much a real-world performance increase that sort of OC would entail.

Yeah, I saw that on Slashdot a few days ago.

Amusingly, to get a Pentium 4 to that level they had to liquid cool it with liquid nitrogen at -105 C.

SPOOFE wrote:I wonder how much a real-world performance increase that sort of OC would entail.

It would need to include a proportional bus/memory increase to be of any value, but the P4 design takes to additional clockspeed quite well.

Yeah. And you know why that is? Because each clock cycle does less work on a p4...

What chips tend toward more work per clock?

EDIT: Work* Typodemons FFS...

Einhander Sn0m4n wrote:What chips tend toward more work per clock?

EDIT: Work* Typodemons FFS...

That's a hard to answer question, especially since you have to define what you mean by "work". The chips that do the most work per clock cycle aren't necessarily the best either, the trick is to strike a balance between clockspeed and Instruction Per Clock. The P4 actually had a great balance going for it during the Northwood days, but since 90nm has stunted clock scaling, the less-work-per-clock Athlon 64 has overtaken the P4 in performance.

With that setup I could run Doom 3 at medium quality! Woot!

Yeah. And you know why that is? Because each clock cycle does less work on a p4...

It doesn't matter HOW it's done, just how WELL.

Einhander Sn0m4n wrote:What chips tend toward more work per clock?

EDIT: Work* Typodemons FFS...

AMD systems (2.4 GHz in an Athlon 64), and PowerPC processors like those in Macs.

The Kernel wrote:

Einhander Sn0m4n wrote:What chips tend toward more work per clock?

EDIT: Work* Typodemons FFS...

That's a hard to answer question, especially since you have to define what you mean by "work". The chips that do the most work per clock cycle aren't necessarily the best either, the trick is to strike a balance between clockspeed and Instruction Per Clock. The P4 actually had a great balance going for it during the Northwood days, but since 90nm has stunted clock scaling, the less-work-per-clock Athlon 64 has overtaken the P4 in performance.

You mean more-work-per-clock .

Now if only someone would also overclock a GPU to 166%

Pu-239 wrote:Now if only someone would also overclock a GPU to 166%

Thats more limited by the memory and efficiency of the chip, in the end, the OC will give diminshing results.

I wonder how much the internal resistance went down when the cooper was cooled down to that temperature?