X6800 @ 3.66Ghz VS AM2 FX-62 @ 3.0Ghz

[Silver Bullet]

Full THG review ... power consumpsion & 10+ pages of benchmarks

THG Tuning Test: Core 2 Extreme vs. Athlon 64 FX-62 (http://www.tomshardware.co.uk/2006/08/02/thg_tuning_test_uk/)

With the usuall results (FX gets beaten up pretty bad).

http://img192.imageshack.us/img192/4429/multi1tv9.png

[AndrewZorn]

Good read... I've had way too many people telling me to "leave overclocking out of this" etc, I guess now I know why. It turns 20% gains into 40% gains. In an AVERAGE of REAL-WORLD situations...

Conroe = too good.

[Eastcoasthandle]

is it just me or is the pc6400 c3 holding it back a bit in the OC dept when gaming???

[AndrewZorn]

didnt even notice air vs water

[Fred_Pohl]

I'm not too happy that they used the STOCK intel HSF and yet tried WC on the FX-62. They could have probably gotten to 3.8GHz on a Tuniq Tower.

Still, the results are crystal clear. Stock vs. Stock or OC vs. OC, C2D is king. But this is news to no one here.

The interesting part is when K8L ES chips show up and we get an idea of how much AMD is going to pull ahead.

I wasn't very happy with how THG used optimal OC settings for the FX62 but not for the X6800. 333FSB with asynch memory is less than ideal. 450FSB with synch memory would have been better. Also, how many overclockers use the stock HSF?

Like you said though, no real news here. It will be interesting to see how far K8L can pull ahead of K8 and how much it can close the gap to C2D. IMO K8L could be as much as 20% faster than K8 but probably less. We'll have to wait until sometime next year to find out.

[Silver Bullet]

No, K8L has some SEROIUS improvements. I'm no AMD fanboy, as many here will tell you, but there is an additional complex decoder and significantly improved SSE2/3 decoder units in K8L as well as a LOT of other architecture improvements. Clock for clock, it should be faster than C2D. But keep in mind that by the time we see K8L, C2D will have been out for almost 1 full year.


Could be a C3D out by then too .. never know i guess... I'd bet intel is really liking the lead they have now and won't be sitting back and letting AMD take it back again anytime soon .. but i guess one never knows .. a year is a long time.

[vengance_01]

No, K8L has some SEROIUS improvements. I'm no AMD fanboy, as many here will tell you, but there is an additional complex decoder and significantly improved SSE2/3 decoder units in K8L as well as a LOT of other architecture improvements. Clock for clock, it should be faster than C2D. But keep in mind that by the time we see K8L, C2D will have been out for almost 1 full year. yup solid points, but lets not forget how high Intel can scale these things. They could easy have 3.0GHZ and higher stock parts. It will be interesting next year.

[Fred_Pohl]

No, K8L has some SEROIUS improvements. I'm no AMD fanboy, as many here will tell you, but there is an additional complex decoder and significantly improved SSE2/3 decoder units in K8L as well as a LOT of other architecture improvements. Clock for clock, it should be faster than C2D. But keep in mind that by the time we see K8L, C2D will have been out for almost 1 full year.

I am aware that K8L has almost as many SERIOUS improvements over K8 as C2D has over Yonah. Attempting to quantify the real-world performance gains to be seen is pure guess work at this time. I'm sticking with my 20% max overall improvement for K8L vs K8 'guess' until proven wrong.

In any case, we have a long wait to find out.

[AndrewZorn]

yes but time and effort do not always mean results. i mean, look at how long intel was behind AMD... and im pretty sure they were *trying* to get better stuff out.

not saying K8L wont top it
especially since its 1yr newer tech
but im not going to count on it

[burningrave101]

K8L will likely have a short lived performance lead if it has one. There will be faster Conroe processors released by next summer and Penryn will released around Nov-Dec 06 most likely. Then the following year Nehalem should be showing up in the summer with a brand new architecture.

[Fred_Pohl]

yes but time and effort do not always mean results. i mean, look at how long intel was behind AMD... and im pretty sure they were *trying* to get better stuff out.

not saying K8L wont top it
especially since its 1yr newer tech
but im not going to count on it

Intel got stuck with a 7GHz processor that they couldn't ramp beyond 4Ghz using current process technologies. Tri-gate transistors, etc. didn't come soon enough to salvage Netburst.

What interests me most about AMD's future isn't the K8L architecture but rather AMD's ability to execute in a timely manner. K8 was initially supposed to hit the market in H1-2002 but was delayed until H2-2003. Even if the first quad-core Opteron K8Ls show up on schedule in mid-2007, how long before dual-core desktop versions show up? For that matter, how long before AMD can even produce 65nm cpus that clock as high as their 90nm cpus? Their current roadmap through Q3-07 doesn't look very promising.