In Tom's review of the x2 6000 they list the memory for the review at DDR2 800, but in the cpu charts at DDr2 750.

Although the x2 5600 and x2 6000 are both Windsor 90nm with the same cache, the chart shows

the x2 5600 outperforming the x2 6000 on the Serious Sam benchmark.

The only thing I can see is the x2 5600 is running at 800 mhz mem speed, while they have the x2 6000 at 750 mhz.

In the review article of the 6000, it outperforms the 5600 on the call of duty bench, and in the charts too.

Anyone understand why not with Serious Sam?

Does 750 Mhz mem speed account for it?
----------------

links:

review: http://www.tomshardware.com/2007/0 [...] page5.html

charts: http://www23.tomshardware.com/cpu. [...] &chart=168

Thanks, Jack, great explanation.

------------------------------------------
interesting small note on a related topic

Reading at Anand on the x2 6000, I did an average of their four game benches (with Oblivion at 1600x1200 for example as one), for the x2 6000 and the e6600.

The e6600 averages 4% better at those 4 games than the x2 6000, which I thought is an interesting number to remember, as occasionally someone will want to know (since overclockers usually buy lower end chips).

We'd need to see the 6000 price very near the e6600 price of course for that to be a decent choice for a non-overclocker.

maximum pc wrote this month: "the x2 6000+ is simply outclassed at every turn by the core 2 E6700" t

Since they did not figure in overclocking and temperature which really makes the 60000+++ look bad i would skip this chip even if you have an existing mobo for it!


xbit wrote: "After the launch of Intel Core 2 Duo processor family, AMD K8 micro-architecture turned pretty obsolete. " i think xbit is flaming?

note: this is not to flame the amd users - i thought this info would be important for those who might be considering an upgrade!

You should probably stop posting. Anywhere.

AJfink, you are my new hero of the day...
for you!

It's not that it doesn't divide it correctly. It's that HT is based on a 200MHz clock which worked perfectly with DDR but with DDR2 you have 266, 333, 400MHz clocks. That means that for DDR2 800 to be below or at 400MHz when doing CPU/div # you only get the whole number at 2800MHZ and 3200MHz.

For 667 it's even wierder. It's 3300MHz, 2967MHz, 2633MHz, etc. When HT goes up to 2000MHz that should make a 400MHz base which will at least solve the problem with DDR2 800.

Hopefully the split power planes will let the RAM run ASYNC again so the rated speed will always be used.

It's not that it doesn't divide it correctly. It's that HT is based on a 200MHz clock which worked perfectly with DDR but with DDR2 you have 266, 333, 400MHz clocks. That means that for DDR2 800 to be below or at 400MHz when doing CPU/div # you only get the whole number at 2800MHZ and 3200MHz.

For 667 it's even wierder. It's 3300MHz, 2967MHz, 2633MHz, etc. When HT goes up to 2000MHz that should make a 400MHz base which will at least solve the problem with DDR2 800.

Hopefully the split power planes will let the RAM run ASYNC again so the rated speed will always be used.

Indeed, K10 should fix the memory dividing "issue." If I had an AMD processor that didn't run the RAM at the full 800Mhz I'd just OC the chip until it hit it.

But that's me.

Sounds sensible. By the time I upgrade from my current setup, I'll probably not worry about that, since I'll likely still be doing what I am doing now, running at base speed with QuietN'Cool in a silent computer, which is pleasant much of the year in a warm climate also.