Whats the point of running your FSB just say at 250mhz and have your ram at 200mhz running on a divider. Ive heard that it can reduce performance a lot if you dont do it right, but if you do the performance increases are pretty bad even because the cpu cant talk with ram at the fsb speed, its limited to the ram speed.
this is primarily only a problem on the nf2, socket A boards that this creates a bottleneck....but still with those boards, if your CPU could be 250x10 but your RAM could only do say 225 that'd still be an improvement over 200x10 regardless of a bottleneck.
Yes I'm pissed off that my RAM is giving me BS and so I'm forced to clock it back down to 200MHz instead of 250MHz which is what I had my CPU at so it was at 1:1, but 1:1 on AMD64s isn't nearly as important as 1:1 on the socket A systems.
How does that work? Yeah sure, the FSB (HTT) runs at 250mhz just say, but it cant transfer data to the ram any faster than 200mhz because the ram is running at 200mhz so the data wont go around the system any faster than 200mhz as the memory is the bottleneck.
The ondie memory controller of the AMD64's is far superior to the north bridge system used in the socket A systems.
if you went a ran a memory benchmark test right now it'd probably be 3gb/sec to 3.1gb/sec or so......you run the same setup on AMD64, lets say 200MHz ram, and regardless if the CPU is set to 200, 250, or even 300HTT, the RAM on an AMD64 will be like 6.5gb/sec memory bandwidth...that's purely because of the ondie memory controller vs the north bridge on the socket A's which handle the memory transfer and really slows the process down...this is also why that 'bottleneck' doesn't matter as much on AMD64's
So basically what I'm getting at, is lets pretend you have your socket A system and you've got it running 300x8 = 2.4GHz and your RAM and CPU both are running at 300MHz........an AMD64 system running at 200x12 will beat it out in terms of memory bandwidth........that's just because the AMD64's ondie memory controller is THAT good over the north bridge system and the AMD64 would still outperform that XP.
Trust me I used to think like you too and would challenge any AMD64 system out there with my XP-Mobile, but really, that memory bandwidth alone will pretty much wipe the XP's clean.
So as long as his CPU speed is pretty close to yours, his would actually be more powerful....run a test like Aquamark or 3dmark01 and you'll probably notice it.
This is what I got in aquamark with my XP-Mobile:
430/1150 6800GT @ DDR400 - 2.6GHz - 71.89 - 59,093 - Aquamark3
you can see the RAM is normal PC3200 speeds, and the CPU is 2.6GHz, this is using 200x13 so theres no bottleneck...now check out my AMD64
430/1150 6800GT @ DDR400 - 2GHz - 71.89 - 65,664 - Aquamark 3
Same GPU speeds, Same RAM speeds, AMD64 is running 6GHz less than the XP and it still scores 6k over it.
This is with my GPU, RAM, and CPU on the AMD64 overclocked quite a bit:
450/1205 6800GT @ DDR510 - 2.55GHz - 78.01 - 77,971 - Aquamark 3
Sooo like I said, the AMD64's are just way more optimized
yea actually with my cpu at 2.5ghz im beating a fx53 in multi media benchmarking and ive almost got it in arithmatic.
I can guarantee you a 2.5GHz XP won't beat out a 2.4GHz FX chip.....they wouldn't make that FX series and consider it the most powerful single core chip out there if an XP a mere 100MHz over could beat it out. I hope you're not using something like SiSoftware Sandra to get these results as their method of benching has been pretty flawed in recent years.