Associate
- Joined
- 29 Jan 2007
- Posts
- 731
I've always had a none overclocked computer for serious work where stability matters and a gaming machine to treat and overclock like crap.
Over the years, I've had some strange overclocking results that defy logic and I'm sure others have too. I was messing around today with my old gaming rig, a A64 Venice on a 754 machine, which I just upgraded to the max 2GB RAM. It's overcloking produced some strange results!
Hardware:
Abit KV8 754
Venice 3200+ 2.2Ghz (stock)
2GB (2x1GB sticks) quality RAM barely works at 200x11 (stock speed) with 3-3-3-8 1T timings. Raise it to 204 it fails, yet 1GB (2x512MB sticks) quality RAM worked flawlessly at 228x11 (2.5Ghz) at 2.5-3-3-8 1T.
Introduce a memory divider (RAM set to 166 instead of 200).....
The same 2GB sticks now work at 250x10 (2.5Ghz) with faster 2.5-3-3-8 timings and, accounting for the divider, the sticks are now operating at 208Mhz as well!
So the 2GB RAM refused to work at 204 3-3-3-8, but has no issues running 208 2.5-3-3-8 and higher when placed on a divider???
This had me scratching my head, and after reading about the Athlon64 I concluded it must be it's built in memory controller which was infamous for getting stressed with too much RAM.
What I think is happening is the higher density 2GB sticks are placing stress on the A64's memory controller. Adding a divider places an extra buffer of latency which relieves that stress, allowing the RAM to work at faster speeds and timings.
Just goes to show you that there are no hard and fast rules when overclocking sometimes, as quirks of hardware means you have to play around for a good result.
Over the years, I've had some strange overclocking results that defy logic and I'm sure others have too. I was messing around today with my old gaming rig, a A64 Venice on a 754 machine, which I just upgraded to the max 2GB RAM. It's overcloking produced some strange results!
Hardware:
Abit KV8 754
Venice 3200+ 2.2Ghz (stock)
2GB (2x1GB sticks) quality RAM barely works at 200x11 (stock speed) with 3-3-3-8 1T timings. Raise it to 204 it fails, yet 1GB (2x512MB sticks) quality RAM worked flawlessly at 228x11 (2.5Ghz) at 2.5-3-3-8 1T.
Introduce a memory divider (RAM set to 166 instead of 200).....
The same 2GB sticks now work at 250x10 (2.5Ghz) with faster 2.5-3-3-8 timings and, accounting for the divider, the sticks are now operating at 208Mhz as well!
So the 2GB RAM refused to work at 204 3-3-3-8, but has no issues running 208 2.5-3-3-8 and higher when placed on a divider???
This had me scratching my head, and after reading about the Athlon64 I concluded it must be it's built in memory controller which was infamous for getting stressed with too much RAM.
What I think is happening is the higher density 2GB sticks are placing stress on the A64's memory controller. Adding a divider places an extra buffer of latency which relieves that stress, allowing the RAM to work at faster speeds and timings.
Just goes to show you that there are no hard and fast rules when overclocking sometimes, as quirks of hardware means you have to play around for a good result.
Last edited:
