It's funny how there haven't been any threads closed recently. All of the sudden, a contorversial newcomer arrives, creates a bungoolio of threads in the benchmarking area, gets everybody mad, and has two closed in a week.
It would be interesting to see how these progressed if they were not killed by the mods... j/k
Anyway, moving right along , I recently managed to get some AS3 under the koolance block. Success is a word I will use, and definitely use for this one. Temps are down and the 1.53GHz CPU will go up to 1.8!
Unfortunately, one of my DIMMS is CL2.5 PC2100, and it doesn't like running above 146MHz with tight timings.. 150 with the most conservative in the world... all with 3.2v pumped into them!
Eventually I'll get to unlock it; it's an AGOGA-Y with a green substrate. That might explain the successful O/C results, as getting into windows at 1.8GHz is no prob, just arriving at the desktop. Upping DIMM voltage helps, so it's a mem limitation, not CPU-related. I'm really confident about this one
Anyway, here's a pic of the mem bandwidth @ 148 FSB... a stable speed for it... 3.2V... really tight timings. Comparing this to Tojo.. I have actually exceeded him... with "lowly" PC2100!
I think I'll bench my o/ced at lunch (10 mins ) and see where I sit for memory bandwidth vs yours (146.7 FSB and no voltage)increase.
Nice to see some fellow XP 1800+ owners doing well
Logan
EDIT - would you believe I'm not sure of the stepping or the compostion of my Palomino? Never actually saw the thing, a friend built the system. Does the stepping give them away (if so I can get it from Sandra 2002 SP-1)? What version of Sandra did you run?
Last edited by Logan[TeamX]; 11-28-2002 at 12:52 PM.
the stepping for my CPU is 6.6.2 (CPU type/model/revision)... but that won't give away the batch number. For that you'll have to look on the CPU itself. Generally if it has a green substrate then it has been manufactured after Feb. 2002 and can overclock better. For both substrates, if the batch code is either AGOGA (mine) or AGOIA, then you're in for a good oc .
NICE Dangit... now I have to go get my memory fast to beat yours... thx alot for ruining a perfectly good weekend of just playing Renagade but now I have to tweak Thaaaaaaaaaaaaaaaaaaaaaanks...
Eight is better; it is one of the features that gave the KT266A its performance advantage over other 266MHz chipsets.
(VIA's official statement, a bit dated so bear with me)
After entering the DDR chipset market with a re-spin of the KT133 chipset, VIA has taken some heat for lacking performance. Though still faster than the SDRAM chipset, the original KT266 chipset proved marginally slower than competing products from AMD and SIS. The addition of buffer depth and the consequent capability to enable burst length of 8 quad words of data has paid off with an average of 70% enhanced memory bandwidth of the KT266A over the original KT266. The increased memory bandwidth of the new revision gives the entire system a healthy boost of about 10% on a system performance level and makes the KT266A the currently fastest DDR chipset on the market.
Last edited by causticVapor; 11-29-2002 at 05:40 PM.
True, but remember, most double-sided DIMMs these days have 4 banks, so interleaving between all of them would allow for them to not be isolated on the four little quadrants of the DIMM, but instead exchange data freely. 8-way interleaving would require two DIMMs to even work. (8 banks) This cannot be done that efficiently without doubling the bus width (making it dual-channel DDR) as it has to wait through all the latencies to interleave between both modules.