Overclocking
For testing the overclocking limits of this memory, I used my trusty Abit KA7 (reviewed here), as it is capable of running RAM at higher speeds than any of the other testbeds here. Even so, modern high-performance RAM is getting more and more impressive, so even the limits of this board are reached occasionally. For that reason, I have included 2 different levels of speed, the highest speed each stick can run a Quake3 timedemo at, and also the highest speed it can run the Sandra memory benchmark at. This should eliminate the AGP port as a possible reason for failing to run the Q3 demo. I benchmarked the v1.0 64MB stick, the v1.1 128MB stick, the v1.2 128MB stick of PC133 and the v1.2 128MB stick of PC150. For reference, I also benchmarked the Corsair PC133 CAS2 memory again (reviewed separately here).

As in the Corsair review, I found that performance between different types of RAM at the same settings was almost identical. This makes sense, that it's more dependent on the motherboard/chipset implementation than the particular stick of RAM. The important difference lies in how fast the various sticks can go before producing errors.

I used the fastest RAM timings available on the KA7 - "Turbo" mode and 4X interleave. To make sure the CPU or HDD's weren't causing any problems, I set the L2 cache on the Athlon to 1/3 ratio, the multiplier to x5, disabled UDMA transfers on the HDD and boosted the voltage to 1.85v. Anyway, graph time:

This big graph shows a few things. Firstly, the KingMax v1.0 stick was not stable enough to run the benchmarks at all. Secondly, the v1.1 stick was only marginally stable. I could run the benchmarks, but I got occasional reboots and freezes. I'm sure on a motherboard it was happier with, the v1.1 stick would clock similarly to the v1.2 one. The PC150 shows itself to be capable of running faster than the v1.2 PC133, but only marginally so. At CAS2 the difference is more pronounced than at CAS3 - still, the 150MHz barrier is one that overclockers are hitting pretty commonly these days and it might be worth the extra bucks to be sure your RAM can get you there. The Corsair spoils the party by clocking up to the limit of the KA7 (and therefore, may possibly be capable of higher results than this), but remember there is a significant price difference between KingMax's offering and Corsair's. You do get what you pay for, but it's a personal decision as to whether the extra bucks are justified in your case.

Conclusions
I admit, I approached this review with some scepticism. Having been frustrated by compatability problems with KingMax before, I was prepared for the worst. However, it really seems that KingMax, with their v1.2 PCB, have sorted out any compatability problems they were having. Based on my experience with the v1.2 PC133 and PC150, I'd have no problem recommending either.

Price wise, the KingMax also fares well. Eyo list the PC133 as $291 inc GST, the PC150 $355 inc GST. This is a pretty hefty price jump for the performance difference I've seen. Compare this to Realtime's price of $455 for the Corsair CAS2 PC133 and, by the graph above, the KingMax looks like something of a bargain. There's no denying the Corsair will go faster, and it's arguably higher quality, being intended more for the server market than home users. Still, as I said before, it's a personal decision as to whether the extra bucks are worth it to you or not.

When all is said and done, the KingMax left me pleasantly surprised. Good value, good performance, lifetime warranty. Recommended.