Another stick of Kingmax RAM arrives at OCAU. We've reviewed quite a few of their products now, covering their PC133, PC150 and PC166 SDRAM. They've certainly had a lot of success in the market lately, going from being considered a fairly generic "cheapy" brand a couple of years ago to a more respected major player now. Certainly in the Australian market they are widely accepted and definitely one of the most popular brands, no doubt partly due to their continuing low cost. However as we've seen in previous reviews, their low cost hasn't attracted a penalty in terms of performance.

This new stick is one of their first DDR SDRAM efforts. The stick is only identified as "DDR-266" which is not as far as I'm aware an official DDR designation. Normally you see the (somewhat misleading) PC1600 or PC2100 moniker which is derived from the bandwidth of the module. There's a slightly ranty explanation of the DDR numbering on the first page of my recent 2-motherboard Apollo Pro 266 comparison here. Anyhoo, DDR-266 is a simpler way of describing memory speed. It means this stick is 133MHz DDR ram, so you get kinda the equivalent of 266MHz, which is normally called PC2100. According to the spec sheet for this module it's CAS2.5, but SiSoft Sandra identifies the RAM chips as CAS2 up to 133MHz and CAS2.5 up to 143MHz (DDR or course).
Speaking of the RAM chips themselves, they are noticeably absent from the back of the PCB on this 128MB stick:

This is because, like their v1.2 PC133 and PC150 we reviewed here, Kingmax are using 16MB DRAMs. This means only 8 are needed for a 128MB stick and they can populate the other side with another 8 for a 256MB stick. Taking a closer look at the chips themselves:

Kingmax are of course using their patented TinyBGA chips with their unique mounting method, as per usual. The 7ns rating of the chips should, as predicted by Sandra, allow the chip to beetle along at 143MHz DDR and be within spec. Speaking of specs..
Overclocking
I should warn you in advance that I only have one DDR-compatible
motherboard here at the moment and the Kingmax is the only DDR
RAM.. so this won't exactly be an encyclopaedic compendium of
comparative DDR benchmarks. But, we can see how the stick goes
on this board which should give you an idea of what to expect.
This board being, of course, the Gigabyte GA-7DXR that fared so
well in the 3-way AMD760 motherboard roundup that Chainbolt
did for us a few months ago. This isn't that actual board from
the review (which is still presumably in Japan somewhere), it's
one I nabbed recently from Secretnet and I'm still getting the hang of.
Very impressed with it so far, though, apart from a couple of
niggles.
Anyway, when testing RAM to high clock speeds you need to eliminate other things as causes for crashing. The GA-7DXR doesn't allow for asynchronous memory speeds, so in order to test your RAM at, say, 160MHz, you need to be running an FSB of 160MHz. This of course is going to play havoc with your AGP card, PCI devices and hard drives if they are intolerant of their respective buses being overclocked. The GA-7DXR also doesn't let you turn off UDMA in BIOS. UDMA lets the HDD controller transfer data directly to main memory, but at high overclocks this data can become corrupted in transit, resulting in hangs and bluescreens during boot etc. Turning off UDMA in the BIOS often lets you boot higher at the expense of higher CPU load and slower transfers from your HDD's. Unfortunately, not an option on the GA-7DXR at this time. I used a Quantum Fireball +KA for testing as it has proven itself very tolerant of overclocking in the past. The GeForce SDR video card I generally use does NOT like high AGP speeds on this motherboard so it was ditched in favour of an ET6000-based PCI card. Very unexciting on the 3D front, but good for high FSB's. You might think these results are not relevant because I'm having to construct such an artificial environment to get the RAM to the speeds we're testing. This isn't true - on other chipsets, including the new SiS offerings, you can run the RAM at FSB+PCI as we're used to seeing in SDR chipsets. This would mean, at the standard FSB for a modern Athlon (133MHz), you could run your RAM at 166MHz DDR - presuming the RAM is capable of such speeds.. which is where we come in.
The CPU used was a Duron 700 that does 900MHz @ 1.85v core voltage. I ran it at 1.85v during this testing and set the multiplier back to 5x.
![]() RAID-0 Shootout ATA-100 vs ATA-133 |
![]() Apple Cooling |
![]() Lian-Li Case Window Kit |