Transcend has recently entered the overclocking ram market with its brand aXeRam
i recieved samples for testing of both the DDR800 CL4 and DDR1066 CL5 2GB kits
the DDR800 is rated at 1.8volts and the DDR1066 is rated at 2.1volts
as a comparison a crucial tracer ddr1066 kit, made by micron themselves, is rated at 2.2volts.
> aXeRam DDR800 Cl4 product page <
> aXeRam DDR1066 Cl5 product page <
the memory looks good with the silver, black and red colouring
...but more importanly what is under the hood?
hmmm...micron D9GMH on both kits...my favourite...in fact the only ic's for overclocking DDR2 in most enthusiasts opinions.
the heatspreaders didn't appear to make perfect contact with all the ic's but on the positive side at least they were very easy to remove as they were not glued down.
many other spreaders will destroy themselves and sometims the ram too when removed...so i was happy...naked micron works very well.
i will however be doing my best to convince Transcend to include the fantastic spreaders available on the DDR1200 kits on all the aXeRam range in future as i think the small extra cost will lead to better overclockability in our hands.
ok so with these ic's being similar to my team xtreem ddr1000 kit and also to the set of crucial tracers that i have here i figured this is going to be great ram.
first off i decided i'm going to make it tough on this ram and the testing motherboard by using all 4 gigabytes at the same time with the stock heatspreaders on.
the test platform is a gigabyte P35-DS3P since its also available from rectron and because P35 chipsets are popular amoung gamers and enthusiast right now.
the software used was dual 32M tests from super-pi
starting off i set 4-4-4-12 timings and stock volts on the board...which turned out to be about 1.85vdimm.
with stock volts and all 4 gigs i reached DDR840
since the 1066 kit is rated at 2.1volts and both kits appeared identical except for the stickers i then applied 2.1vdimm at 4-4-4-12 timings.
the result was stability at DDR940
hmmm...i prefer rounder numbers so i pushed 2.2vdimm, which is what the aXeRam DDR1200 kit is rated at, and hit ddr1000.
i can't say i'm that comfortable using 2.2vdimm for 24/7 use unless i know the heatspreaders are top shelf like those on the DDR1200 kit.
i decided i'm not going to drill many more volts into this ram, that i became fond of very quickly, as rumours abound of micron not always handling anything higher over the long term.
instead i rebooted and set 5-5-5-15 timings for the ram and dropped back to 2.1vdimm.
this resulted in an easy DDR1066 stable for both kits.
so as suspected both kits are perfectly stable at ddr1066 and 2.1volts.
now micron runs warm at high speed and high volts so i decided to haul out some ghetto spreaders that had been lying around since winbond BH-5 days.
using arctic silver ceramique, high surface area aluminium heatspreaders and some active cooling made me comfortable pushing further. i only had enough spreaders for 2 stick of ram so i chose the ddr800 kit for some extra torturing.
2.2vdimm made an easy DDR1200 possible, making these sticks similar to the official DDR1200 kit.
you have to love DDR800 reaching DDR1200 with such ease.
from this speed onward super-pi 1M test was used as a very basic stability tester...as a general rule 1M will run 10mhz faster than dual 32M.
staying with 2.2vdimm resulted in the 1M test completing all the way up to DDR1266...a little more volts at 2.25 achieved DDR1290 and the highest number i saw was DDR1299.
nothing at all would get me any further. i tried binning each stick to find the absolute best one, (super-pi 1M only requires one stick of ram to work at full speed) all the memory dividers available in the bios that would allow me up there and all the voltages from 2.25 to 2.45vdimm...nothing could break the DDR1300 wall unfortunately and the behaviour seen lead me to believe it wasn't the ic's holding me back but rather iether the memory pcb or a motherboard/bios limit.
the memory is built on brainpower 6 later pcb's of exactly the same model used on my team xtreem. in fact without heatspreaders on they would be indistingushable from eachother except for the actual production batch codes on the IC's themselves.
actually i went to the length of flashing the aXeRam to both team xtreem and crucial tracer spd files...unfortunately this didn't help me reach DDR1300 iether.
anyway i was extremely pleased with the cas latency 5 performance of this memory even if the cas 4 performance wasn't as great as it could have been.
it seems the particular batch of IC's on this run are excellent at cas5.
just to prove this to myself conclusively i tested both the team xtreem 1000 and crucial 1066 at 2.2vdimm cas5 and niether could reach the DDR1200 the aXeRam DDR800 could.
these aXeRam sticks will definately be my cl5 benching kit for as long as they're in my possesion as they're faster than anything i've ever layed my hands on.
Reply With Quote
Bookmarks