Buffalo Technologies isn't going to be the first name many of you think of when coming up with memory manufacturers that produce high-end overclocking RAM. They are however starting to really make their presence known in the enthusiast sector with high-quality memory kits at some of the most reasonable prices in town. Buffalo introduced their FireStix line back in the DDR hay-days and have carried that brand over into 2007 and the peak of DDRII performance.
These modules were provided to me by Buffalo but are a normal run-of-the-mill sample sent from their retail division. In fact, as you will see soon, they are kind of an odd set that somehow slipped through the production cracks. When I informed Buffalo of the oddity, they sent out a replacement module but I had already completed the testing with the first set of sticks so that is what I will be posting results from today. Here is the complete list of specifications for these modules.
I had trouble finding retail outlets that carried this memory and right now, Newegg.com doesn't even have this PC8000 kit listed. You can always grab Buffalo memory directly from the manufacturer and their on-line store has this kit listed at $280 right now. Sometimes their on-line store has some pretty good deals so if your looking into this kit, don't rule them out and check their price at time of your ordering. Now that we are all aware of what the subject of todays work is going to be, let's have a look at the modules and get to some overclocking.Code:* Part #FSX1000D2C-K2G * 2GB kit (2 x 1024MB) * DDR2-1000 5-5-5 @ 2.1v * 2x 240-pin DDR2 DIMMs * Non-ECC, Unbuffered * Aluminum heat spreader with thermally-conductive adhesive * Lifetime Warranty
Package:
Click for full size...
Buffalo sends out their modules in the standard blister pack with a simple cardboard insert that is universal with all of the FireStix line of memory. The model designation comes by way of a sticker on each module and a third sticker on the insert as seen on the right hand side about half way up in the first image. It is quite obvious that the oddity I was referring to earlier is the fact that one module has the newer heatspreader and one has the older heatspreader. They are pretty much the same, just a slight difference in color, logo, and sticker. So if you buy a set of FireStix modules, don't be surprised if you get mismatched heatspreaders, I assure you that they are the same sticks underneath.
Contents:
Click for full size...
The heatspreaders Buffalo uses on these modules are pretty basic and the thermal tape used is the thicker grey adhesive material that is evident in the fourth image. I would prefer to see the thin clear thermal material that Super Talent and Corsair have been using as of late that is heat cured to the heatspreader and ICs. That method would promote better heat transfer but also makes it difficult to remove the heatspreaders without causing damage and makes it impossible to re-attach them.
Either way, with the grey thermal tape used, it is quite easy to pull the heatspreaders up with a little bit of heat to reveal some lovely little D9GKX ICs below. The PCB appears to be produced by Dynamic but specifications seem to be scarce on the internet. I really would like to see Brain Power PCBs for better high voltage stabalization or some fancy 8-layer boards.
Intel Setup:
Click for full size...
The intel setup will be my standard Asus P5B-Dlx setup that involves some pretty intense cooling on the CPU by way of a Chilly1 single stage unit and a Noctua NC-U6 heatsink on the northbridge. I will also be using my DDR Maximizer as always and a 120mm fan hanging overhead to keep the heat generated by the modules and chipset away. For a few of the higher voltage runs, I did switch to air cooling by way of a Thermalright Ultra-120 as they were run at a later time so you will notice much higher CPU temps. I also had to substitute the E6300 for an E6600 at one point while my beloved 4GHz E6300 got RMA'd .
AMD Setup:
Click for full size...
I know I lied before by saying that the AM2 testing was going to be done but this will definitely be the last review as I no longer own this setup. There isn't anything special about this setup and it will be replaced by an Asus P5N32-E SLI 680i board in future reports. Here are the complete lists of hardware used for each setup:
As always, there are a couple disclaimers that one has to be aware of with these results.Code:AMD Setup: Intel Setup: MB: DFI LanParty NF590 SLI-M2R/G MB: Asus P5B-Dlx Wifi CPU: AMD64 X2 4000+ (CCBYF 0620UPMW) CPU: Intel C2D E6300 (L630A864) CPU Cooling: Thermalright Ultra-120 CPU Cooling: Chilly1 SS GPU: Biostar 7300GT GPU: Biostar 7600GS NB Cooling: Noctua NC-U6 NB Cooling: Noctua NC-U6 PSU: OCZ GameXStream 700W PSU: Silverstone Zeus 560W HD: Seagate SATAII 80GB 8MB NCQ HD: Seagate SATAII 80GB 8MB NCQ OS: Windows XP Pro OS: Windows XP Pro EXTRA: Hipro-Tech.com DDR Maximizer Ambient Temperature: 22-23C
A - SmartGuardian reports VDIMM 0.11v lower than it actually is on the AM2 results
B - What you see is what you get. I am only claiming dual32M stability and that is what I show in the screenshots. Please don't say "that isn't a stability test" or remark that I should be running something else for stability. Dual 32M is a great starting point for finding memory overclock stability and is all I intend on showing in my overclocking reports.
We can now see what these sticks are capable of, starting off with 3-4-3-8 results on the Intel setup. I choose to clock 3-4-3 instead of 3-3-3 because tRCD of 4 is faster than tRCD of 3 on this board. For those that are wondering, 3-4-3 clocks about 5-10MHz higher than 3-3-3.
3-4-3-8:
Intel:
Click link for full size screenshot...
365MHz :: 2.00v | 385MHz :: 2.10v | 405MHz :: 2.20v | 420MHz :: 2.30v | 435MHz :: 2.40v | 450MHz :: 2.50v | 460MHz :: 2.60v
3-3-3-8:
AMD:
Click link for full size screenshot...
365MHz :: 2.00v | 380MHz :: 2.10v | 400MHz :: 2.20v | 415MHz :: 2.29v | 425MHz :: 2.39v | 435MHz :: 2.50v | 445MHz :: 2.60v
4-4-4-8:
Intel:
Click link for full size screenshot...
460MHz :: 2.00v | 485MHz :: 2.10v | 515MHz :: 2.20v | 540MHz :: 2.30v | 560MHz :: 2.40v | 575MHz :: 2.50v | 585MHz :: 2.60v
AMD:
Click link for full size screenshot...
470MHz :: 2.00v | 500MHz :: 2.10v | 525MHz :: 2.20v | 540MHz :: 2.29v | 560MHz :: 2.39v | 575MHz :: 2.50v | 580MHz :: 2.60v
5-5-5-8:
Intel:
Click link for full size screenshot...
540MHz :: 2.00v | 560MHz :: 2.10v | 570MHz :: 2.20v
More CL5 funny action on my P5B-Dlx. There is nothing I could do that would get over 570MHz at 5-5-5. Sub timings and more voltage did nothing, despite CL4 running up to 585MHz. Clearly there is something with the memory controller on my board that doesn't like CL5 as this is not the first kit with symptoms like this. The modules are very stable at 570MHz as seen in this Prime / Win Memtest stability screenshot.
AMD:
Click link for full size screenshot...
515MHz :: 2.00v | 540MHz :: 2.10v | 570MHz :: 2.20v | 590MHz :: 2.29v | 600MHz :: 2.40v
That is about all the testing I have and will do with this kit. No DDR1200 4-4-4 but these modules certainly did like running 3-4-3 on the P5B-Dlx and almost reached DDR900 at 3-3-3 on the AM2 setup. Overall, this kit did just about what I expected it to do being D9GKX.
Hopefully this report shows some of those folks out there that think D9GKX is always better than D9GMH...highly binned D9GMH tend to beat most D9GKX that I have found and obviously Buffalo isn't getting the highest binned D9GKX from Micron. Perhaps my kit just wasn't great...which makes perfect sense as I have terrible luck with memory for the most part
As always, questions, comments, and concerns are certainly welcome. Thanx for looking.
Bookmarks