Home / Opinions / Memory Roundup / Buffalo Firestix FSX5000V PC4000

Buffalo Firestix FSX5000V PC4000

Price listed above is per 1x 512MB Module.

Price Per MB - 19.5p

Buffalo did rather well in our last roundup - its PC4300 CL2.5 modules proved to be a very accomplished all round product despite the rather plain looks. For this reason I was quite pleased to hear that the company had released an enthusiast range of high-end memory called “Firestix” and was quite looking forward to running them through our test rig.

Firestix come in plastic semi-rigid blister style packaging a-la-Corsair. They are decked out with flame red aluminium heat spreaders, which feature the Buffalo and Firestix logos along with the part number, module type and CL rating on a separate sticker.

Thermal tape is used between the memory chips and heat spreader and spring retaining clips are used for added security.

Rated to run at 250MHz with 3-8-4-4 timings at 2.6 to 2.8v, I’m told that Firestix rely on Hynix B memory chips which are fairly common in similarly specified modules.

SPD data wasn’t read accurately in our test system and was set to 2.5-7-3-3. At 200MHz, performance proved to be off the pace slightly, as can be expected from these timings. It could be argued that I should have forced even better timings than these at such a low frequency but it must be remembered that we’re not basing these memory roundups on enthusiast users. A mainstream user is unlikely to manually adjust memory timings so unless I see an obvious mismatch all testing is done using the pre-programmed SPD settings.

Overclocking using the SPD settings got me to a mere 220MHz which was rather a disappointment. Even 3-4-4-8 only just got me past the rated 250MHz to 258MHz, which was again rather a letdown – this is actually lower than the limit reached last time around with Buffalo’s mainstream PC3200 and PC4300 parts. Naturally I tried a range of voltages but nothing above 2.8v improved the situation and lowering it only made matters worse.

Of course no two memory chips are alike and I may have just landed myself a poor sample, something I’ll definitely investigate if I can obtain a fresh pair of modules to test. That said, performance isn’t a long way off what Hynix B equipped memory modules seem to be delivering at present, but it is usually a touch better than this. It could also be a poorly designed PCB but that would be speculation at this stage.

comments powered by Disqus