OCZ PC5000 EL (DFI nForce4 special)

http://www.bit-tech.net/preview/hardware/ocz_pc5000_el_dfi/1.html

 

I understand some reviews are intended to be briefer than others but over all I feel a little let down by the review despite there being some interesting information. Synthetic benchmarks (like the used Sandra & SuperPi) are a pet-hate of mine. I do see them as having some value and place but IMHO they offer almost nothing compared to real-world benchmarks such as encoding, zipping & gaming. After all that's what people are (or should be) interested in and will actually use their PCs for and that's what really counts. Unfortunately the review used just those 2 synthetic benchmarks.

Including gaming benchmarks would have ideally merited something more than a 6600GT but by far the most vital thing would be realistic res & details (ie. how people actually play their games). That's a detail most reviews neglect. For mobo, CPU and RAM reviews I understand we'll see more deviation at low res & details because it largely nullifies the gfx card, so I don't mind them being included but the meaningful settings should certainly be used in tandem.

While I'm nit-picking I would have preferred to see a WinChester or even Venice core CPU used. ClawHammer is pretty ancient now and a little less relevant but all credit to the reviewer for practically stating as much in the review.

The reason the OCZ PC3500 EB outperformed the Corsair when both were at the same speed and timings is the 'EB' (Enhanced Bandwidth) in its title. OCZ worked a little technical magic with EB technology to enhance the RAM's perf in a similar way as using lower latency increases perf. In a P4 the EB CL2.5 performed on par with a non-EB CL2.0 but then the P4 is less dependent upon latency so it isn't a perfect example.

My biggest concern is that people are going out needlessly buying expensive RAM. Esp P4-style fast yet high latency RAM which really doesn't suit the Athlon64 and simply isn't required even when o/c'ing. My understanding of the A64 archy is certainly not perfect but it doesn't have to be as it's results that count. My understanding is that due to the memory controller being integrated with the CPU unless the RAM can run at full CPU speed (eg cache) then it's always effectively running on a divider, there is no need at all for RAM to be synced to HTT (there is no FSB).

All A64 users need (inc o/c'ers) is low latency PC3200 which is now pretty common and more importantly cheap. Sure faster is better so long as it's not at the expense of latency but costs can certainly escalate for that extra 1-2% perf. I can't see how OCZ expect to charge over £200 for 1GB of PC5000 specifically targeting A64 chipsets because it's latency, not raw speed, which is key regardless of whether you o/c or not. Twice the price for P4 style memory is not a selling point for AMD users and PC5000 3-4-4 is blatantly P4 memory in my book.

Don't get me wrong, I'm just airing my views and have found almost all recent Bit-Tech reviews to be absolutely first rate (honestly, no creeping). Of course it's not like this review sucked at all, it just missed the point for me. There are a few plus points such as experimenting with higher tRAS which is commonly optimal for AMD CPUs. Another fine example is trying to keep the CPU speed the same in order to really focus on the memory and that is something often missed by reviewers. Sorry for the rant, just MHO as ever!