Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
CPU Intel P4 840 D P4 820 D P4 630 P4 640 P4 650 P4 660 P4 670 AMD Athlon64 3500+ 3700+ 3800+ 4000+ X2-3800+ X2-4200+ X2-4400+ X2-4600+ X2-4800+ 1-Way Opteron Opteron 144 Opteron 146 Opteron 148 Opteron 150 Opteron 152 2-Way Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Opteron 248 Opteron 250 Opteron 252 2-Way Dual Core Opteron Opteron 270 Opteron 275 nVidia GF 7800GT GF 6800GT GF 6600GT ATI R X850 XT PE R X850 XT R X800 XT PE R X800 XT R X800 XL Memory Corsair Crucial Kingston Mushkin OCZ What are you shopping for?

150 MHz and Beyond

Is it possible to push the performance beyond the gain from PC-133 and CAS-2? In one word: yes.

There are a few problems with benchmarks of this kind, though. Benchmark results reflect the entire system configuration, including changes in bus and clock speed. As trivial as this may sound, it does pose some serious limitations regarding the availability of possible bus speed settings if the clock speed is to remain constant. Possible combinations of external CPU speed and multiplier settings at 1 GHz are:

Beyond 133 MHz, it is not possible to run the system within specs, at least not at the present state of technology, since the highest available bus divider (PCI divider) is currently ¼. As result, going beyond 133 MHz external CPU frequency will also increase the PCI and AGP bus frequencies which, by itself, may cause a slight increase in system performance. Further, most devices, including SDRAM are optimized for operating frequencies centering around 33 MHz or multiples thereof. Therefore, odd frequencies as, e.g. 143 MHz, often perform even worse than one might expect.

Time for some more benchmarks

Increasing the bus speed to above 150 MHz pushes the limits of even the best DIMMs currently available, at least at 2:2:2 settings. I haven’t found any commercially available modules yet that would work reliably at 154 MHz. I did have access to some experimental prototypes, though, that run at 154 MHz and 2:2:2 latencies.

SiSoft Sandra CPU and FPU memory bandwidth (MByte / sec) at 154 MHz external CPU and memory frequency (1 GHz). Scores from memory bus running at 100 MHz and 133 MHz (2:2:2 each) are carried over from the previous charts as reference points. At the 3:3:3 latency setting, scores fall behind the those achieved at 133 MHz and 2:2:2 settings. With shorter latencies (2:2:2), however, there is, once again, a significant increase in both CPU and FPU bandwidth scores. Increasing tRCD (RAS-to-CAS delay) latency to 3 cycles while maintaining CAS and precharge latencies of 2 (2:3:2) has only a minor impact on the benchmark scores. The important finding is that running at CAS-2 and 150 MHz or above, still causes a substantial increase in performance.

Memory intensive real world applications as Expendable show the same patterns as seen in the synthetic benchmarks albeit not as dramatic. The simple reason is that the memory bandwidth starts to saturate the power of the CPU. In any case, the frame rates speak for themselves.

Expendable frame rates at 154 MHz external CPU and memory speed (1 GHz clock speed).
Again, the predominant finding is that, with increased bandwidth, latency becomes an increasingly important factor for the overall system performance.

next page:    => 166 MHz, who would have thought =>