Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
Prices: Mainboards ABIT ASUS Chaintech Shuttle Soyo Tyan CPU Intel P4 2.4C-800 P4 2.6C-800 P4 2.8C-800 P4 3.0-800 P4 3.2-800 AMD AthlonXP XP 1700+ XP 2000+ XP 2400+ XP 2500+ XP 2700+ XP 3000+ XP 3200+ Athlon64 Athlon64 3200+ Athlon64 FX-51 Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Memory Corsair Crucial Kingston Mushkin OCZ Search Prices:

LOSTCIRCUITS	SHORTCUTS: A matter of 8 cores Dual Independent Buses FBDIMMs or back to the MTH FBDIMMs, Take 2 The S5000XVN Test Configuration Memory Bandwidth Latencies Intercore and cache communication DVD Shrink, & Mainconcept H.264 Futuremarks Cinebench, POV Ray TrueSpace and Render Power Power Consumption Final Thoughts Give Us Some Feedback on this Review
Intel's V8-Demo System "Octopussy"
(Review by MS, May 31, 2007)

Memory Subsystem

Because of the FBDIMM architecture, we'll spend a bit more effort on the memory subsystem. As mentioned on the last page, conventional benchmarks such as SiSoft Sandra may not take full advantage of the FBDIMM system, on the other hand, as also mentioned earlier, the entire argumentation about FBDIMMs providing 1.5x the bandwidth of a standard memory system because of the "dual ported" nature of the bus interface is a moot point since at the end, there are still standard memory chips that are not capable of concurrent reads and writes.

Unfortunately, Cachemem does not run on Vista-64 either, therefore, we have to rely on SiSoft Sandra's latency tool for the latency measurements as well.

Buffered Memory Bandwidth

Memory Bandwidth [MB/sec], higher is better

Unbuffered Memory Bandwidth

Memory Bandwidth [MB/sec], higher is better

Interestingly, turning on "buffering and prefetching" in Sandra results in a performance hit, that is, the unbuffered benchmark results are better than those with buffering enabled. Keep in mind though that the AMB and the chipset, let alone the dual independent host bus architecture abide by slightly different rules than what we see in conventional x86 systems.

Memory Bandwidth [MB/sec], higher is better

In comparison to other systems, particularly the ccNUMA enabled Quad FX system, the FBDIMM architecture looks somewhat poor with the unbuffered (UB) results being markedly better than the buffered transfers. However, we cannot stress the point enough that we are not running SPEC, UA or ART benchmarks that, even tough synthetic as well, may yield totally different results (average of three runs each).

(BX80557E6300)

next page: => Latencies =>