Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
Prices: 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:

** using 2 x 1024 MB (1GB DIMMs), SiSoft Sandra did not return any benchmark scores. A single 1G DIMM worked without problems but gave marginally lower scores than a comparable 256 or 512 MB DIMM. All other benchmarks we tried worked flawlessly with 2 x 1 GB DIMMs. We need to mention here that the 1024 MB DIMMs are built using x4 chips that are officially not supported by the chipset. However, the modules are assembled by tying together the address and command pins of the two chips and combining the I/O pins (4 bit + 4 bit) to an 8 bit width seen by the controller. The only chipset we have seen so far that had difficulties with this configuration is the nForce2 chipset. The problem with SiSoft Sandra may simply relate to being confronted with a single 2 GB "ÜberDIMM" which is beyond the current PC convention (limited to 1 GB/DIMM). Since Sandra has to assess the density of the DIMM in order to calculate the appropriate memory space for benchmarking (1/2 of the total system memory) the return of a value beyond what it knows will look wrong and the benchmark will terminate.

Performance

SiSoft Sandra

Of course we were curious how the 128bit mmeory interface affects the overall memory bandwidth available to the CPU. Moreover, since we were boldly claiming on the previous page that there was no support for CAS 1.5 on the DRAM level, we wanted to put proof to the pudding by showing appropriate benchmark results.

SiSoft Sandra memory benchmarks (MB/sec), (buffering enabled) for CAS 1.5, 2 and 2.5 in either single channel (S) or dual channel (D) mode. We know that in DDR the impact of CAS latency is somewhat negligible but we don't believe that there is no difference between CAS-1.5 and CAS-2.5. Still Streaming benchmarks can hide latencies extremely well which is why we turned off prefetching for the next set of benchmark.

SiSoft Sandra memory benchmarks (MB/sec), (buffering disabled) for CAS 1.5, 2 and 2.5 in either single channel (S) or dual channel (D) mode. If prefetching is turned off, we should see some impact of CAS latency on the overall bandwidth, however, the benchmark results are identical for all settings. Changing tRCD from 2 to 3 should show the differences but once again, results were identical for all settings which leaves us with the only possible conclusion that the BIOS settings are, once again, eye candy only and without functionality.

next page:    => ContentCreation Winstone2002 =>