|
|
Benchmarks - results
All tests were run at least three times to ensure accurate results. Your should focus on comparing the SB62G2 9700pro and DFI Lanparty rows if you want to see how well the stock SB62G2 with fast display adapter compares against the normal desktop system. Overclocking results are also very interesting but I think that you already got the picture what this baby can do from that 301FSB that I was able to achieve. Here is quick rundown about the used markings:
| SB62G2 Int | Barebone system running with stock speed using the build-in display adapter |
| SB62G2 9700pro | Running with stock speeds with ATi's 9700pro display adapter |
| SB62G2 9700pro 3126MHz | 268FSB and XMS4000 memory modules |
| SB62G2 9700pro 3618MHz | 301FSB and XMS3200LL memory modules |
| DFI Lanparty | DFI's LanParty PRO875 running with stock speed, ATi's 9700pro and XMS4000 memory modules |
Numbers
3DMark03
Futuremark's more or less de'facto benchmark that you can see used in almost every review. Equipped with the latest 340 patch this is what we got; SB62G2 running with its integrated display adapter was... well, slow. You may observe the similar trend in all the benchmarks but for gamers this graph tells it all.
On th other hand, setups with 9700pro show very little difference. >1200MHz overclock makes its difference visible but it is clear that the display adapter is the limiting factor this time. Wish I had more powerfull GFX card... *wink* *wink*
Quake3
Oldie Goldie but still a valid tool in many benchmarks. Overclocking gives a nice performance boost as expected and the first setup using the integrated graphics adapter feels almost playable.
PCMark2002 PRO
Take a look at first and second setups. Because the Intel Extreme Graphics share the memory with the system you can notice a slight slow down on memory benchmark. Then take a look at the last setup and compare it with the second. Stock systems with 9700pro run neck to neck and there is no indication why anyone should favor physically bigger system over the smaller one when performance is an issue.
SiSoft Sandra - CPU MMBench
Faster clocks -> Better results. Stock systems are again very close to each others.
SiSoft Sandra - Memory
Latest Sandra versions are a great tools to test performance of memory subsystems. As you can see, while core clock speed difference is 400MHz between the overclocked systems the memory performance is better when running memory modules 1:1 with the CPU. You just have to remember that we are using 3:2 on this 301FSB system and this gives only ~200MHz for the modules so the performance isn't the greatest even with low latency settings.
Pifast
While this test isn't about graphics, the setup with Intel Extreme Graphics2 falls behind. Sharing memory brings disadvantages as we already noticed from previous tests. Brute calculation force of CPU brings clear victory to our 301FSB system. Personal best for me at this time. :)
LanParty was bit slower in every test compared to stock SB62G2 with 9700pro. This proves that these small barebones are ready for same tasks as you would handle with other desktops. Rendering, CAD, games, making of music etc. function the same as they would with those big tower systems that you may now have. Using your systems as a file server might be the only real reason why you should aim for those regular or server towers. Nothing stops you from installing couple of 320 Gigabyte drives inside the SB62G2 but then one should pay close attention to the case temperatures because new drives tend to be warmer than the old 5400rpm ones.
 | Pages: 1 2 3 4 5 6 7 8 |  |
Content in english!
Sisältö suomeksi!
