Overclocking Performance - 3D and Multimedia
After looking at the results from our benchmark suites, we were interested to see what would happen in more real-world test scenarios. We were particularly interested in multimedia performance, as that's one of the areas where increased processor speeds can be useful. We also wanted to see if CPU throttling would come into play at the higher overclock settings like it did in Folding@Home comparisons. Here are the results from Cinebench 9.5 along with several encoding tests.
Interestingly enough, nearly all of these tests continue to scale up through the highest clock speeds, although in QuickTime H.264 and Windows Media Encoder tests the difference between Bin-5 and Bin-4 is very small. We expected the single core Cinebench results to scale almost linearly with clock speed increases, but with the higher demands of two CPU cores we thought CPU throttling might become an issue on the multi-core Cinebench test. That wasn't the case, but that doesn't necessarily mean you'll always get higher performance at Bin-5 in 3D rendering applications. The multi-core Cinebench test only takes about half a minute to finish, so it could simply be that there wasn't enough time for the CPU to get warm enough that throttling was necessary. The results from the video encoding tests seem to support that theory, as the two tests that take the longest to run (Windows Media Encoder and especially QuickTime H.264) start to show smaller improvements beyond the Bin-3 setting.
After looking at the results from our benchmark suites, we were interested to see what would happen in more real-world test scenarios. We were particularly interested in multimedia performance, as that's one of the areas where increased processor speeds can be useful. We also wanted to see if CPU throttling would come into play at the higher overclock settings like it did in Folding@Home comparisons. Here are the results from Cinebench 9.5 along with several encoding tests.
Interestingly enough, nearly all of these tests continue to scale up through the highest clock speeds, although in QuickTime H.264 and Windows Media Encoder tests the difference between Bin-5 and Bin-4 is very small. We expected the single core Cinebench results to scale almost linearly with clock speed increases, but with the higher demands of two CPU cores we thought CPU throttling might become an issue on the multi-core Cinebench test. That wasn't the case, but that doesn't necessarily mean you'll always get higher performance at Bin-5 in 3D rendering applications. The multi-core Cinebench test only takes about half a minute to finish, so it could simply be that there wasn't enough time for the CPU to get warm enough that throttling was necessary. The results from the video encoding tests seem to support that theory, as the two tests that take the longest to run (Windows Media Encoder and especially QuickTime H.264) start to show smaller improvements beyond the Bin-3 setting.
Facebook
Twitter
RSS
20 Comments
View All Comments
Gary Key - Tuesday, May 8, 2007 - link
It is a typo on the chart. The numbers reflected are the total score, not the individual break out on FPS.
JarredWalton - Tuesday, May 8, 2007 - link
Fixed. SupCom is a generated score from the perftest map (with an edited benchmark script). Sorry about that.yacoub - Tuesday, May 8, 2007 - link
what?Article says:
"We weren't able to run our latest gaming benchmarks (S.T.A.L.K.E.R. and Supreme Commander) on all of the laptops, so performance results for those games won't be included here."
JarredWalton - Tuesday, May 8, 2007 - link
The XPS M1710 OC scaling charts included SupCom and STALKER results. Just not the other laptops (although I might be able to run the benchmarks on a couple laptops still).yacoub - Tuesday, May 8, 2007 - link
I gave up on waiting for laptops to reach reasonable prices. Ordered a nice c2d setup to replace my aging A64 rig and did it for under $425. CPU, RAM, and Mobo. My 7900GT is still enough for now, but when the 8800GTS 640MB hits $350 without rebates I'll probably scoop one of those up too. So still under $800 for a full system upgrade.And since I can remote in to my home machine from work and my work machine from home, I really have little need for a laptop, though I do have a company-provided laptop for travel if I really needed to use it. On that flash games (tower defense, etc) are enough to keep me entertained if I'm that desperate to sit in a hotel room(?!). Most likely an mp3 player or a book is all I need in-flight and I'll be out doing things (business or tourist related) when I'm traveling so uber high-end gaming laptops at exorbitant prices just don't really have a use for me, or I'd imagine for most folks.
Ender17 - Tuesday, May 8, 2007 - link
The graphs would be a lot easier to read if they were labeled with the actual CPU speed instead of Bin 1, Bin 2...redbone75 - Tuesday, May 8, 2007 - link
Agreed. Just as easy to put 2.33 - 3.16 as it is to do Bin-0 - Bin-5. Actually, you save a character :)JarredWalton - Tuesday, May 8, 2007 - link
Given that the clock speeds are more of a request than an actual result, I didn't want to use those. I couldn't actually see if throttling was occurring during the game benchmarks, but the scores seem to indicate that the CPU was throttling at the Bin-4 and Bin-5 results on some games.The names I used came from discussions with Dell, where they referred to the clock speeds as "Bin + 3", but I used a plus sign instead. Given that the scores are all pretty close on many benchmarks, I didn't think too much about it.
Zsuu - Saturday, February 21, 2015 - link
Good looking Dell <a href="http://www.lapt0p-hu.eoldal.hu/">notebook&...Zsuu - Monday, February 23, 2015 - link
These laptop is very good http://www.lapt0p-hu.eoldal.hu/