Improved memory speed for E-450.

About a year ago I purchased an E-450 based machine with DDR-1333 memory. For some reason it was only running at 533 MHz and it ran that way for months. Late last year I updated the BIOS and loaded the optimum defaults. Upon restarting, the memory was running at the rated 667 MHz.

Now I am interested in buying a new RPG but all of the gaming tests show results for the E-450 that are the same as the E-350. This suggests the tests were run on E-450s with underclocked memory.

Is there any source of gaming test results that were done using properly configured machines as mine is now?

 

http://forum.notebookreview.com/gam...9-amd-e-350-e-450-intel-i3-2367-compared.html

 

Thanks, Meaker, but that material is known to me and it seems obvious that the E-450 is being tested with underclocked memory. What I am interested in is a test performed on a machine with 667 MHz memory clock. Have you seen any like that?

 

That is 667MHz memory clock, DDR = Double Data Rate = 667 * 2 = 1333MHz.

 

i'm sorry but your point is unclear to me. The machine was sold with DDR3-1333 memory which should have been clocked at 667MHz but for some reason virtually all such machines were running at 533 MHz from the factory. This caused almost every E-450 based system to perform about the same as an E-350 when it should have been a small but measurable amount faster.

Now that the underclocked memory problem has been rectified I am looking for real gaming tests performed on E-450 based machines with the memory running at the correct speed to help me select the best RPG for my needs. I'm not interested in how well the software runs with memory or other clocks set to lower speeds.

 

DM1z with E-450 was at 1333MHz. If you look at the benchmarks the E-450 performed ~ 15-20% faster than then E-350. 1066MHz to 1333MHz result will be ~ 5% FPS improvement. The extra 100MHz of GPU boost (600MHz vs 500MHz) will give you an extra 8-12%. That also shows in the benchmark results. See this comparison with the AMD A8 GPU 6620 comparison with varying RAM speeds and CAS timings: http://forum.notebookreview.com/gam...g-benchmarked-various-ram-configurations.html

Which laptop do you have? It could have affected just one brand of laptop.

 

There are many brands but few manufacturers. A quick online search reveals the problem applies to many makes and models using the E-450. Speculation is that this was done to reduce power consumption.

My memory does report its type as DDR3-1333. This information is probably being read from the SODIMM's ROM. Before upgrading my BIOS the Catalyst Control Center was also reporting a 533 MHz memory clock speed instead of the expected 667 MHz.

By the way, you weren't trying to run background tasks like "fraps" while trying to measure the speed of these machines, were you? Every additional task that is running will slow the machine and skew the result.

I rarely use benchmarks. They are an indirect and often inaccurate measure of something that can be measured directly. This is especially true when comparing independent GPU cores with CPU driven EU's. The EUs will generally benchmark higher than they perform in the real world. What counts more than a thousand benchmarks is how the software really runs on an actual machine. That is what I want to know before purchasing expensive and non-refundable software.

 

I've been doing this stuff for over 20 years, I'm well versed in "how to benchmark" and what affects it. If FRAPS was used in every case, then it would have an equal effect. But anyhow FRAPS has nearly zero impact on performance unless you're recording video.

There is no way to fairly compare products without a benchmark. There's no repeatability. In-game benchmarks are designed to showcase the performance of that game. So what you see in the benchmark is fairly close to the performance you should see in game. Even recording FPS from a game sequence is a benchmark. Anything that is repeatable for the sake of comparison is a benchmark. It's synthetic benchmarks like 3DMark that aren't real indicative of actual game performance. While it's a good measure of raw gaming power, it's not really indicative of how it will run in a certain game. What I usually run are in-game benchmarks that are repeatable.

In some instances, like BF3 I usually record FPS real-time over an extended period on the same map in multiplayer to give users a feel for how it responds to real in-game performance, here's examples in a few reviews I did (BF3 and Skyrim):

http://forum.notebookreview.com/sag...10-clevo-w110er-first-look-review.html#gaming
http://forum.notebookreview.com/sag...o-p150em-review-lots-680m-benchmarks.html#bf3
http://forum.notebookreview.com/gam...g-benchmarked-various-ram-configurations.html

 

Well seeing as you did confirm the memory clock was set to 1333 MHz and my experience is no where near 20 years with "this stuff" i will just have to go with your results.

Looking at your data it is clear that my machine is not fast enough to run any of the newer games so I won't buy any. It is good that you posted the informaton for millions to view before they made the mistake of buying software they couldn't run.

it's like a public service or something.