Worse performance on desktop than laptop

[mrtetz]

Hey guys, I am hoping someone can help shed some light on an issue I am having. I've been planning to buy FF14 once I got a new home theater PC built, and that has finally happened. I tested on my laptop, and got a benchmark of about 2400 with laptop high settings, but on my new desktop it's worse. The same settings get about 1750. I'm so confused because the desktop SHOULD be better.

Here are the specs, and I have the saved benchmarks as well. What I am assuming is it could be a memory or hard drive issue but I am really not sure how to best diagnose that. Any helps or tips would be great, I don't plan on playing a ton of games, nor do I expect to be able to max out FF14 on this, that wasn't the intent for the build, but I did expect it to perform better than a 2 year old laptop.

Laptop:
Windows 7 Home Premium 64-bit (6.1, Build 7601) Service Pack 1 (7601.win7sp1_gdr.130828-1532)
AMD A8-3520M APU with Radeon(tm) HD Graphics
5609.906MB
AMD Radeon HD 6620G(VRAM 3046 MB) 8.17.0010.1191

Score:2434

Screen Size: 1280x720
Screen Mode: Full Screen
Graphics Presets: High (Laptop)


Desktop:
Windows 7 Ultimate 64-bit (6.1, Build 7601) Service Pack 1 (7601.win7sp1_gdr.130801-1533)
AMD A8-6600K APU with Radeon(tm) HD Graphics
7374.984MB
AMD Radeon HD 8570D(VRAM 4095 MB) 8.17.0010.1247

Score:1736

Screen Size: 1280x720
Screen Mode: Full Screen
Graphics Presets: High (Laptop)

[khyler]

You do know benchmark test scores are based on the slowest component in your system. Also, I would suspect that each operating system has its own benchmark table. Thus, Ultimate 64 has different benchmark than Home Premium. And I wouldn't be surprised to discover that benchmarks for laptops have a different table than desktops.

[Raist]

hmm... that is interesting. The 6620 has slower clocks, but more pipes (400 vs. 256 on the 8570)--might be just enough extra width to give it a slight edge on fill rate somehow. Would be interesting to get some numbers on the actual extended bandwidth and such on them, but that's probably hard to nail down since they are both using shared system memory. Might be something else in the APU's and motherboards that is coming into play as well. On-die graphics can be weird that way sometimes.
http://www.notebookcheck.net/AMD-Rad...G.54675.0.html
http://www.notebookcheck.net/AMD-Rad...D.96587.0.html

Some motherboards will have a feature for throttling the GPU clocks automatically when needed, sort of like the Intel Speed Step options. My Gigabyte board calls it Robust Graphics Booster, or R.G.B., and has Auto, Fast, Turbo modes. Might be some other hidden tweaks in the PC's BIOS like that might give it a little more punch. Another common feature is one for tweaking the DRAM CAS/RAS memory timing to a performance profile. If you can do this, it might make a nice improvement too since you are using shared memory (might show up with a reference to JDEC mode/profile, if not it might need to be manually over-ridden to match the timings listed on the memory sticks or data sheets).

Basically, your PC might be set to the slower, safer, widest compatibility settings by default and not taking full advantage of your hardware capabilities--especially if you got some performance grade components.

{Edit:}
Actually found them listed in the Passmark database... and they are surprisingly close in scores--570 for 6620G, 617 for 8570D. May not have been many samples on them though. Just seemed odd that they were that close.
http://www.videocardbenchmark.net/vi...HD+6620G&id=40
http://www.videocardbenchmark.net/vi...+8570D&id=2553

What's really unsettling is that much older GPU solutions for laptops scored higher than that 8570D APU. The old Mobility 4870 (with much older CPU architecture... talking around 2009 hardware) scored 729, enough to put it i n the High End chart.

[mrtetz]

Raist, thanks for the feedback and ideas, I really appreciate it. I wasn't expecting anything earth shattering from the A8 in my HTPC, but I was hoping for at least equal to the performance from the laptop.

I checked the BIOS and you were right about some configuration changes. Apparently my RAM was being detected at 1333 so I fixed that and also tweaked a few other settings. It's now getting almost the same score as the PC.

I guess all in all I can't complain given that the entire unit cost $350. Thanks again for the help on getting this squared away! Now to finally get this game installed.