A few memory benchmarks

[DrPepper]

Today I was bored and decided to do a comparison between DDR3 800mhz and 1833mhz. This isn't really a full, thorough benchmark but it would be a waste not to post the results I got.

Core i7 - 2.66ghz
Ram - 800mhz 6-6-6-18
Everest Memory Read - 11549
CPU Queen - 22219
FPU Julia - 11206
Company of Heroes (DX10 Highest settings, Resolution 800x600)
- Average 119.2
- Maximum 249.7
- Minimum 48.5
Cinebench - 15483

-----------------

Core i7 - 2.66ghz
Ram - 1833mhz 9-9-9-24
Everest Memory Read - 17181
CPU Queen - 22272
FPU Julia - 11263
Company of Heroes
-Average 134.2
-Maximum 283.8
-Minimum 60.8
Cinebench - 15804

Conclusion
Well as you can see a 1033mhz increase does yield some benefit in the game I chose. However CPU benchmarks were hardly affected by the increase in memory which makes sense since it's a CPU benchmark. Thanks for reading

-----------------

Benchmark Number Two

1333mhz 7-7-7-21

Cinebench
OpenGL 8312
Single CPU - 4076
Multi CPU - 16332

Unigine

16xAF + 8xAA 1920 x 1080 -

DX11 - 21fps 531
DX10 - 21fps 529
DX9 - 32fps 809

Everest Ultimate v5.3

Memory Read - 14102MB/s
Memory Write - 11541MB/s
Memory Copy - 15599MB/s
Latency - 57.2ns

CPU Queen - 23017

1866mhz 9-9-9-24

Cinebench
OpenGL 8531
Single CPU - 4139
Multi CPU - 16535

Unigine

16xAF + 8xAA 1920 x 1080 -

DX11 - 21fps 525
DX10 - 21fps 530
DX9 - 32fps 810

Everest Ultimate v5.3

Memory Read - 17401MB/s
Memory Write - 14937MB/s
Memory Copy - 20198MB/s
Latency - 48.6ns

CPU Queen - 23102


Conclusion
Well as you can see there obviously is a nice increase in memory benchmarks but at my native resolution of 1920 x 1080 unigine see's no increase. The differences between the benchmarks are solely down to a margin of error, e.g. cache misses and background services.

[BraveSoul]

interesting, i am assuming game settings were identical in both tests

[DrPepper]

Originally Posted by BraveSoul

interesting, i am assuming game settings were identical in both tests

yay someone read it

Yeah settings were identical. I should do graphs for 800 1066 1333 1600 and 1833.

[BraveSoul]

yeah absolutely,,, how about this,, 800 6.6.6.18 vs 1600 12.12.12.36

[mlee49]

How about 1066, 1333, 1600,1800

[DrPepper]

Originally Posted by BraveSoul

yeah absolutely,,, how about this,, 800 6.6.6.18 vs 1600 12.12.12.36

well timing's would be

800mhz - 4-4-4-16 ideally but realistically 6-6-6-18
1066 - 5-5-5-18
1333 - 7-7-7-21
1600 - 8-8-8-24
1833 - 9-9-9-24

I'm not sure what timings would be best but I'd mess around with it first. I'd also add a few more tests.

[Mussels]

i've been saying for a while that memory doesnt make huge impacts in FPS games, and been getting ignored a lot - you show a small increase (but noticeable in benchmarks, if not in gameplay) from 800 to 1866 - 800-1066 is negligible, but you always see people advocating it on DDR2

[Lionheart]

Quiet interesting, how about 1600 vs 1866

[Mussels]

Originally Posted by CHAOS_KILLA

Quiet interesting, how about 1600 vs 1866

you could almost math it out yourself.

it'd be a fairly small difference, 5FPS at most (on maximum FPS)

[DrPepper]

At decent resolutions such as 1920 x 1080 there would be about a 5 fps difference between 1333 and 1833. I might give it a go tonight though.

[Binge]

Originally Posted by Mussels

you could almost math it out yourself.

I was just going to suggest that

[Bo$$]

i think is less about speed more on timing with the ram

[DrPepper]

Timings aren't as important anymore imo. However if your speed is 1333mhz and your timing's are 9-9-9-24 then you might as well go back to DDR2.

[DrPepper]

New benchies are up.