AMD Trinity A10-4600m with 7660G IGP Tested with Various RAM Configurations

In similar manner to my AMD Llano GPU testing, I have done a similar test comparing the AMD's Trinity A10-4600m with integrated GPU 7660G with various RAM configurations to test its performance based on speed, latency, and single/dual channel configuration.

I used 4 sets of RAM sticks to manage the 5 configurations. The base RAM was 2x4GB DDR3-1600 CAS11.

The configurations tested were:

(1) 2x4GB DDR3-1600 CAS11 - Samsung 1.35v
(2) 1x4GB DDR3-1600 CAS11 (single channel - on stick of above RAM) - Samsung 1.35v
(3) 2x4GB DDR3-1600 CAS9 - Kingston HyperX 2133 @ 1600
(4) 2x4GB DDR3-1333 CAS9 - G.Skill
(5) 2x2GB DDR3-1066 CAS7 - Asint



The games/benchmarks tested were:
(1) 3DMark Vantage
(2) 3DMark 11
(3) Allbenchmark Catzilla beta 23 - Cat score
(4) Crysis Medium 720p DX9
(5) Dirt 3 Medium Preset 720p
(6) Hawx 2 High Settings 720p
(7) Just Cause 2 Low/Med Setting (no AA) 720p
(8) Resident Evil 6 Low Settings 720p
(9) Skyrim Medium Settings NoAA 720p

Test system:
HP 6475b
A10-4600m with 7660G IGP
Various RAM configs as above
Intel X25-M 120GB SSD
1600x900 14" LCD
DVDRW


CPU-Z details below in spoiler:

Spoiler

Samsung 2x4GB DDR3 1600 CAS11


Samsung 1x4GB DDR3 1600 CAS11


Kingston DDR3 2133 @ 1600 CAS9


G.Skill DDR3 1333 CAS9


Asint DDR3 1066 CAS7

Details of game results below in spoiler tag:

Spoiler

Average performance difference when compared to baseline of 2x4GB DDR3-1600 CAS11




Details of performance difference for each benchmark in spoiler tag:

Spoiler

Summary:
RAM at same CAS between 1333MHz CAS9 and 1600MHz CAS9 is about 7% improvement
RAM with tighter CAS timings 9 vs 11 resulted in about 4% improvement
RAM running Single Channel vs. Dual Channel results in about 30% GPU performance degradation
RAM running 1066MHz vs 1600MHz results in about 9% improvement for each 266MHz increment.


NEXT: Results with Overclocked GPU

 

Reserved for future post.

 

Thank You for the testing. Are you able to program the RAM SPD of the HyperX? I just wonder if you can get it down to 1600 CL8-8-8-24-32 or even program it for 1.6V?

Also, do you have a memory bandwidth benchmark?

Heres to hoping that the Piledriver memory controller scales much better to get past the bottleneck that is keeping the faster RAM from getting larger gains.

 

I have not been successful programming SPD of RAM, unfortunately. But that's not to say it can't be done, as I haven't tried very hard. Plus the system probably won't run at anything but 1.5V or 1.35V anyhow.

I think there will be diminishing returns with faster RAM since there will be less benefit due to GPU speed and bandwidth, although I'd love to be able to run it at 2133MHz to see what it can do.

I can do a RAM bandwidth test, AIDA64.

 

HTWingNut I tried to find a way to PM you but I guess this forum doesn't support that.

Next week I am replacing my A8-4500m with an A10-4600m(found one online for $45, not a bad deal at all)

I am currently running CAS9 1333 ram, based off your experience is it worth it to spent the money on CAS9 1600 ram?

Also do you know if this APU will support 1866?

Thanks.

 

You need 5 posts to PM someone I believe.

And I don't think it will support DDR3 1866.

You can see that 1333 CAS9 to 1600 CAS9 resulted in about 7% performance improvement. Up to you if it's worth it or not.