- Back to homepage
- Related articles
  Related articles

Gskill RipjawsX 4GB 17000 CL9D-4GBXMD Review

Memory by leeghoofd @ 2011-02-25

Gskill introduced their magical Flare ram kit at the same time as the release of AMD's Hexacore Thuban CPU. By using PSC ICs, better known as powerchips, achieving new ram speed heights on the AMD 890 platform. On the AMD front it has been quite calm lately, though Intel launched their new "ahum" mainstream Sandy Bridge CPU early this year. With a bucketload of available brand new ram dividers on the P67 motherboards, you can squeeze every drop of performance out this platform. When exploring the Sandy architecture we noticed a sweetspot of price/performance around the 1600mhz ram speed mark. Though for the enthousiasts every Mhz counts. Today we test the brand new 4gb 2133Mhz CL9 kit. Quite an interesting kit as it's not the high end CL 7 kit, and thus less pricey. Though could we achieve similar speeds ?

2133Mhz testing

First up, as usual is SuperPi 32M. Giving a good indication regarding bandwith. Take note that the CPU is run at 4500mhz. Only the main timings listed in the charts, were set in the bios. The rest was autodetected.

I've limited the ram voltage to 1.65 to explore tighter timings. The RipjawsX were able to run as low as CL7-9-7-24 through all of our tests. Which is not bad at all as they are rated CL9-10-9-28 at 1.5Vdimm.

Testing the more expensive Flare kit gave us a lowest latency at 2133Mhz of CL6-9-6-24 at 1.675. The latter are marked with an * in the charts.

Running the rams out of the box is barely slower than running them with pretty tight timings. This is pretty on par as what we concluded in the Sandy Bridge review. Once over 1600mhz, the timings don't matter that much anymore as the bandwith is so huge. But for a bencher a 2 sec gain means going from Top 20 to Top 10. With the Flare kit at CL6 timings we gain another second.

The Superpi 32M results didn't improve drastically with tighter timings. The AIDA memory bandwith benchmark tells the same tale. Small gains, nothing earthshattering, but you can have the +/- same performance as a kit that costs almost half as much (17000CL7 retails at +/- 150 euros)

AIDA latency measured shows a nice gain, especially for the Maxxmen benchers this can give quite a nice boost.

More synthetic results. Both the 3Dmarks benefitting from the ram tuning. Albeit the results are again not that spectacular. Put it against rams running at 1600mhz, it would be a nice boost.

Cinebench Release 10 (64bit version) is scaling too, yet again a tiny improvement going from stock to the tightened/tweaked timings.

The X264HD encoding test gains a whopping 1fps going from CL9 to CL6. Not even worthwhile some might think. For an encoder it might be worth every penny. Let's put some games to the test :

Take note that the games were run at high to very high detail, but at a limited resolution of 1280 x 1024 ( to avoid the GPU bottleneck). Scaling is visible, worthwhile for daily gamers. Sorry nope, nothing to see here...

Let's move onto the required voltages for the above settings.

All information and graphics contained in Madshrimps are sole property of the Madshrimps crew and may not be reproduced or copied in any manner without written permission from us.