GSKILL FLARE - powerful amd performance kit

[youngpro]

hey guys

Here is a little info on the new gskill flare kit, in the past i have always been very anti-psc, i really dont enjoy trying to clock these kits....

On the AMD thuban platform these sticks are actually very impressive, good clocks, good performance product and best of all, home users can achieve these clocks on air, you don't get much gain from cpu and ram both on ln2 than you do with all air, volts don't really scale at all either...

the result? a perfect kit for 1050mhz 24/7 clocks on your well overclocked thuban cpu...

what i always love about getting back onto the amd platform is how logically things work, and thuban memory clocking is no exception to that. it is just the same as deneb except you get this nice boost past 950~ right up and beyond 1100mhz..

all the following results are done with air on cpu and rams, stock cooler on cpu, a few scattered noctua fans..

Setup:
GSKILL FLARE 16000 7-9-7-24 2T
ASUS Crosshair 4 Formula
ASUS M4A89TD PRO
Corsair 1000W
AMD Thuban 1090










definatly a good looking module and if you are running anything asus rog or evga (im thinking p55) then they will look good in your rig..







My performance results with both ASUS motherboards are identical so I will present one set of results, give or take 5+ mhz if you put these sticks in either board. My stability test is based on 10 loops of LinX at the pictured settings.

First let me say 2T is not required, if you want to run 4 modules then to push the same frequencies it might be required, but on both these asus boards I tested I could do everything at 1T, 2T gave me a few more MHz but nothing dramatic.


Max validation is 1109 MHz, but retardly cpuz has overriden my max memory validation because at a lower memory speed the CPU SPEED is higher..


7-9-7-20 1T
1075 mhz pass
1100 mhz fail




7-8-7-20 1T
950 mhz pass
975 mhz fail


this actually crashes here at loop 7 but gives a good indication that there is hardly any scaling with these modules at cas level, PSC is totally TRCD bound.. so forget your cas 6-11-6 world record, go push your TRCD instead, cas is so BH5



6-8-6-20 1T




6-7-6-18 1T
850 mhz pass
875 mhz fail



6-6-6-18 1T
750 mhz pass
775 mhz fail



Some notes on this kit, price point is good for the performance, sub timings you can go fairly aggressive, and for anything except 32m this kit should be a real aggressive kit in competitive benching... i found i could hit all the high frequencies without really sacrificing any nb, with earlier PSC kits you had to have a very loose nb and trcd to push the kits..

[eva2000]

Hey James

PSC seems to be improving

[TheGreenFoX]

Nice tests... Would this be a good Deneb kit for CAS6 1600Mhz? :-)

[Royalk]

6-7-6-18 850MHz is really nice.

[hiwa]

let's dance man, new PSC are funny

[Tomasis]

very good mem for every $ what volt? stock or more?

how is linx on mem .. same as prime95? how much stable on 32m pi?

[Smartidiot89]

What's the difference between the 1800mhz 7-8-7-24 and 2000mhz 7-9-7-24? Identical but different in binning?

Nice results, glad to finally see som good AMD-specified kits out there

[tool_462]

What's the difference between the 1800mhz 7-8-7-24 and 2000mhz 7-9-7-24? Identical but different in binning?

It is G.Skill so probably just random binning from what I've understood. Based on their other kits, if they have the same ICs they usually clock the same.

[Smartidiot89]

It is G.Skill so probably just random binning from what I've understood. Based on their other kits, if they have the same ICs they usually clock the same.

Thanks

[Pt1t]

let's dance man, new PSC are funny

2000 CL6 for X58 are doing almost the same



http://wd.ch-img.com/1183821-gskill-...0cl6-kit-2.jpg

[dinos22]

not bad man these new PSCs

[hiwa]

2000 CL6 for X58 are doing almost the same


http://wd.ch-img.com/1183821-gskill-...0cl6-kit-2.jpg

yeap, i know three PSC ics : XDT,XEE,XDE seems the best one is XEE

[youngpro]

yeah i am fairly impressed, they are getting better slowly...

What's the difference between the 1800mhz 7-8-7-24 and 2000mhz 7-9-7-24? Identical but different in binning?

Nice results, glad to finally see som good AMD-specified kits out there

performance? real world performance 2000 7-9-7 is going to kill 7-8-7... im not sure you can really compare these.. even for 32m 2000 is going to kill 1800..

if you are comparing directly 2000 7-8-7 for a hyper kit and 2000 7-9-7 for a psc kit.. then maybe 2-3 seconds for 32m at 4ghz.. real world performance different, hardly anything...

if you are buying hypers for your gaming rig YOUR WASTING MONEY... check out psc or bbse

[BeepBeep2]

yeah i am fairly impressed, they are getting better slowly...




performance? real world performance 2000 7-9-7 is going to kill 7-8-7... im not sure you can really compare these.. even for 32m 2000 is going to kill 1800..

if you are comparing directly 2000 7-8-7 for a hyper kit and 2000 7-9-7 for a psc kit.. then maybe 2-3 seconds for 32m at 4ghz.. real world performance different, hardly anything...

if you are buying hypers for your gaming rig YOUR WASTING MONEY... check out psc or bbse

Umm on AMD, the difference is minimal.

7-9-7 is = to 8-8-8. 6-8-6 is equal to 7-7-7. You get what I'm saying.

AMD is different than intel when it comes to timings and memory badwidth.

[dinos22]

have you done the testing beep with those specific clocks?

[BeepBeep2]

have you done the testing beep with those specific clocks?

A bit.

My data is lost.

Changing Cas latency from 6-6-6 to 7-6-6 did almost nothing.

[dinos22]

you really should actually do some testing before calling out someone like james about his comments as he is very experienced with RAM testing
i'm not saying you are wrong but you need to qualify exactly what you mean at specific clocks and timings and what benchmarks do get affected

AMD rigs definitely like tighter timings but your statement is too broad to be meaningful

[BeepBeep2]

Don't take what I say too seriously.

Calling G*Skill flare a powerful AMD performance kit though,...I don't really agree with.

I like the results though, and PSC keeps getting better and better.

[Dahaka]

Sorry, wrong thread

[PcCI2iminal]

good job as always James

im testing the same Flare mixed with Pefect Storm on P55

[youngpro]

great start man, everyone is loving the 8gb at the moment, i might mix in 4GB of GSKILL PID and see how my results go..

Don't take what I say too seriously.

Calling G*Skill flare a powerful AMD performance kit though,...I don't really agree with.

I like the results though, and PSC keeps getting better and better.

well man im not so sure, what sort of timings and frequency are you looking for on the AMD platform... i am i guess looking more at a solution someone can run in a rig.. if you want the HIGHEST PERFORMANCE you will always look to a hyper kit, and thats why they have a premium price..

but at 1100mhz even on a hyper kit youll be looking at 7-8-7 and this kit will be okay at 7-9-7.. so the question is, for that extra few seconds in 32m and that tiny gain in 3d benchmark, are you willing to pay the premium? and for your gaming rig, or 24/7 workstation, is that a good option?

im just saying

[Oliver]

Thanks for posting James, much appreciated and as always nothing to say about the work other than

True in terms of memory 24/7 wise, it can`t be argued that its wise to spend $$ on killer mem, versus other performance upgrades the money could be used on (or saved)

Benchwise, ya hypers is THE bomb, but also comes with a pricetag after.

24/7 and performance vs $$ is the force in G.Skill and it`s kinda there brand now, close to "full" performance, at a very attractive price.

keep pushing it

[BeepBeep2]

great start man, everyone is loving the 8gb at the moment, i might mix in 4GB of GSKILL PID and see how my results go..



well man im not so sure, what sort of timings and frequency are you looking for on the AMD platform... i am i guess looking more at a solution someone can run in a rig.. if you want the HIGHEST PERFORMANCE you will always look to a hyper kit, and thats why they have a premium price..

but at 1100mhz even on a hyper kit youll be looking at 7-8-7 and this kit will be okay at 7-9-7.. so the question is, for that extra few seconds in 32m and that tiny gain in 3d benchmark, are you willing to pay the premium? and for your gaming rig, or 24/7 workstation, is that a good option?

im just saying

The Hyper kit in my sig can be had for less than $200 USD

[Oliver]

STT 2200C8 is nice if its available at your place, but again its not screened as CMGTX2 and thus you can get a less killer kit, tought kit consitency is ok ~.

at geizhals.eu the 2200c8 kit is listed in three shops, cheapest is 274$, other two is 312$.

Availability is also one of the key points in G.Skill succes, along with the price/performance it makes a really attractive kit.

Alone in Denmark, gamers and overclockers are all over G.Skill memory.

[websmile]

@ Oliver You are right, even the kits listed are more or less not available, I asked some weeks ago- so I imported my STT from US, incl. the EU-Tax, currency conversion and shipping I paid about as much as the Flare would have cost at the cheapest offer in germany- still a lot of money and lot of work

P.S. Interesting tests nonetheless- good running ram on AMD