Established in 1989, G. Skill is synonymous with memory modules that deliver great performance and have very impressive overclocking capabilities. What stands out with G.Skill memory is that for modules that can overclock the way they can, the price of their memory is often lower then similar modules offered by their competitors which don’t nearly have as good overclocking potential. G.Skill’s latest memory entries – the RipJaws and Ripjaws X series – have been praised by many memory enthusiasts.

Today we are fortunate enough to bring to you a review of the Ripjaws X series’ latest DDR3 8GB (2×4GB) kit. Let’s take a closer look.

Packaging & Aesthetics

The packaging is the normal hard plastic packaging used by most RAM manufacturers. The default speed of 1866MHz, timings 9-10-9-28, command rate of T2, and recommended DRAM voltage of 1.5v are all shown on the sticks themselves.

The Ripjaws X kit we received is a 2 x 4GB kit with red and black aluminum heat spreaders, which protect the black PCBs. The look of the RAM as a whole is something we feel would look especially great in a board like the Asus Rampage IV.

As with many high end memory modules, the heat spreaders themselves can be removed and replaced with third party heat spreaders of your liking.

Next Page: Stock & Overclock Tests

Originally, I planned to test the memory kit subjectively with many benchmarks reflecting more real-world benefits of the high speed memory.  Based on similar tests published around the net in the past, I didn’t expect to see a huge increase in overall, everyday type performance.  So I also planned to test the memory in extreme conditions to give a better idea of where the memory should truly shine, with extreme benchmarking.

Well, as I wrote in the review, my testing was met by so much difficulty getting the DIMMs to run at rated speed on various CPUs.  The Bloomfield testing I knew didn’t have much hope, but after I got the i7 970 operational with marginal success, I was fully expecting the DIMMs to be stable in dual channel mode on the P55 platform with the Lynnfield CPUs.  I knew that Corsair only expected them to operate on Lynnfield, so after having tested two, and seeing that neither was capable…I had to give up on the the “everyday testing”.  (All these CPUs were paid for out of my pocket, and I could not afford anymore.)  The single run with the Xeon CPU was completed earlier and was my only chance to test the DIMMs with a sub-zero IMC.  Having troubles with that also discouraged me from trying again.  In the interest of getting the info to you in a timely manner (and knowing that 99% of my readers do not overclock with LN2), I decided to forgo further testing and get the article to press.

Well, in response to all the feedback, I’d like to clarify that these DIMMs are clearly capable of incredible performance.  But the purpose of yesterday’s article was not to just show off what an extreme overclocker could do with the right conditions….I wanted to give the bulk of my readers the more valuable information about the truth of high end memory.

I had the i7 970 under LN2 tonight while testing a graphics card on LN2.  The 3D testing ended prematurely, and I had a few liters of LN2 leftover, so I threw the GTX3 DIMMs back in the rig and spent about two hours trying to maximize memory speed for SuperPi 32M testing.  I was finally met with success in reaching the DIMMs rated speeds and timings, and in fact I was able to do a bit better.  I hit DDR3-2430 8-11-8-27 88 1T in triple channel mode…pretty impressive stuff.  I think there is still more in them with more time (and LN2) and a better CPU.

*The CPU was not maxed out for this run.

SuperPI 32M with DDR3-2430

When I got down to the last few drops of LN2, and didn’t have enough to attempt another 32M run, so I spent the last few minutes pushing the speeds up with PiFast.  This was not the max, only the second attempt, then the pot got too warm as I ran out of LN2 and I had to call it quits.

PIFast with DDR3-2500

This doesn’t change anything about my conclusion yesterday, I knew these DIMMs were special….that point was never argued!  The problem is that you’ll need a special CPU and possibly even some extreme cooling to get what you paid for.

i7 970 under LN2 & Dominator GTX3 DIMMs

Thanks again for all the great feedback!

Begin Rant:

“Absurd:  the quality or condition of existing in a meaningless and irrational world.”

Welcome to the absurd world of premium random access memory.  Here today I have some of the latest and greatest from a company which needs no introduction as they have defined this segment for years.  There are many memory vendors from whom I have purchased RAM throughout my time as a computer builder and user, but none stand out to me like Corsair.  Why is that?  Why is it that my friends who don’t know diddly-squat about their computers refuse to buy anything but Corsair?  Let’s not kid around, to play with this brand (and more specifically the Dominator product line) you have to open up your wallet.  Corsair doesn’t even flinch each time it announces a new product with pricing that seems to scream highway robbery, and their pricing adjustments as time goes on seem less influenced by market trends than other brands as well.  Are Corsair customers getting what they pay for?  Is their product really superior?   

Dominator GTX3 & fan

The Dominator GTX3 memory is sold at Corsair’s online store as individual 2GB DIMMs which you can combine into pairs for Lynnfield based systems, or in sets of three for Bloomfield or Gulftown based systems.  This memory is rated to operate at DDR3-2400 at 9-11-9-27 timings.  Corsair has only validated the speed and timings of these DIMMs with the P55 platform in 2×2GB configuration; a 6GB kit on the X58 platform is not a validated configuration.  The price? A whopping $189.99 per DIMM (where’s my “yikes” smilie?)!  In addition to the price of the DIMMs, Corsair recommends buying a cooling fan as well ($35)…so the total for the kit I have here today is $604.96 plus tax and shipping (if applicable).   

That’s a fairly tall order if you ask me!  Why?  Well, Intel’s CPUs contain the integrated memory controller (or “IMC”) which operates this memory, and the controllers are only rated by Intel to operate at DDR3-1333 and DDR3-1066 for Lynnfield and Bloomfield/Gulftown respectively.  So concerning Bloomfield/Gulftown operation, you’d be asking the memory controller to operate at a 125% overclock!  With Lynnfield, we’re talking about an 80% overclock.  If that sounds like a lot, that’s because it is!   

This was intended to be a formal review, but I’ve grown frustrated with that intended goal.  The problem isn’t the memory, but the IMC on the seven CPUs that I’ve used to attempt to test with. Yes, I said seven.  I tested one Clarkdale i5 670 dual core, which obviously didn’t stand a chance.  If you run a Clarkdale CPU and you’re eyeballing high end memory like this, STOP IT!  High speed memory like this, is not going to run well with Clarkdale.  I also tested with two Lynnfield CPUs, one i5 750 and one i7 860 on the Gigabyte P55A-UD7 motherboard.  I also tested with two Bloomfield CPUs on the Gigabyte X58A-UD9 and the eVGA Classified E760, both were i7 920s.  Lastly I tested two Gulftown based 32nm CPUs on X58, a Xeon X5677 quad core and an i7 970 six core.  With the four X58 based tests, I tried in both triple channel and dual channel mode.  In every single attempt, I failed to achieve stable operation at the rated speed and timings specified by Corsair.

Next: The Rant Continues

I would like to point out first that G.Skill has a current supported motherboard list for these modules consisting of the following boards, ASUS M4A89TD PRO, ASUS M4A89GTD PRO and ASUS M4A88TD-V EVO/USB3. For this purpose, although we did try them out on an ASUS Crosshair IV Formula and they appeared to work fine, we decided to stick with the officially supported ASUS M4a89TD PRO. Also an important note, currently the only officially supported CPU’s are Phenom II X6 Thubans, which we stuck with for official support reasons.

I would also like to point out at this time that after doing some research, GIGABYTE has recently posted on their site a revision 2.1 890FXA-UD7 that claims to support the 8 gig variant of these modules at they’re rated specs. Unfortunately, we have not been able to acquire one of these boards thus far for testing and G.Skill has not updated there site with any info on this.

Let’s take a look at the modules and what you get.

The G.Skill Flares:

Included with these particular modules is a fan that attaches to the module’s heatspreaders. It unfortunately uses a 4 pin molex connector; not too excited about that.

As we can see here the box clearly states the motherboards officially supported and tested by G.Skill.

As usual with all ram reviews I like to point out possible clearance issues so users are completely aware. These particular DIMMs will indeed interfere with some large heatsinks.

You will want to be sure that whatever you are using for cooling can be oriented both ways to ensure clearance.

Continue on for Overclocking, Testing and the Conclusion. >>

Sorry about few days delay here, but what you can do about it. Nothing really… So i bough used PSU & Cellshock Black Edition C7-12800 4×1gb sticks. Okay, well first ill introduce my pc:

Phenom II x2 550BE , True Black, Radeon HD3850 512mb, Asus M4A78T-E.

Let’s start with PSU, bought it for 52 euros, not bad deal because it was Thermaltake ToughPower 750 watt. Few notes thought, it missed few modular cables… not big deal for me. So i started to put back together my pc, after while i booted it up first time… well tryed. No go, no matter what. So i plugged hdd’s of first and tryed again. WoW! It was alive, but the reality hitted on my face pretty quickly. Erm what? so what to do without hdd’s? Nothing. Putted back again everything, no go… ugh… then i unplugged 6 pin pci-e cable from radeon.. and wow it was alive again. You know what happens next.. Doh. So i was like in situation where i have to choose between them, ermm no. Contacted seller, got my money back. He tested it too, he’s pc was working allright. Umm what? Strange problem then, but let’s head to Cellshocks.

Opened up first kit, booted fine with 1333mhz and CL7. Good working… Thought i did not try memtest or superpi at all. Tryed 1600mhz and no go just bluescreens.

After 2 hours of trying different slot’s latencys and all the setting i quess, i had to give up! Opened other set, unplugged first ones and attached second  ones to mobo, it fired up to at 1333mhz. Nice, moved up to 1600mhz and cl7. Working ! oh yes finally some good stuff ahead. I did not expect it after first kit. Moved up to 1800mhz, again working. So i was finally getting decent clocks on memory and with cl7 @ 1.9v rated, which is stock voltage for these sticks. Tested some superpi 1m &8m and passed!

Well im happy, at least something is working and hitting good numbers on screen, gotta test these tomorrow more.  But after these backdraws i got something good and feeling happy right now.

Now some updates to G.SKILLS which i have been testing lately. My mind have changed few times when it comes to these.

120+ euros, pretty decent clocks on AM3 platform. I migh say; best bang for buck, but with few exceptions. These are very good for intel, for AMD you might not be lucky at all with these sticks, when it comes to your mobo&cpu. With C3 cpu’s your mems might go little further, maybe. Cheaper CPU’s might have weak memcontroller and that’s why getting near to AMD memspeed limit, might be impossible which is not your mems fault.

PS: Next i will post about Cellshock C7-12800 black editions, BeQuit psu and x2 4870. Bought these as used. Also i might make and order for Corsair Domitar GTX2 memorys,  200 dollars each… Ouch.  But hey, 6-6-6 latencys and 1800mhz for AMD AM3 platform… not bad! As those need only 1.65v for it. Ooh i will se some wet dreams about them. Below G.SKill update. Before weekend, if i have time enough to run some superpi 32m test

But here’s my latest SuperPi 32m run with 1.6v & 7-8-7 timings. Yeah had to loose timings bit as my cpu barely can hold +1800mhz.

Check it out. Not bad afterall and keep in mind, only 1.6v which is good for 24/7 use. These might go little further, but as i dont wanna go over rated voltages. Otherwise i could get easily more out of these.

Well i dunno where to start, eh?

Uhmm yes, i got few days ago new memorys for my AMD AM3 platform… choice were easy, just pick up cheap and fast memory.

I ended up on G.Skill Ripjaws DDR3-2000 F3-16000CL9D-4GBRH kit, price tag of 125 euros. I though now i can start oc*ing my setup, but oh boy how wrong i was. As the kit fly’s on intel chipsets, it wont fly on AMD chipset. Im barely getting some decent numbers: CL7 and 1600mhz and stock voltage. Well that isn’t bad if many would say, but… you have to start 1333mhz divider and raise htt to 240 and after that lower htt multiplier and nb multiplier to get it running 24/7 stable. Did i mention about 1600mhz divider? it wont boot with it, duh!

Cas Latency 6 is also no go for these on AM3 platform. Well these stick’s can stay for while for future use. I Do believe these have Elpida BBSE chips, but im not sure about it. Duh, it’s 01:00 AM here, have to go to sleep, tomorrow i will post some more what i did buy today

I’m Out

Jerry

But here’s a screenshot with dirty quick run.

I’m not very big on chit chat so lets get right to the testing after seeing a shot of the ram.

System Specs:

AMD Phenom II X4 955 BE

Asus M4A79 Deluxe

G.Skill F2-9600CL5D-4GPI (1200 5-5-5-15 2.1v)

PC P&C 1200w Turbocool

Thermalright Ultra-eXtreme 120

Testing

First up, I tossed the timings and voltage to spec, left everything else auto and ran some Prime 95 blend to stress the ram with the IMC.

Next I tested scaling with voltage

800 4-4-4 1.8v                          900 4-4-4 1.8v                     960 4-4-4 1.9v

980 4-4-4 2.0v                   1032 5-4-4 2.1v                          1066 5-5-5 1.8v

1100 5-5-5  1.8v              1150 5-5-5 1.9v

1200 5-5-5 2.0v              1226 5-5-5 2.0v

Conclusion

This has got to be one of the easiest and most versatile sets of ram I have oced with on AM2+.  The fact that it exceeds some of the 2×1g  “overclocker” sets I have used makes it that much sweeter. I personally am not a fan of  high voltage, and a relatively low-voltage set of memory doing speeds that require far more voltage with other sets of memory is a very welcome change indeed.  All in all, this set gets a big thumbs up and appears to compliment the AM2+ platform quite well. It’s nice to see with the majority of the manufacturers making the jump to DDR3 that G.Skill did not forget about the enthusiasts that still use DDRII. My only complaint would have to be the tall heatspreaders causing interference with the inner slots and high performance heatsinks such as the popular TRUE 120 and Xigamtek 1284 on AMD platforms where the heatsink can not be oriented in a different direction.  For users of heatsinks like these, you will most likely be forced to remove the heatspreaders on the memory if you wish to use them in your boards ‘primary’ memory slots.

On a final note, I would like to add that the Asus board I tested this on has some teething issues with ram clocking at the moment with the 955 CPU.  I’m quite sure that this is bios related and may have hindered the full potential of this set of ram.  Unfortunately, I did not have a different board to test with at present time due to me getting a little frisky in the overclocking department with my other AM2+ board.

Many thanks to G.Skill for the review sample.  If you’re looking for the best 4GB kit of DDRII available on the market today, look no further than the 4GB PC2-9600 PI’s, which can be found here at Newegg.

]]> http://www.techreaction.net/2009/05/18/gskill-f2-9600cl5d-4gbpi-ddr-ii-reviewed-on-am2/feed/ 1