OC Report :: Corsair Dominator 2x1GB PC3-14400 CL7

[3oh6]

I thought I would kick off 2008 with a fresh OC Report and who better to do it with than Corsair and their flagship DDR3 kit, the TWIN3X2048-1800C7DFIN. We are all familiar with this Freemont California based memory manufacturer and in particular their support staff. Members from the Corsair support staff are seen floating about in almost every major enthusiast web site including this one. The friendly likes of Redbeard, Yellowbeard, and XS's very own Bachus_Anonym are among the most approachable reps that I know of on the internet and are always ready to help. Eric/Ryder, you are most definitely one of the most approachable too, that is why I said 'among' for the Corsair cats...as for you Tony, if it’s one of your good days then you can be on the list too . I won't be needing their assistance today but I may need some pharmaceutical assistance because I have finally gotten my hands on some of the best clocking DDR3 currently available and my excitement is hard to disguise.


The Dominator line of memory was introduced some time ago with the emergence of DDR2 high frequency and low latency kits. I can't recall exactly which kit was the first to sport the Dominator brand but I remember the TWIN2X2048-10000C5DF being the first I saw on display at the Computex trade show back in June of 06. At inception, the Dominator line only recruited the very best of Corsairs memory for membership but shortly after release, the TWIN2X2048-6400C4D (now retired), had a change in ICs that brought them down to the 'average' level. This eliminated a little bit of luster for the Dominator name. The TWIN2X2048-8500C5D was recently announced to have dropped Micron ICs and it too has retracted another chuck of the Dominator dominance from its name.

Even still, the Dominator line of Corsair memory has proven its worth in the enthusiast community and the first DDR3 offering supporting tuxedo black heat sinks is definitely at the top of its game. The memory we are looking at today technically aren't the first Corsair Dominator DDR3 modules but an updated version to include Intel XMP profiles for the latest X38/X48 chipset. I am of course talking about the TWIN3X2048-1800C7DFIN, 900MHz of CL7 goodness.

Directory:

• Specifications, Features, & Pricing
• Package & Module Photos
• Asus Maximus Extreme Setup
• Dual 32M OC Report
• Benchmarking Results
• Deep Thoughts...By 3oh6

Before we get started, I have to state that these modules have been provided to me directly from Corsair as a review sample. As with all manufacturers I present, I have been assured these are not a hand-picked or cherry-picked sample. They are simply your run-of-the-mill Dominator retail kit and the results should be compared with other user’s results to provide a complete picture of how this memory tends to overclock.

[3oh6]

I won't get into too much discussion about the specifications but we can see these modules are rated for 900MHz or DDR3-1800 at 7-7-7-20 timings with 2.0v. The Dominator DHX (Dual-Path Heat Exchange) heatsink is the identifying trademark of the Dominator line but has proved to be so beneficial that Corsair has extended its use to mid-range products in the DHX line which sport a silver DHX heatsink. The unique DHX heatsink is designed in conjunction with a special PCB to not only remove heat from the ICs outside surface area, but also through the PCB side of the IC where heat is transferred from the IC through the ball grid array solder joints.


Without going into great detail, we can see the photo below showing the backside of a Micron mobile DDR BGA package (D9DKJ). Now D9GTR/GTS are not going to be the exact same as far as rows and columns of solder joints go but for those un-familiar with how a BGA IC connects to the PCB, it does the trick. These solder balls not only secure the IC to the PCB and transfer the electrical signals, they also transfer heat to the PCB according to Micron. The DHX PCB is designed to interface directly with its own heatsink fins to alleviate this heat build-up in the PCB and in turn lower operating temperatures of the ICs. This is obviously where the name, Dual-Path Heat Exchange (DHX) comes from since there are two paths for the removal of heat. There will be some photos of the dual heat sinks in just a moment, for now, let's get back to the specifications.


The first CPU-Z screenshot displays the SPD profile of these modules. There are three JEDEC SPD profiles that allow these modules to POST at relaxed timings/voltage to ensure compatibility so the end-user can at least get into the BIOS to adjust settings. I can confirm that these modules have no problem POSTing in an Asus P5K3-Dlx with setup defaults loaded (BIOS 0604). There is also a XMP profile that stands for the Intel specified Extreme Memory Profile. This is a standard developed by Intel to provide similar capabilities to that of NVIDIAs EPP profiles. With X38/X48 motherboards you simply select the XMP profile of your choice, in this case #1, and the motherboard will adjust the settings accordingly, including voltage.


In fact, with this kit on the Asus Maximus Extreme, it automatically loaded the XMP profile on BIOS defaults. FSB was set to 450 with a 7X multi for 3150MHz and the vCORE was left to the default VID of the CPU. The memory is then obviously set to 1:2 for 900MHz and 7-7-7-20 1N timings. Heck, a Performance Level of 6 was even set which put performance of this automatic overclocking right up there with a manual setup by someone who knows what they are doing. The vDIMM was also adjusted to 2.0v but unfortunately this puts the vDIMM at 2.1v on the Maximus Extreme. This is a product of the motherboard and not the memory. Overall though I was quite surprised that the BIOS defaults (BIOS 0803 BTW) would choose the XMP profile and it was 100% stable at BIOS defaults. My first experience with XMP is quite positive because if I was just an average user, I would have been quite pleased with this ease of overclocking. Without XMP, it would take some effort to get the same result manually adjusting settings.

Prices/Availability at time of posting:

• NXSource.com - $616.27 CDN
• NCIX.com - $649.99 CDN
• NCIXUS.com - $614.55 USD
• Newegg.com - 554.00 USD
• ZipZoomFly.com - 553.00 USD
• MWave.com - 553.00 USD

Once again, we get screwed up here in Canada with the prices but at least it is available and it isn't too bad in comparison. Naturally, like it is will almost all DDR3, this kit is priced very high. But do take into account the type of pricing we saw with the high binned DDR2 kits when they first came out. Even now, the cheapest PC2-10000 kit of Dominators I could find is $475USD at the low end of the price list so for this memory to be coming in at $550USD isn't that bad at all. Heck, it could be considered damn good when you put it into perspective considering how mature the DDR2 market is. So all you price whiners can stick a sock in it...if you want to play, you have to pay.

[3oh6]

Click for full size...

I am not sure if the retail kits of these modules come in the same kit but Corsair certainly takes care of their press kits. The modules come encased in styrofoam with the Dominator fan in a separate box to the right. Even the Dominator fan is heavily protected inside its smaller box with an air pocket making sure the fan doesn't move an inch during transport.


Moving on to the modules themselves, we can see no changes to the DDR3 dominator heatsink from the DDR2 counterparts aside from the addition of the Intel Extreme Memory logo and DDR3 on the tip. There isn't really much to discuss because we should all be familiar with these heatsinks having seen them on Corsair modules for over a year and a half now. The PCBs used are obviously unique to any other manufacturer’s memory due to the DHX features. The gold colored strip along the top of the PCB is where the inner silver heatsink interfaces with the PCB which we will see close up in the next couple of photos.


As just mentioned, the part number on these PCBs is going to be unique to Corsair memory. Looking through a macro lens we have the ability to see the ICs underneath the heatsink without the need for removal. The backside of the module contains standard thermal material between the heatsink and the PCB since there are no ICs on that side. On the other side, however, the ICs are contacted by the heatsink with a very thin thermal material. We can barely see the material which looks like almost a clear glue. This material is baked onto the ICs and removing the heatsinks will generally result in the IC coming off with it. One thing of note is that at the bottom of the IC we don't see any thermal material and the IC isn't even touching the heatsink. This obviously isn't helping with heat removal and I would like to see complete contact on the IC from top to bottom. The last photo is a nice close-up of the interior heatsinks providing cooling for the specially designed PCB. With the steps Corsair has taken in module cooling technology, it is no surprise to see them amongst the top memory manufacturers, especially for enthusiasts.

[3oh6]

Click for full size...








I still haven't had anything cold on this motherboard yet and this memory has been the only kit to grace the DIMM slots of the Maximus Extreme so there are no photos or results from any phase or dry ice work. I will certainly add those when I do go cold with this board and these modules. I have also bypassed the Dominator fan in favor of 120mm fans in order to help with the cooling of that huge heat-pipe mess of a chipset/PWM heat sink. During testing, as we can see, I had a couple different fan setups but at all times there was a 120mm fan over top of the Dominator memory.

Hardware used:
MB:          Asus Maximus Extreme (BIOS 0803)
CPU:         Intel C2D E6850 (L724A479)
CPU Cooling: Thermalright Ultra-120
GPU:         Biostar 8600GTS 512MB DDR3
NB Cooling:  Stock
SB Cooling:  Stock
PWM Cooling: Stock
PSU:         Silverstone Zeus 560W
HD:          Seagate SATAII 80GB 8MB NCQ
OS:          Windows XP Pro SP2

Ambient Temperature: 22-23C

The motherboard obviously isn't volt modded as it likely won't be necessary for this ROG board. I also haven't changed the chipset cooling so this motherboard is literally straight out of the box. So far, it has performed admirably but as mentioned, it still hasn't been cold so I will wait until I get some benching down and the board sees sub-zero before I praise it like my monster Asus P5K3-Dlx.

[3oh6]

Before looking at the results, I will again tackle some disclaimers that I wish more people would discuss when providing memory results. Too often we see crazy results with no context like the fact that the motherboard over-volts memory by 0.10v and only the BIOS selected voltage is listed skewing the actual numbers quite drastically.

To start off with, my Maximus Extreme over-volts the BIOS selection by exactly 0.10v through the entire voltage range. The actual voltage is picked up correctly by software so the Everest voltage reading for vDIMM, and all other voltages for that matter, are correct. The Everest vDIMM reading is identical to what I am picking up with a DMM. This makes it very nice for screenshots in what I am doing here. All voltages listed are not the BIOS selected but the actual voltage that the memory is receiving.

Second, this is only dual 32M SPi stable. As always, I am not claiming anything more or less. Prime, OCCT, HCI Win Memtest, and distributed computing stability are going to be lower. At some clocks, substantially lower, but a lot of the time it won't be too much lower if any at all. The DDR3 modules I have worked with seem to be a little weird at times with some sticks showing dual 32M SPi as the first sign of stability issues passing hours of Prime or HCI but failing dual 32M SPi. Other kits have shown some instability in 3DMark 01 before anything else with a couple small adjustments to secondary timings alleviating the in-stability. So there is a lot to think about when going over various DDR3 overclocking results. I simply test for dual 32M stability because it tends to be a great gauge of 24/7 stability and is relatively quick compared to other stability tests. Time is always a factor when it comes to stability testing.

Lastly, I will repeat that this kit of memory was not purchased from a retail store but sent directly to me from Corsair. This is a review sample but Corsair does not influence review samples in any way and I am confident that these results will be about the average of what all other retail kits will do. Now, with all of the BS out of the way, let's finally see what this memory can do.



6-6-5-X 1T:
Intel X38:
Click link for full size screenshot...
752MHz :: 1.90v | 778MHz :: 1.95v | 792MHz :: 2.00v | 812MHz :: 2.06v | 828MHz :: 2.10v | 848MHz :: 2.16v

With this being the first kit of memory for me on this new Maximus Extreme, it took a while to get things going but once I did, this memory took off running. At 6-6-5, the memory really seemed to be off to a great start with loose secondary timings but when it came time to tighten things down, this memory refused to play nice. In particular tRFC absolutely refused to allow this memory to clock well at less than 60. At the next lowest BIOS setting of 48, this memory refuses to co-operate at any timing set as we will see shortly. At these frequencies I was certain that tRFC of 48 would be more than fine but even at 750MHz, tRFC 60 allowed for much higher memory clocks. Notice the use of PL5 right up to 432FSB...I am loving this motherboard. It is just a shame that the X38 chipset won't allow a PL lower than a single value from CL because 500FSB+ with 8-7-6 will only allow a minimum of PL7. P35 would allow PL6 at 500FSB+ when CL8 was being used.


7-6-5-X 1T:
Intel X38:
Click link for full size screenshot...
870MHz :: 1.90v | 896MHz :: 1.95v | 914MHz :: 2.00v | 930MHz :: 2.06v | 950MHz :: 2.10v | 966MHz :: 2.16v

Making the jump up to 7-6-5 provided a large jump in frequency starting at 870MHz with 1.90v and steadily climbing up to 966MHz at 2.16v. I am sure this kit is a 1000MHz 7-6-5 kit but with a little higher voltages than I want to use just yet. Overall, the 7-6-5 timings seemed to react in a similar fashion to the 6-6-5 timings and again, tRFC 60 was needed for any kind of decent clocks. It just seems that either this motherboard or this memory doesn't like tRFC 48 and that is the end of it. With further testing in another motherboard and more memory in this one I will be able to determine which one doesn't like tight tRFC but I would imagine it is simply this particular kit.


8-7-6-X 1T:
Intel X38:
Click link for full size screenshot...
940MHz :: 1.90v | 972MHz :: 1.95v | 984MHz :: 2.00v | 1000MHz :: 2.06v | 1014MHz :: 2.10v | 1026MHz :: 2.16v

The 950MHz dual 32M SPi wall that I found on my P5K3-Dlx was not only passed but absolutely blown away by this kit on the Maximus Extreme. Secondary timings needed to be loosened up early on in the 8-7-6 testing but that is more a product of the chipset not being able to handle tight timings at such high frequencies for dual 32M stability. With single 32M runs, secondary timings are able to run at tighter settings. However you look at it, DDR3-2052 dual 32M stable at only 2.16v is mighty impressive and has an ear to ear smile tack welded to my face.

[3oh6]

8-7-6-X 1T HCI Win Memtest:
Intel X38:
Click link for full size screenshot...
1000MHz :: 2.13v

Normally I don't show this kind of stability testing in my OC Reports but I just finished it for another thread and thought it was kind of impressive so why not post it here for those 24/7 stability guys. It could probably be stable at a higher frequency but I was just looking for 1000MHz on the nose. Yeah it's only 2 hours but that should equate to 24/7 stable and it is definitely Prime stable if it passes HCI from what I have found in past experience.

[3oh6]

Well there you have it…my first work with the X38 chipset on the Maximus Extreme and my first work with a kit of Dominator memory from Corsair. If I was to compare these results to the Crucial Ballistix OC Report on the Asus P5K3-Dlx it would be a no brainer that this kit wins at every timing set hands down. Of course that isn't exactly a fair comparison because the X38 chipset has been reported to clock DDR3 better than P35. It certainly handles DDR2 high clocks much better than anything before it. I will of course find some time to test the Ballistix in the Maximus Extreme for an equal comparison but I think I will find this kit of Corsair PC3-14400 Dominator to still be ahead by a reasonable margin at all clocks.

At this time of posting I have not had the chance to do any phase change or dry ice benching with this kit on this motherboard but you can be sure there are plenty of sessions that this memory will see in the future. The Benchmarking Results section in the post previous to this one is where you will want to look for all those high clock results that will focus more on performance than stability.



As always, if you have any questions, comments, or concerns about this report, the memory, or the motherboard used; don't hesitate to ask. Also don't forget to keep an eye on this thread if you see it pop to the top of the list as I will be updating with future results. If there is something in particular you want to see from this setup, just let me know and I will do my best to accommodate. Thanks for your time and have a great new year...my 2008 is off to a great start!

[saaya]

and another really nice review jody!
Cant wait for more results, looks like a really nice kit!
this is on an x38 board? did you test the mem on the same board as the crucial mem as well? Maybe you could make a quick p35 vs x38 mem performance/clocking article? you should have a lot of the results you would need for that already right? and it would be really interesting imo.

[Gautam]

This stuff is indeed quite nice. Take em to 2.25v and watch them scream at 1800C6....you're doing way better than I could with v1.1's at CAS 7...940-950 seems to be tops for that old revision, at least for me.

[Oliver]

Really good report, there is gts on your kit eh? or have you good a kit with good gtr?.

Keep pushing it.

This stuff is indeed quite nice. Take em to 2.25v and watch them scream at 1800C6....you're doing way better than I could with v1.1's at CAS 7...940-950 seems to be tops for that old revision, at least for me.

Maybe because there is gtr on there old rev and there is gts and there new.

[3oh6]

thanks guys!

this is on an x38 board? did you test the mem on the same board as the crucial mem as well? Maybe you could make a quick p35 vs x38 mem performance/clocking article? you should have a lot of the results you would need for that already right? and it would be really interesting imo.

yes, the Maximus Extreme is X38. i haven't tested them on the P5K3-Dlx and likely won't. but...i have penciled in some face time with the Maximus Extreme for the Ballistix kit so watch for that thread to pop back up with some updates. i may end up running this Corsair kit on the P5K3-Dlx for some benches but based on preliminary performance testing, the Maximus Extreme will be my new main ride for 32M benching.

GTS or GTR, no idea. based on what has been coming out of camp Corsair though, i would venture to say GTS. perhaps one of the friendly Corsair reps could chime in and clear it up. it was a review sample sent out less than two weeks ago if that helps.

[saaya]

This stuff is indeed quite nice. Take em to 2.25v and watch them scream at 1800C6....you're doing way better than I could with v1.1's at CAS 7...940-950 seems to be tops for that old revision, at least for me.

pretty sure thats a veriation from stick to stick.
if theese sticks like high voltage they should run 880-890 665 with 2.25-2.30v

and oliver, why are you always asking everybody if they use gtr or gts?
in my expirience gts clocks the same as gtr...

[Mean Machine]

Things sure have happened since May when I tried out a pair of 1333MHz Corsair ES sticks, rated at C9. They wouldn't for their lives go past 1800MHz regardless of timings or voltages.

[EnJoY]

Great write up, and truly a testament to your testing abilities and thorough attention to detail!

[saaya]

Things sure have happened since May when I tried out a pair of 1333MHz Corsair ES sticks, rated at C9. They wouldn't for their lives go past 1800MHz regardless of timings or voltages.

micron? cas8 right? sounds more like you had some very good quimonda in your hands.

[Mean Machine]

micron? cas8 right? sounds more like you had some very good quimonda in your hands.

Probably wasn't any Microns, no. It was CAS9, a pair of these: http://www.bit-tech.net/hardware/200...24_1333c9dhx/1

Judging by the review, they actually oc'd pretty well for a pair of those, as the max they hit was 1520MHz... Thinking of it, I'm not sure if it was 1800MHz, could've as well been 1600. Point being, Micron rocks, I want a pair of those! Not willing to pay for them, though.

[JANJWJMC]

Great test Jody

Drr3 is the future and i tinking of buying mn new setup with ddr3

But mn question is wen i look at the memory compatible list
of different manufactures i see that the recommendation is
to but only 1gig or one slot of ddr3 why is that ?

Not only i see it at Asus and co but also at Micron an co .

[Oliver]

pretty sure thats a veriation from stick to stick.
if theese sticks like high voltage they should run 880-890 665 with 2.25-2.30v

and oliver, why are you always asking everybody if they use gtr or gts?
in my expirience gts clocks the same as gtr...

Afaik, gts is like good gtr, you know there is "normal" gtr and slightly better gtr.

gts is like good gtr and does not get under the "good" gtr clocks.

800 6-5-5-15 2.0v is standard for gts,but 1800 7-x-x-x 2.0v with CAS7.

gtr can also run that but not all gtr afaik.

[Rob94hawk]

Great review. My Corsair 1800C7DFIN did not dissapoint @ 1800/7-7-7-20-2T Untill I start going higher I have the fans off for now.

FYI there is no XMP profile in the BIOS for the GA-X38T-DQ6.

[saaya]

Afaik, gts is like good gtr, you know there is "normal" gtr and slightly better gtr.

what makes you think so?

gts is like good gtr and does not get under the "good" gtr clocks.

yeah, but have a look at the speeds they are binned
its binned at so low speeds and so relaxed timings that they could as well use one bin and then use that for 1066 777 and 1333 999 etc... and maybe they do

[Oliver]

what makes you think so?.

Some information i have got from M.Beier and you know there is seen examples of gtrs that overclock better than some other gtrs.

But maybe gts is holding a limit where good gtrs are?.

yeah, but have a look at the speeds they are binned
its binned at so low speeds and so relaxed timings that they could as well use one bin and then use that for 1066 777 and 1333 999 etc... and maybe they do

Dont know if they do, but maybe you are right .

You do have alot more knowledge in this area than me.

[dinos22]

nice work Jody

very thorough and good speeds
keep pushing it hahahah

[3oh6]

thanks for the comments guys, can't say that i don't enjoy putting these reports together

Great write up, and truly a testament to your testing abilities and thorough attention to detail!

thanks, i had to do something with all this free time over the holiday break.

But mn question is wen i look at the memory compatible list of different manufactures i see that the recommendation is to but only 1gig or one slot of ddr3 why is that ?
Not only i see it at Asus and co but also at Micron an co .

that is something i don't know. to be honest, i don't think i have ever looked at a manufacturers compatibility list. i always thought of that list as somewhat of a waste because you can always make any memory work in any motherboard whether it was manually adjusting timings or flashing SPD profiles. i would imagine it is just because no manufacturers have tested 2GB DDR3 sticks...its not like there are a whole bunch. i do no of one person who got a 2x2GB kit of micron D9GTR DDR3 and they ran real well in a P5E3:

http://www.ocxtreme.org/forumenus/showthread.php?t=2251

FYI there is no XMP profile in the BIOS for the GA-X38T-DQ6.

interesting, i was completely shocked when the motherboard actually set everything correctly on its own using the XMP profile and was 100% stable getting the memory running at spec. the kicker was that i didn't even have to choose the XMP profile in the BIOS, right after a CMOS clear, first boot was at the XMP profile. very impressive stuff from Intel. i never had a good experience like that with EPP. i guess that is what you get from a $350 motherboard

nice work Jody

very thorough and good speeds
keep pushing it hahahah

[saaya]

oliver, just take any random micron ddr3 kit and check what speeds and timings you can run with 1.5v and you will see for yourself, the different bin levels are almost identical

lol dinos, how come i get the impression your laughing whenever you post a reply? you always seem to be in a really good mood, whats your secret? ^^

yeah the compatibility lists from mfgs are rather useless imo...
it means that at least one board, of some revision of the board, with some bios revision, did work
which is at least something, but it doesnt mean every kit of said mfg will work in all of those boards.
the only thing they are good for imo is to see if the mfg has optimized their bios for a certain memory CHIP.

interesting to hear that xmp worked well... im rather skeptic towards xmp... the bios needs to be programmed properly for it to work and its a lot of extra work only few board mfgs will be willing to do i think. if you use a different cpu the bios has to select a different lower multiplier for xmp to work properly without ocing the cpu clockspeed. from what ive heard this didnt work perfectly on a few occasions.
It would be nice if intel would provide the bios code necessary to adjust the board for every combo of xmp mem and cpu and board mfgs only had to inluce this code in their bios... but i think its not possible cause every bios and board works slightly different :/

[dinos22]

oliver, just take any random micron ddr3 kit and check what speeds and timings you can run with 1.5v and you will see for yourself, the different bin levels are almost identical

lol dinos, how come i get the impression your laughing whenever you post a reply? you always seem to be in a really good mood, whats your secret? ^^

2xkids
lots of xxx
and plenty of good memory from Corsair
who wouldn't be happy hey?

whoever said X38T-DQ6 does not have XMP profile i say look again
however the GB board i have here is a bit of a dog no matter what bios
the asus P5E3 DLX i bought from Kevin(OPB) is a lot more promising but i am not happy yet until i see 2GHz cas7 benchmarks so i have to test all the bioses
does anyone know how the voltmod vcore out of this stupid mofo of a board lol and also why is asus being stupid and not just release a bios with full vcore (YES IT IS POSSIBLE THROUGH BIOS FOR THEM EEEEEK) and they don't allow downgrading bioses so i can't play with 0503 bios bachus used for his benches CRIES

Jody >> i have a corsair 4GB kit to test on this board too....we'll see how that goes soon
need that vcore for the board so i can slap my kingpin tube and LN2 gawd damn it