Corsair Vengeance 8GB 1866MHz C9 Review

[Ket]

Well fellow readers just 30 minutes ago a UPS man delivered the replacement Vengeance modules. Are they any good? Absolutely not. This replacement kit is actually WORSE than the kit I sent away. At least the old kit could load Windows @ 1866 9-9-9-25 1T 1.65v, this replacement kit BSODs instantly after putting in my password at the login screen with 9-9-9-25 1T timings @ 1866 1.65v. Less voltage just results in the system failing to POST, period. If this isn't bad enough I've simply been sent another Rev. 8.16 kit despite Corsair knowing the problem and me being told it would be probable the kit would be replaced with a Rev. 5.11 kit. Before I consider using my .177 on this utter failure of a product I will try 8-10-8-25 1T timings @ 1866 1.6v-1.65v and see what the Corsair guy who contacted me in the beginning has to say.

Again I will reiterate, right now for just a extra £20 or so you are FAR better off with a 4x4GB kit of G.Skill Ripjaws. This Vengeance kit is far more trouble than its worth, I'm not going to spend a single penny more on RMAs. This is Corsairs problem to sort out, not at my financial expense.

Ed - Forget these Vengeance memory modules, even 8-10-8-25 1T @ 1866 1.65v BSODs. Never in my life have I used such a piss-poor product.

ED 2 - and here is the PM I've just sent to our Mr. Corsair man.

I've just got the replacement Vengeance kit and this replacement kit is absolutely atrocious. Once again I tried 9-9-9-25 1N timings @ 1866 1.65v and the system just BSODs at the login screen. I can't even get into Windows like I could with the previous kit. Lower voltages just result in the system failing to POST, period. I also didn't get a Rev. 5.11 kit, just another Rev. 8.16 kit. To be honest, I'm getting tired of being sent sub-standard memory. For the same money I've now spent on this terrible memory I could of got a kit of the most stellar G.Skill F3-12800CL7D-8GBXH. I want something done about this situation, immediately. I'm not spending a single penny more in RMA costs on a product that quite frankly, is the worst product I've ever had the displeasure of testing.

[Formula350]

Ugh, that sucks

What a great example of why the memory industry has me worried. Aside from DDR4 being slated for 2013 release, with expectations only being a market adoption of 12% by 2014, but full integration by 2015. We are only about to enter the 4th quarter of 2011 FFS! Seems like ever since the DRAM price-fixing scandal that IC quality has tanked, a seeming indicator of being told "Yea, alright, we were bad... We'll drop prices now to what they should be, but hey, we're still going to rake in the same amount of cash because we'll stop making anything good. KTHNBAI!"

I know that 4GB modules aren't as capable as their 2GB & 1GB counterparts, but is the performance gain from the increased system capacity really offsetting what is lost from the high speeds with low timings? I'll admit my ignorance, but it just doesn't seem likely :\

EDIT: ... NOW JEDEC says mid-2012!! (and I only was there to check something totally unrelated to all this)

[Ket]

Any half decent 8GB memory kit makes up for the speed loss, yes. A half decent CHEAP 8GB memory kit would be able to handle 8-9-8 timings around 1.65v at 1866, but have no OC headroom. A half decent 8GB memory kit REASONABLY PRICED would be able to handle 8-9-8 around 1.65v at 1866, with enough OC headroom to touch 2000MHz @ either 8-9-8 or 9-9-9 timings depending. If you buy G.Skill Ripjaws, I'd put money on their CL7 1600 8GB kit doing 8-9-8 @ 1866 with 1.65v.

[Smartidiot89]

@Ket: I've so far only had limited experience with Vengeance modules but out of 64GB none wanted to run at rated specs in Dual-channel @ 1866 MHz with AMD A8-3850 "Llano", but we do have one "red" Vengeance-kit which will and also supplies OC-headroom to speak off...

Always thought this was a Llano problem but after reading your post it looks like I will go elsewhere for new modules which I am planning on purchasing. I am by no means an extreme memory overclocker but what you are describing just sounds awful.

[Ket]

Honestly, there is no word in the English language to describe how poorly things are going right now. I'm no mug, I tested a huge number of system configurations when bug testing the Asrock P67 and Z68 boards, the problem is undoubtedly with the extremely poor quality of the Vengeance memory. I tested no-name memory that performed far better than both of these Vengeance kits I've tried. I may as well let the cat out of the bag now, the Corsair guy while trying to help all he can made a very bizzarre request when talking to him when I had the first Vengeance kit in for testing, he asked if I could bump up BCLK and test the modules at 2000MHz with 10-9-9 (or was it 9-10-9 timings?) Even after I already told him 1866 wasn't even stable with 9-9-9 and 9-10-9 was flaky. With almost identical timings to what I was asked to try anyway, only at a much slower frequency, I questioned what the point in that would be as 1866 wasn't even stable with very lax timings so theres no reason or sence in trying 2000MHz if 1866 isn't even stable. To which I had... no reply on the matter from the Corsair guy. All I'm after right now is for Corsair to give me a kit that does what they claim, thats not asking much surely?

[BeepBeep2]

Any half decent 8GB memory kit makes up for the speed loss, yes. A half decent CHEAP 8GB memory kit would be able to handle 8-9-8 timings around 1.65v at 1866, but have no OC headroom. A half decent 8GB memory kit REASONABLY PRICED would be able to handle 8-9-8 around 1.65v at 1866, with enough OC headroom to touch 2000MHz @ either 8-9-8 or 9-9-9 timings depending. If you buy G.Skill Ripjaws, I'd put money on their CL7 1600 8GB kit doing 8-9-8 @ 1866 with 1.65v.

Ket,
You need Ripjaws 2000 8-9-8 or Flare 2000 8-9-8 to get 2000 8-9-8. (The 2300 9-11-9 kits might do it too...) You are expecting too much. Very few Hynix BFR will do 2000 8-9-8.

A half decent cheap kit is capable of 1866 9-10-9, and half decent reasonably priced kit should do 1866 8-9-8. Only top bin stuff will do much more.

By the way ket, I hinted at using 2T for the Hynix modules, see if it helps...you are being a bit daft imo as far as the overclocking goes. Let go of the frustration and try a different combination of timings.

[Hondacity]

buy some gskills and show us your expected numbers

[Ket]

Ket,
You need Ripjaws 2000 8-9-8 or Flare 2000 8-9-8 to get 2000 8-9-8. (The 2300 9-11-9 kits might do it too...) You are expecting too much. Very few Hynix BFR will do 2000 8-9-8.

A half decent cheap kit is capable of 1866 9-10-9, and half decent reasonably priced kit should do 1866 8-9-8. Only top bin stuff will do much more.

By the way ket, I hinted at using 2T for the Hynix modules, see if it helps...you are being a bit daft imo as far as the overclocking goes. Let go of the frustration and try a different combination of timings.

How many times do you have to be told that 2T timings have already been tried? Its right there in my very first post I've tried countless memory modules of various capacities, the 8GB kits I did test, even the no-name DIY dual channel stuff (IE, two sticks bought at the same time but not tested for DC by the supplier) managed 9-9-9-25 1N @ 1866 1.62v. The Vengeance kit will not even run 1866 @ 9-9-9-25 1N with 1.65v and they are supposed to work with 1.5v at that speed. Even 9-10-9 XMP settings are a bust, stability is very flaky.

Corsair advertise the Vengeance kit as being for overclockers. I didn't expect the memory to OC very well anyway so I already lowered my standards when going into doing the review being satisfied if the modules hit 9-9-9 or better at their rated frequency, which neither kit will do. Corsair are flase advertising, and that is illegal.

[BeepBeep2]

How many times do you have to be told that 2T timings have already been tried? Its right there in my very first post I've tried countless memory modules of various capacities, the 8GB kits I did test, even the no-name DIY dual channel stuff (IE, two sticks bought at the same time but not tested for DC by the supplier) managed 9-9-9-25 1N @ 1866 1.62v. The Vengeance kit will not even run 1866 @ 9-9-9-25 1N with 1.65v and they are supposed to work with 1.5v at that speed. Even 9-10-9 XMP settings are a bust, stability is very flaky.

Skimmed through thread. Thought you recieved Hynix kit like you were supposed to, I now see you didn't.
Of course you tried 2T on the Nanya kit, however the post at the top of this page has 1T written all over it which is why I said that.

No-name stuff isnt binned tight, looks like you got good sticks for the off brand kit... and you should know before buying that corsair saves its good IC's for its top end Dominator GT and GTX series...

[Ket]

I also said the new kit I got is Rev 8.16 like the old kit Corsair either supply what they claim, or their in trouble for their false advertising. All their marketing has to do is not claim things the kits will never do after all.

[Don_Dan]

I could understand your frustration about all this in the beginning and I like your honesty, as the same thing happened to me in the DDR2 era, when Corsair's 6400C4D review samples were built with Micron D9 ICs while my kit ( and any other retails kit I've seen ) were built with ProMOS chips. However, now you're drawing some conclusion that are just plain wrong...

If that 1066 ECC kit can hold 7-8-7 or 8-8-8 @ 1.6v 1600, thats actually not bad and just shows the much more rigorous testing ECC modules go through.

You know that this has nothing to do with "rigorous testing" but with what chips are used on the kit in question.

Although changing chips on these modules is an unpleasant surprise for anyone reading the reviews and deciding to buy a kit, what do you expect from a 1866 9-10-9 kit? In Germany it's listed for 58€, with similar kits from G.Skill and GeIL listed for 52€ and 51€ respectively. Thus it's one of the cheapest DDR3-1866 kits around!
As we all know, DDR3 memory has never been cheaper than now, so it makes sense to me that manufacturers try to cut costs wherever they can, changing to cheaper chips and a weaker bin is the first move to do. I know it sucks that us memory addicts can't fully benefit from the cheap prizes, but faster memory has also significantly decreased in price and you can still get higher speed grade for your money, than, say, 4 months ago.

Any half decent 8GB memory kit makes up for the speed loss, yes. A half decent CHEAP 8GB memory kit would be able to handle 8-9-8 timings around 1.65v at 1866, but have no OC headroom. A half decent 8GB memory kit REASONABLY PRICED would be able to handle 8-9-8 around 1.65v at 1866, with enough OC headroom to touch 2000MHz @ either 8-9-8 or 9-9-9 timings depending. If you buy G.Skill Ripjaws, I'd put money on their CL7 1600 8GB kit doing 8-9-8 @ 1866 with 1.65v.

In Germany the 1600 7-8-7 G.Skill kit is listed for 114€, which is twice as much as the Corsair kit, in other words a 100% price premium. Not sure how pricing is in the UK, but that's what I'm seeing here. I'm pretty sure the G.Skill kit will do much better than the Corsair kit, but it's not really comparable in my opinion as it's more expensive and will very likely use Hynix chips.

So, is your new kit stable at its advertised speed?

[l0ud_sil3nc3]

buy some gskills and show us your expected numbers

this

[Hondacity]

if the xmp specs don't work, it could be asrock bios not tweaked for 4gb / high density sticks.

[badatgames18]

@Ket

not sure if you've already tried or if it will work for the particular modules you are using, but with lower end kits (not sure on the IC) that i've had in the past... i found that they would refuse to post or be absolutely unstable unless i increase tRC to something like= tRAS+tRP+4 or more.
I"ve also found, although it decreases performance by somewhat ... increasing tRFC gave some stability. With my UD4 B3 and kingston greys(2x4gb kit) i had to manually configure every single timing to get 9-11-10 stable @ 2133. I don't know why... maybe it was my motherboard not playing nice, but eventually after hours of googling how to manually configure esoteric timings... i got it fully stable.

[BeepBeep2]

I also said the new kit I got is Rev 8.16 like the old kit

What part of "Thought you recieved Hynix kit like you were supposed to, I now see you didn't." do YOU not understand?...
...
...
I'm misunderstanding a lot but you just repeated what I said I realized.

IMO it was foolish of you to buy a kit that cheap and expect top notch ICs. Binning costs money, the better chips cost money, better brainpower PCB's like Gskill uses cost money...just because you fell for marketing fluff doesn't mean it's the end of the world.

[Ket]

I was saying that I did say I got yet another Rev. 8.16 kit back in my previous posts before you posted. Anyway, expecting 9-9-9 timings at stock 1866 with +15% voltage is hardly expecting "top notch" ICs :/

[PiLsY]

Try 11-11-11-30-2T with the third timing set at 1-4-4-AUTO using 1.575v. Bandwidth sucks but my mates v8.16 sticks will do 2133mhz (2172mhz actually).

[BeepBeep2]

I was saying that I did say I got yet another Rev. 8.16 kit back in my previous posts before you posted. Anyway, expecting 9-9-9 timings at stock 1866 with +15% voltage is hardly expecting "top notch" ICs :/

When I said you didn't get a Hynix kit it was in reaction to you talking about getting a Rev 8.16 Nanya kit again. How else would I have known?
When ram likes TRCD +1, why would you expect 9-9-9? Gotta learn what the IC's like before trying random timings...8-9-8 will go furthest in CL8, 9-10-9 will go furthest in CL9 and 10-11-10 will go furthest in CL10...I can tell just by your posts, not having owned the sticks myself.

This kit is probably similar to GSkill's 2133 11-11-11 kit in performance...

Things you can obviously try-
Scaling at 9-10-9, 10-11-10, 11-11-11...believe it or not SB likes higher clocks a lot more than latency, 2133 10-11-10 or 11-11-11 will probably spank even 1866 9-9-9.

[Formula350]

I've been doing a lot of reading in regards to DDR3, timings and subtimings, which as a result I am going to ignorantly suggest trying: a tREFI of around 900, tFAW of 20 with an RTL of around 50 and a relaxed tRFC. BeepBeep's 9-10-9 point is also valid. Even if what I'm suggesting doesn't sound like it'll help, what do you have to lose?

[Royalk]

Ket can you post a picture with the kit with showing ver 8.16? Thanks.

[TsunamiJuan]

Ket have you managed to get your RMA cleared up with corsair yet?

I took me 3 weeks to mine cleared up, they tried to stick me with a triple stick set of 1866 in exchanged for a dead 2400mhz module.
The last 4 sets of dominators I have gotten aswell seem to clock like these for the cas9 kits of 2133 dominator gt. Its really quite sad.

[chew*]

Like i said before i think peoples expectations are just a tad to high.

This is some low bin hynix according to spec timings ( 9-11-10 2133 )

Without cold this IMC on this chip can't do it and this IMC is known good to 110.3 bclk with 2133 divider........

Realistically on air with high density I have no clue what people are expecting but this is what they should expect......and stop smoking that good stuff. TRCD is not gonna do 9 without the use of good drugs........

[chew*]

If you buy G.Skill Ripjaws, I'd put money on their CL7 1600 8GB kit doing 8-9-8 @ 1866 with 1.65v.

Put your money where your mouth is. I will take that bet, I have the 1600 7-8-7, 2133 9-11-10 and 2133 9-11-9.

None will do what you "think" they should, not at least in anything I would deem stable ( prime 95 blend + games after verifying prime blend )

Only set that might do that is the 2200 9-11-9 and thats well out of the same price range of any of the above, almost double in fact.

Ive tested 8-9-8, 9-9-9, 8-10-8, 9-10-8, 8-10-9 1T or 2T doesn't matter, realistically your looking at 9-10-9 1T @ 1866 and maybe 100 mhz more depending on quality of IC bin.

As far as corsair suggesting try 2000 mhz he probably referred to 2133 divider and lower than 100 bclk, some boards dividers especially asrocks can be busted........as well as there memory support can suck the big one.

[Leeghoofd]

Gonna give the sniper 1600C7 kit a run for it's money Brian, "think" they can do 1866 CL 8-9-8 though on SB stable... but they are on a diff price level then the kits you mentioned...


if not I eat my last pair of socks :p,

oh too bad I don't have any more pairs left.......................................

[bassplayer]

Betcha I can get 1866 8-9-8 out of my Dommies :p