Corsair Vengeance RGB PRO DDR4 4000 MHz

[cadaveca]

So in simple terms the argument you just made is that the performance in overclocking of these sticks is utterly irrelevant, despite the fact you've spent an entire thread arguing that you can back up your claims about overclocking, mentioned it and features relating to it on multiple pages of the review, listed it as a pro point in your conclusion and spent an entire paragraph of said conclusion extolling the sticks overclocking virtues.

Instead, you would like us all to believe that you never had any intention of making any claims about these things. Immediately after making some big claims about these things.

What's really sad about this is how completely your own first post in this thread contradicts you, glossing over the iCUE integration in order to extol how these sticks are for "Serious overclocking too!" (Exclamation point your own).

Exactly, yes. Why is that a problem? All of those claims, that seems unproven to you, are all true. It's not my job to prove everything, like I said. I will rely on my fellow reviewers to do that for me.

I want to take a moment and say thanks too, because clearly this is important to you. That's great.

However, many people complain about our scoring system. This is why it doesn't makes sense to some...they are picking one or two factors about a product to determine their overall score rather than looking at everything as a whole. It's really all about the experience these things offer you. Anyone can run benchmarks and post results and claim whatever. That's not what I'm doing here. Until you can understand that, I understand why you feel the way you do. That's cool. Sorry, but it is what it is.

the way I see it, if I take your approach, then companies should only ever make one, product, and it needs to be the best. If it's not, then it sucks. That's not how the world works though.

 

[damric]

Memory tuning is by far the most difficult aspect of overclocking, and to reach the maximum headroom of a kit can take many, many hours of stress testing. I've literally spent weeks tuning some kits, so I can see the hesitation about fiddling.

But Dave, pretty please spend an hour or so overclocking these RAM kits in your reviews. Even if you just test one voltage setting. You should be using the same platform and CPU for all of your kits anyway since memory controller affects latency and bandwidth performance, so no need for testing with dozens of CPUs. Just give us a quickie of what the best bench stable frequency and CAS timing you can get at XMP volts without trying too hard.

 

[cadaveca]

Memory tuning is by far the most difficult aspect of overclocking, and to reach the maximum headroom of a kit can take many, many hours of stress testing. I've literally spent weeks tuning some kits, so I can see the hesitation about fiddling.

But Dave, pretty please spend an hour or so overclocking these RAM kits in your reviews. Even if you just test one voltage setting. You should be using the same platform and CPU for all of your kits anyway since memory controller affects latency and bandwidth performance, so no need for testing with dozens of CPUs. Just give us a quickie of what the best bench stable frequency and CAS timing you can get at XMP volts without trying too hard.

Makes sense when reviewing an individual kit and not the product line for sure. What you are asking for I don’t call Overclocking either but that is truly semantics. It doesn’t provide meaningful info, to me, unl as you are testing many kits and posting the average. Like I said, there are some reviews that have that but I do test every kit far beyond that. I’ve posted the screenshots in the past when requested too.

 

[emissary42]

But Dave, pretty please spend an hour or so overclocking these RAM kits in your reviews. Even if you just test one voltage setting. You should be using the same platform and CPU for all of your kits anyway since memory controller affects latency and bandwidth performance, so no need for testing with dozens of CPUs. Just give us a quickie of what the best bench stable frequency and CAS timing you can get at XMP volts without trying too hard.

I second that. Would already be a lot more interesting for me than no tests beyond XMP at all

Also a Thaiphoon SPD screenshot would be nice, for any future memory reviews too.

 

[cadaveca]

Also a Thaiphoon SPD screenshot would be nice, for any future memory reviews too.

Done. I've got several reviews already done and waiting, and for a couple of those I'll add in the Taiphoon screenies, but not all as I gotta swap rigs as I'm on to another testing phase with a different rig ATM


I will say again, OC testing without having several kits and averaging the results doesn't tell anyone anything except what the one kit I got can do, IMHO. Add in cherry-picking, and I see absolutely no reason to include it. I do do that testing, because its fun, but not a review priority at all right now. Separate review for memory OC in progress, so I hope that's enough.

 

[emissary42]

Thank you

I think no one was demanding of you to give us a stastitical average of what overclock to expect from this kit (or any other). And most of us are already aware that samples from manufacturers are cherry picked more often than not. However just showing if this kit is able to for example run DDR4-4000 16-16-16 1T or DDR4-4133 17-17-17 with tight subs below 1.4V and the performance difference that makes versus the basic XMP, would have probably been interesting for your readership regardless. And again it is just a humble suggestion, no one wants you to force changing your review style or general workflow.

 

[cadaveca]

Thank you

I think no one was demanding of you to give us a stastitical average of what overclock to expect from this kit (or any other). And most of us are already aware that samples from manufacturers are cherry picked more often than not. However just showing if this kit is able to for example run DDR4-4000 16-16-16 1T or DDR4-4133 17-17-17 with tight subs below 1.4V and the performance difference that makes versus the basic XMP, would have probably been interesting for your readership regardless. And again it is just a humble suggestion, no one wants you to force changing your review style or general workflow.

Of course, the sticks do do 4000 C16 without an issue at all, @ 1.35V. Should 4000 MHz sticks all do this? or should it be more around 1.4-1.45? To be honest, I was only really expecting C17 @ 4000. Getting C16 was a red flag to me, but maybe I'm wrong about that and Corsair is just binning these really well. Either way, nice sticks for clocking for sure, LEDs or no.

And yeah, I mean, I think you and I kinda see things the same way, so I wasn't really directing that 2nd comment at you, but anyone who might read.

 

[basco]

i think this 4000c16 is better then average with good volts too.
but you are right about binning ram+cpu+mainboard(yes motherboards too if ya want hwbot memfrequenzy records) its getting ridiculous.

hard times for a memory tester-thanks for your effort!

 

[emissary42]

Of course, the sticks do do 4000 C16 without an issue at all, @ 1.35V. Should 4000 MHz sticks all do this? or should it be more around 1.4-1.45?

Depends on how well they are binned. I had worse, even higher rated ones. So I'd say it is a pretty good result, especially considering their far from exciting specifications.

While other kits might need a little bit more voltage than that, they could still scale really well frequency wise and/or be able to hit really low timings at extreme voltages. And yes, while scaling graphs are very useful they also are a lot of work. So I don't blame anyone who decides against making them part of their reviews to keep their sanity

To be honest, I was only really expecting C17 @ 4000. Getting C16 was a red flag to me, but maybe I'm wrong about that and Corsair is just binning these really well.

I don't think you need to worry about flags popping up because of that. It is only an issue for me (and probably most others), if someone doesn't state using a sample from the manufacturer while trying to sell it as a retail review.

And yeah, I mean, I think you and I kinda see things the same way, so I wasn't really directing that 2nd comment at you, but anyone who might read.

It is all good.

The performance numbers for 4000 16-16-16 1T seem a bit low to me. However I don't know how much the i5 and its lower frequency contributes to that and only have results with my 7700K to compare it to. A tRFC of 700 also does not exactly hint at tweaked sub-timings, so the values might actually be in the expected range.

(ASRock Timing Configurator is a nice tool to display the relevant timings in your screenshots and also works with Non-ASRock motherboard)

 

[cadaveca]

So I don't blame anyone who decides against making them part of their reviews to keep their sanity

The silly part is that I still do all that testing, looking at scaling, because its interesting to me? ROFL. I don't mind the time spent since I have multiple rigs and it's not a big deal to hit buttons... but there WAS a point where I had those graphs in my memory reviews... almost a decade ago!


Like you know, we have 17-19-19 as a base spec for 4000. It's an automatic thing to set up those timings manual with subs and then see what's what.

As to the bandwidth, yeah, low, because yeah, not fully tweaked. That was just the qualifier run (boot, test XMP, boot test 17-19-19, boot, 16-16-16, and then the 15-16-16 too see if something crazy is going on and still haven't found THAT kit yet) and I've got the screens of the tweaked sub-timings too, but those are for later.

Timing Configurator is GREAT!!! I test on ASRock boards for X299/X399, but ASUS for Z370. I prefer timing configurator over memtweakit, anyway. So there are options for sure.

 

[las]

I'd never buy CL17 or above.
Sweet spot is 3200/CL14 or 3600/CL15-16.

 

[Mussels]

I've got the 3200Mhz variant of this RAM, and it overclocked to 3800Mhz on my first attempt (i think my IMC was crapping out, as it passes memtests but goes unstable at idle)

Mine turned out to be Hynix A die, but due to getting good clocks on ryzen anwyay i'm not worried about that

iCue was glossed over, as part of the big appeal for their RGB implentation is that its software linked to *all* corsair RGB products - when i type on my keyboard, or speak into my mic - the ram can react, or be programmed to respond to various other stimuli within the corsair ecosystem (For example, i can set the RAM sticks RGB to temperature of the ram itself, my H115i water cooler, or as a reaction to a corsair mouse/keyboard/headset. Not every device supports every reaction yet, but they're being added in at a fast pace since iCue itself is quite new)

Theres even a giant ad in the review screenshots showing farcry 5 and the game integration thats missed, where this RAM can have its lighting controlled by external software - you catch fire in farcry 5, your ram changes an animation to match. It may be totally pointless, but holy shit is it awesome having an entire synched RGB setup

I maaaaaay be a corsair fanboy for the fact that my external and internal hardware can be synced together.

 

[cadaveca]

Mine turned out to be Hynix A die, but due to getting good clocks on ryzen anwyay i'm not worried about that

Thanks for highlighting why I made no mention of the type of IC in the reviewed sticks... because not all these Vengeance RGB PRO sticks use the same IC!!!

iCue was glossed over, as part of the big appeal for their RGB implentation is that its software linked to *all* corsair RGB products - when i type on my keyboard, or speak into my mic - the ram can react, or be programmed to respond to various other stimuli within the corsair ecosystem (For example, i can set the RAM sticks RGB to temperature of the ram itself, my H115i water cooler, or as a reaction to a corsair mouse/keyboard/headset. Not every device supports every reaction yet, but they're being added in at a fast pace since iCue itself is quite new)

Yeah, for me the iCue software integration is really the reason to get these sticks. Clocks, timings, whatever.... ah, whatever... but that software, and how it all works with everything sure is fantastic, and if you have other Corsair goodies that support iCue, I really think you'd be missing out on a lot of you had any other ram. Looking at your screenshots, you are the perfect example of this, so thanks for taking the time to show that!

 

[GlacierNine]

Thanks for highlighting why I made no mention of the type of IC in the reviewed sticks... because not all these Vengeance RGB PRO sticks use the same IC!!!

Which of course makes my point perfectly - and in fact I even made this same point myself much earlier - because the ICs aren't guaranteed, the overclocking headroom isn't either, which is exactly why your review shouldn't be making claims about overclocking headroom that aren't substantiated.

But thank-you for reviving the issue only to tell me what I already told you.

With that said though - the reason they do that in such a coded way is so that they can if needed change the IC used with an inferior quality one that still meets quoted specs. Its a meaningless "statement" about the chips over clocking potential once you bear that in mind.

 

[cadaveca]

But thank-you for reviving the issue only to tell me what I already told you.

Um, perhaps you failed to read @Mussels post that states his 3200 MHz sticks are doing 3800 MHz.… That's a considerable 600 MHz (~18%) overclock, no? Corsair has binned these sticks to have overhead, confirmed by @Mussels RETAIL set...

Mine turned out to be Hynix A die, but due to getting good clocks on ryzen anwyay i'm not worried about that

So what now? He's getting those clocks on Ryzen, even...

Corsair says:

each module is built using carefully screened ICs for extended overclocking potential.

Users trying to clock these sticks with boards that do not support extended memory frequency can't blame the sticks... it's the motherboard at fault.

 

[Mussels]

hell i can get random posts above 3800 as well, my IMC just doesnt seem to like it (or i need to tweak the timings more) as i pass tests, but get random crashes at idle.

For now out of laziness i've settled for 3533 fully stable at low voltages and heat, but yeah - they overclock.