T-FORCE VULCAN & G.SKILL Flare X AM4 Overclocking @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
I have 16g (2x8) G.Skill Flare X 3200 in my 2700x rig and my 2600x rig and both sets are solid and well worth the $$.
 
Yep, G.Skill Flare X 3200 in my 2700X newly minted build as well. Very happy with it so far.
 
I have The G-skill Flare x 3200 16GB in my ASUS TUF B3550M-plus gaming.

My experiences have been a little mixed.

Up until and including BIOS 3401, it was just a question of setting the XMP profile, and everything would be good at 3200.

With any of the later BIOS'es, I have not been able to get stability higher than 3066, using the XMP profile.

I think I could find stability if I tuned timings manually, but this is something that should just work, and doesn't.

Otherwise I'm quite happy with the Asus board, but the memory stability has been a letdown.
 
I have The G-skill Flare x 3200 16GB in my ASUS TUF B3550M-plus gaming.

My experiences have been a little mixed.

Up until and including BIOS 3401, it was just a question of setting the XMP profile, and everything would be good at 3200.

With any of the later BIOS'es, I have not been able to get stability higher than 3066, using the XMP profile.

I think I could find stability if I tuned timings manually, but this is something that should just work, and doesn't.

Otherwise I'm quite happy with the Asus board, but the memory stability has been a letdown.

I don't think you can fault the memory in this case. Seems obviously a motherboard BIOS issue. I'd flash back to the previous BIOS version if all was good.
 
I wanted 16GB modules (2x16GB). Even though my sticks are Corsair LPX 3200 modules they are for the Intel processors (not AMD). Even the XMP profiles won't post on my 1700 with the latest BIOS for my x470 ASUS motherboard. I can get 2666 out of them and that's what I'll have to live with (since I wanted 16GB modules).
 
Please forgive my ignorance, but what is it about this RAM that makes it "for the AMD Ryzen AM4 platform"? Is it not able to be used in an Intel system?
 
At DDR4 3400MHz, things started to unravel. I had to loosen the timings and start throwing voltage at the memory in order to achieve stability. At 18-19-19-39 @ 1.5v
That's some really lose timings for ram that should be able to do 3466c14 easy with that kinda V or 4000C16 on a Intel system.
Did you try increasing the V to the memory controller? SOC ~1.2v

Setting the sub timings right is of high importance to getting the highest frequency out of RAM with Ryzen especially on MB with shoddy BIOS that don't get it right when left to Auto

The Rzen RAM calculator can help make it easier for those new to overclocking RAM on Ryzen.
https://www.overclock.net/forum/13-...lator-1-1-0-beta-2-overclocking-dram-am4.html
 
Last edited:
Please forgive my ignorance, but what is it about this RAM that makes it "for the AMD Ryzen AM4 platform"? Is it not able to be used in an Intel system?
It works with both but the sub timings have been tweaked for higher compatibility\performance with first gen Ryzen.
 
I will add, that I've sold tons of the FlareX 3200 16GB (2X8) kits and saw a huge amount of them return as faulty. That said, the Trident Z RGB 3200 2x8 kits I've sold have been amazing and rarely comes back faulty.
 
I will add, that I've sold tons of the FlareX 3200 16GB (2X8) kits and saw a huge amount of them return as faulty. That said, the Trident Z RGB 3200 2x8 kits I've sold have been amazing and rarely comes back faulty.
Were they actually faulty? or simply not compatible with the AMD CPU\MB at XMP vs Trident Z which may be more commonly used with Intel which can handle much higher memory speeds more easily.
 
I bought a couple kits of the T-Force Dark 3000 when on sale at Newegg for $139 each (16gb 2x8). Zero problems in either of my Ryzen boards (Asrock AB350+R1700 and X470+R2700x) using xmp profile, 3000 mhz speed. Will do 3200 pretty easy.

I totally recommend them when on sale, and I would buy them again. At $170, maybe not so much. The heatsinks are kinda cheesy, but I don't care.

In the past I have tried to stick with Corsair and Kingston, but their ddr4 stuff is high. I've done Patriot here and there because it always seems to be on sale at Fry's when I am in a pinch.

https://www.newegg.com/Product/Product.aspx?Item=N82E16820313807
 
Its interesting, Ive been coming here for a good long time now and I don't ever recall any strictly memory based reviews.

Now I can read about headphones , gaming chairs , memory, watercoolers.........the dearth of GPUs has changed the landscape.

I miss the case reviews.
 
Nice write up and good info. It seems that 3200 tight timings are a nice sweet spot.

I have 2 , 2x8 kits of Team Dark Pro 3200 cas14 and both run at the rated speeds for me. One kit is in the Asrock B350 itx with a R5 1400 and the other is in my sig rig whch is a 2600 on the CH7. Oddly enough the CH7 needs 1.4v to stay at 3200 c14 stable. XMP all day on the Asrock board and 1400. I could only get 2933 c14 on either kit with a 1600 I had. IMC silicone lottery at play there me thinks.

I paid 179 for one kit and 189 for the other months apart. Right now its high at 209 sadly

https://www.newegg.com/Product/Prod...12&cm_re=team_dark_pro-_-20-313-712-_-Product


Do not own but here is the 3466 kit

https://www.newegg.com/Product/Prod...26&cm_re=team_dark_pro-_-20-331-226-_-Product
 
Great way to benchmark memory is to measure FPS in Fallout 4. That game is limited by RAM speed. I think that if I had DDR4 at 4600 instead of 3200, I'd have no reprojection in VR.
 
  • Like
Reactions: Meeho
like this
Had you tried tightening the timings instead of increasing the ram multi? Sometimes ram doesn't work well at a higher speed, but can manage slightly tighter timings. Otherwise, I've found that bclk overclocking works well to reach a higher ram speed, but it also overclocks the cpu and pci bus so it's not as fair a comparison, and can cause instability in other areas (making troubleshooting a bit more involved).
 
I wanted 16GB modules (2x16GB). Even though my sticks are Corsair LPX 3200 modules they are for the Intel processors (not AMD). Even the XMP profiles won't post on my 1700 with the latest BIOS for my x470 ASUS motherboard. I can get 2666 out of them and that's what I'll have to live with (since I wanted 16GB modules).

You need modules specifically for Ryzen. Don't know why, but that's just the case. Lots of incompatibility problems with RAM that isn't. One of the reasons why I chose the Flare X, it just works good with Ryzen.
 
In general getting b dies such as the flair x is much easier for sure. Also more expensive.

2400/2666 kits of almost anything seem to work on all the builds I’ve done.

Also many other kits are working well now at higher speeds. Some need more tweaking obviously.

I set up an x370 Taichi build with a 1700 and 16gb of GSkill ripjaw V 3200 c16, which is a Hynix kit. Works at rated speed at 2933 all day at xmp on several boards. Now at 3200 xmp with newer bios for AsRock and MSI boards I’ve tried.
 
You need modules specifically for Ryzen. Don't know why, but that's just the case. Lots of incompatibility problems with RAM that isn't. One of the reasons why I chose the Flare X, it just works good with Ryzen.
Oh.. I'm aware of that. I was just wanting 16GB modules. I tried... Didn't get there. I'll take 2666 though. A LOT better than the 2100 it defaulted to. :)
 
Oh.. I'm aware of that. I was just wanting 16GB modules. I tried... Didn't get there. I'll take 2666 though. A LOT better than the 2100 it defaulted to. :)

Ah, didn't realize getting 16GB modules was a problem. I was only shopping for 2x8 kits.
 
I don't think you can fault the memory in this case. Seems obviously a motherboard BIOS issue. I'd flash back to the previous BIOS version if all was good.

I am aware of it, and thats also why I mentioned the BIOS versions. My point was that the Ryzen compability sticker doesn't guarantee compability, not even in brand name boards.

Downgrading is a pain, as the BIOS does not support going to older versions, so you have to use special software for the downgrade.

I am rather disappointed by Asus because of this, but I'll live with it, as I'm buying a 2700x to replace my 1700 in this board (and the 2700x needs a newer bios than 3401).
The 1700 will live on in another build :)

And perhaps I'll have better luck with the new CPU, and mem speeds...
 
I have to admit, RAM and RAM overclocking is one of the aspects of our builds that I am least well read on.

I'm currently building a Ryzen 5 2400G APU for my fiance. She doesn't game or do anything fancy on her system, but I still wanted to make most of the system, and I've read that RAM speed is crucial to APU performance.

Since the ASRock x470 motherboard I bought supports DDR4 3466 OC modes, I bought a 2x8GB set of RAM sticks that claims to support it. Best price was for this T-Force Dark kit. We'll see how it goes...
 
Last edited:
If you're buying gamer bling ram for zen, 3200C14 or bust, nuttin' but fine b-die there. (includes anything better like 3600C15 etc)

XMP on amd is still going to be a maybe at best, so [H] up and do your own timings.
 
You need modules specifically for Ryzen. Don't know why, but that's just the case. Lots of incompatibility problems with RAM that isn't. One of the reasons why I chose the Flare X, it just works good with Ryzen.

Hmm.

I did not realize this.

My first Ryzen based system was a first gen hexacore I built for my stepson when his old FX-8350 gave out. I just got whatever cheap RAM I found, some Mushkin DDR4 2400. Seems to have worked OK, but I haven't tried overclocking. He doesn't need it for what he does.

As I mentioned above, for my current build for my fiance in a Ryzen 5 2400G APU I went with the T-Force Dark quite accidentally. I had no idea it was a "designed for Ryzen" kit until right now when I googled it.
 
I really like my FlareX running 3466 cas14 on an ASRock B-350 itx. Still, the T-Force looks to be a better value.
 
I'm still skeptical if ram overclocking is worth a damn. Even in these dedicated memory bandwidth tests the gains are negligible. I wish there were real world tests to see if it is worth the time to fiddle with it.
At this point I'm happy if a ram works at rated speed without me having to manually tune timings.

Never heard of T-FORCE but it seems to do what it says so I guess good value ram.
 
Hmm.

I did not realize this.

My first Ryzen based system was a first gen hexacore I built for my stepson when his old FX-8350 gave out. I just got whatever cheap RAM I found, some Mushkin DDR4 2400. Seems to have worked OK, but I haven't tried overclocking. He doesn't need it for what he does.

As I mentioned above, for my current build for my fiance in a Ryzen 5 2400G APU I went with the T-Force Dark quite accidentally. I had no idea it was a "designed for Ryzen" kit until right now when I googled it.

If he is gaming at 1080P, and to a lesser extent, 1440P, you are leaving performance on the table keeping his ram at 2400. I can't recall for certain but I remember you said he had a fairly decent GPU, like ~1060 ish kinda power? Try getting the ram to at least 2800, it will give him a decent boost in minimum FPS, and help his frame times, especially at 1080P.

As for the ram you picked for your Fiance (how do you have a stepson and a "Fiance btw? We need to take this to GenMay so you can reveal your plans at marrying multiple people :-D ) , if you use the latest BIOS you should be able to at least get it to work at 3200. Those are Hynix based sticks IIRC, so you want the latest BIOS and you want to use the Ryzen Ram calc to set your timings manually. People give Hynix sticks a bunch of crap, but with the latest AGEA code, they seem to work just fine @ speeds up to 3200.

I have a 3200 C16-18-18 kit I was running with my old 1600 @ 2933 C14 across the board with no issues @ 1.4V and a slight SOC Vcore bump. With the move to x470 and the latest BIOS, I can do 3200 C16 across the boards. Haven't tried anything higher yet, I just started playing with the Ryzen ram Calc. I had the wonderful luck of G.Skill sourcing multiple ICs for the ram I have. It can either be Samsung D Die, or Hynix so that, along with the fact I could do 2933c14, has kept me from messing with it a whole lot in the past.

I bought it in January, and I just didn't see $65+ worth of extra performance to go with a B die kit at the time. Now that they can be had on sale for much cheaper prices, it makes it a better choice.
 
I'm still skeptical if ram overclocking is worth a damn. Even in these dedicated memory bandwidth tests the gains are negligible. I wish there were real world tests to see if it is worth the time to fiddle with it.
At this point I'm happy if a ram works at rated speed without me having to manually tune timings.
2933 vs 3400 on a 2600 vs a i5 8400 with ram at 2666 in 36 games.
https://www.techspot.com/review/1627-core-i5-8400-vs-ryzen-5-2600/

On a i7 6700k I have tested photoshop and lightroom and 2133 vs 3866 can bring up to 20% performance increase depending on the task
while games see up to 30% increase in CPU performance which combined with a CPU OC adds up to ~40% in total.
3200c14 to 3866c16 is ~7%
 
I ended up killing my old motherboard (water leak)
So instead of buying a new processor, I ended up with an "ASUS TUF B450-plus gaming". Had to cancel the order for the processor, as I cant afford both right now.

Basically this is the updated version of my old board, and I chose the full ATX version instead of mATX.

The board is improved on many points. Better VRM colling, more fan connectors, and so on.

But the best thing is, now the memory works flawlessly, at stock speeds. I'm even running it at 3333 right now, completely stable. 3400 makes errors, 3466 wont even boot.

And the CPU seems more reliable when OC'ing, needing less voltage to get to the same speed.

The CPU wont be updated for now, but this is working so well, that perhaps I could hold out until Ryzen 2 gets released.
 
Last edited:
They also OC-ed the cpu in that test, there is no way to tell how much or little ram speed matters from the graphs.
I would estimate close to 50\50 maybe a bit less. Some games will benefit more than others.

These are some of the tests I have done. Be nice if I had Ryzen to play with and maybe some dual rank ram or four single rank.
https://www.overclock.net/forum/180...0c11-2133c9-ddr4-2133c15-3000c12-4000c17.html
2871260.png
 
  • Like
Reactions: M76
like this
I would estimate close to 50\50 maybe a bit less. Some games will benefit more than others.

These are some of the tests I have done. Be nice if I had Ryzen to play with and maybe some dual rank ram or four single rank.
https://www.overclock.net/forum/180...0c11-2133c9-ddr4-2133c15-3000c12-4000c17.html
View attachment 100814

So I'm going to assume you selected the game where it benefits the most? Of course I'll ignore the downclocked results. You didn't OC a 2133 RAM to 4000, right? I'm not saying don't do it, if that's your hobby, and it's fun to you. I'm still not going to waste hours to fine tune a dozen ram timings for a few fps more at the top end. Even if there are a few games where you can gain 50/50.

BTW I know it's hoping for too much, but have you seen where someone tested how much memory speed matters in cinebench for example? Because if you can gain 50/50 there as well then consider me converted.
 
I was using 3200c14 RAM which defaults to 2133c15 till you enable XMP
3600C15 RAM doesn't cost a huge amount extra for those that cant be bothered overclocking although it probably wont work without tweaking on AMD it's good for coffeelake

ARMA is a CPU bottlnecked game and open world. Two things that will increase the demand for higher RAM speeds.
Most games see large gains around this but not all I know Crysis 3 benefits more from overclocking the CPU than the RAM unlike most games that prefer RAM overclocking

RAM speed makes no difference to video rendering tasks like cinbench maybe 1% tops.
This is probably due to it being a orderly process the cache has what it needs ready to go and so the CPU doesn't have to wait on the RAM.
But it can help photo editing.
https://www.pugetsystems.com/labs/articles/Puget-Systems-Adobe-Photoshop-CC-Benchmark-1132/
eb106f9873cf97cffa0485df9a52c910285bd53c020a38fe4e05a14be2e90957.jpg
 
  • Like
Reactions: M76
like this
Every b die kit I've owned does 3466 cl14-14-14-14 at 1.45.. these are all gskill kits a trident z 3200 kit cl 14, triz 3600 cl15 kit, and the flarex 3200... All the same, all b die...I would like to think the article could have been a bit more thorough in this respect given the huge price break for the team kit...

Is there going to be a follow up?
 
Every b die kit I've owned does 3466 cl14-14-14-14 at 1.45.. these are all gskill kits a trident z 3200 kit cl 14, triz 3600 cl15 kit, and the flarex 3200... All the same, all b die...I would like to think the article could have been a bit more thorough in this respect given the huge price break for the team kit...

Is there going to be a follow up?

Following your post I tried setting mine to 3466 / 1.45, with the 3200 DOCP setting. The computer boots, and I can work on it for a few minutes, RAM test 1.1.0.0, never gives errors. But eventually the PC freezes up.

Turning speed down to 3400 makes the PC stable, so still higher than what I could do before.

Do you do anything special to make it stable at 3466?
 
Funny you should say that I just got the gskill rgb 3600 kit and once again 3466 without too much worry..I do set the procst to 60 ohm if that helps and 1.45 might be to much or too little this kit only needs 1.4 but I have it at cl14-14-14-14 1.475 3533
 
Mine is of course only a 3200 kit.

I have tried setting Procodt to 60, and 1.475V @3466.

It does seem to have helped, but I'll have to run for some time, to see if its reliable / stable.

Edit:

Nope, didn't work, still get memory errors / crashes after a few minutes.

3400 seems to be where I'll be at for now. To much trouble for small gains when I try to go higher :)

But thanks for the advises, I'll probably try to work some more with it later on.
 
Last edited:
Back
Top