Ryzen 3600X Tested Tight Ram Timings 2133/2400/2933/3200/3733/4000/4200

[gerardfraser]

Ryzen 3600X Tested Tight Ram Timings 2133Mhz 2400Mhz 2933Mhz 3200Mhz 3733Mhz 4000Mhz 4200Mhz 2560x1440 Ultra settings.

There are plenty of 1080P and lower test,these test are aimed at people who blew there cash on $500+ graphics cards and are not quite sure if they should upgrade their ram.I say no if you game at higher resolutions.

Battlefield V I noticed that above DDR4 4000Mhz average FPS started to climb.Which is weird because memory clock/fabric clock/memory controller switch from 1:1:1 to 1:1:2 so memory controller gets cut in half.

A Plague Tale Innocence had two areas where FPS lows dropped to 16 FPS and 7 FPS in the same spot on every test

All games gave the exact same game play every test.

Video of game areas tested

Spoiler

RAM TIMINGS USED FULL NO AUTO

Spoiler

♦ 2133Mhz (16GB)CL10-10-10-10-21
♦ 2400mhz (16GB)CL10-11-11-11-21
♦ 2933mhz (16GB)CL12-14-13-13-26
♦ 3200mhz (16GB)CL14-14-14-14-28
♦ 3733mhz (16GB)CL16-17-16-16-34
♦ 4000mhz (16GB)CL16-18-17-17-36 -(memory clock 2000x2/fabric clock 1800x2/memory controller 1000x2)
♦ 4200mhz (16GB)CL16-18-17-17-36 -(memory clock 2100x2/fabric clock 1800x2/memory controller 1050x2)

COMPUTER USED
♦ CPU - AMD 3600X With MasterLiquid Lite ML240L RGB AIO (Fans 55%)
♦ GPU - Nvidia RTX 2080
♦ RAM - G.Skill Trident Z 16gb DDR4 (F4-4000C18D-16GTZ) (2x8GB)
♦ Mobo - MSI X470 - Gaming Plus
♦ SSD - M.2 2280 WD Blue 3D NAND 500GB
♦ DSP - LG 27" 4K UHD FreeSync Gaming Monitor (27UD59P-B.AUS)
♦ PSU - Antec High Current Pro 1200W

VIDEO INFORMATION
► FPS Monitoring : MSI Afterburner/RTSS
► Gameplay Recorder : Nvidia Shadowplay
► Edit Videos : VSDC Free Video Editor http://www.videosoftdev.com/

 

[MartinPrince]

Dude, well done with the testing though that font is terrible to read.

 

[Davedree]

I think 3600 c16/18 will be the sweet spot

 

[~>Dg<~]

3200 looks the best for price performance though.

 

[tamzzy]

@gerardfraser thanks for this!

could i ask, seeing as we (we as in the forum) also spec for newbies not spending thousands on a new rig...to test 3200c16 and 3000c16 as well please?
(don't really need graphs, just numbers will be fine if it makes it any easier)
would give a clearer indication whether the loss of cas latency would negatively impact performance.

 

[adrichardson81]

That's really useful - thanks! What voltages were you using? I tired to get my 8pack 3200 c14 kit to run at 3600 c16, but it wasn't boot stable even at 1.39v. I was chasing 3600 for the higher infinity fabric clocks, but from your testing it doesn't seem to have much impact on gaming.

 

[drunkenmaster]

3200 looks the best for price performance though.

Yes and no, micron e-die £60-70 for 16GB, 130-140 for 32GB, you can buy either the 3000Mhz cl15 or 3200Mhz cl16 Crucial ballistixs with AES in the code and you'll get micron e-die. Most seems to do 3600Mhz cl16 and 3733Mhz at cl16/cl17, which seems to cover both general limits on the fabric speed, some can get their chip to 3733Mhz 1:1, some 3600mhz, and as was said before the chips launched that 3733Mhz 1:1 was the sweet spot is basically what everyone is finding. Average frame rate almost no gain, hugely higher speeds with non 1:1 doesn't really help, you get some rare games with silly inefficient heavy bandwidth usage but as said not many.

At less than half the cost of Samsung b-die it's a no brainer really. Overall Zen 2 scales with memory bandwidth/latency drastically less than Zen 1 or Intel does performance wise so the absolute best of the best memory is bringing tiny tiny improvements overall.

 

[MartinPrince]

At less than half the cost of Samsung b-die it's a no brainer really. Overall Zen 2 scales with memory bandwidth/latency drastically less than Zen 1 or Intel does performance wise so the absolute best of the best memory is bringing tiny tiny improvements overall.

Which is great so no longer a need for exotic memory to get decent performance out of the CPU. I picked up some brand new Corsair 16GB 3200Mhz for £75.

 

[~>Dg<~]

Yes and no, micron e-die £60-70 for 16GB, 130-140 for 32GB, you can buy either the 3000Mhz cl15 or 3200Mhz cl16 Crucial ballistixs with AES in the code and you'll get micron e-die. Most seems to do 3600Mhz cl16 and 3733Mhz at cl16/cl17, which seems to cover both general limits on the fabric speed, some can get their chip to 3733Mhz 1:1, some 3600mhz, and as was said before the chips launched that 3733Mhz 1:1 was the sweet spot is basically what everyone is finding. Average frame rate almost no gain, hugely higher speeds with non 1:1 doesn't really help, you get some rare games with silly inefficient heavy bandwidth usage but as said not many.

At less than half the cost of Samsung b-die it's a no brainer really. Overall Zen 2 scales with memory bandwidth/latency drastically less than Zen 1 or Intel does performance wise so the absolute best of the best memory is bringing tiny tiny improvements overall.

was just basically going on value. some will think oooh we need 8 pack ram at 130 quid when some £60 ram will be nigh on the same performance in games. witihn a couple of frames. on budget that extra £70 can nearly pay for a 450 motherboard

 

[Davedree]

Yes and no, micron e-die £60-70 for 16GB, 130-140 for 32GB, you can buy either the 3000Mhz cl15 or 3200Mhz cl16 Crucial ballistixs with AES in the code and you'll get micron e-die. Most seems to do 3600Mhz cl16 and 3733Mhz at cl16/cl17, which seems to cover both general limits on the fabric speed, some can get their chip to 3733Mhz 1:1, some 3600mhz, and as was said before the chips launched that 3733Mhz 1:1 was the sweet spot is basically what everyone is finding. Average frame rate almost no gain, hugely higher speeds with non 1:1 doesn't really help, you get some rare games with silly inefficient heavy bandwidth usage but as said not many.

At less than half the cost of Samsung b-die it's a no brainer really. Overall Zen 2 scales with memory bandwidth/latency drastically less than Zen 1 or Intel does performance wise so the absolute best of the best memory is bringing tiny tiny improvements overall.

Do you know if the £40 per 8gig Ballistix Sport LT (BLS8G4D30AESBK ) can hit 3600mhz at cl14/15?

 

[tamzzy]

Do you know if the £40 per 8gig Ballistix Sport LT (BLS8G4D30AESBK ) can hit 3600mhz at cl14/15?

Probably not.

 

[Davedree]

Probably not.

Yeah i'm thinking it's too much to push for, but maybe c16/17 ?
Still for £70 I think i'm going to go with it

 

[tamzzy]

Yeah i'm thinking it's too much to push for, but maybe c16/17 ?
Still for £70 I think i'm going to go with it

3600c16 is doable. various forumers have done it on the e-dies.
of course, as always, YMMV.

 

[Murphy]

You wouldn't find much difference because you changed the column access strobe (CAS/CL) as speeds increased, not saying that's wrong, that you shouldn't have, that the benchmarking is irrelevant, that it could've been done any other way, or anything else negative about the testing. I'm just trying to point out that when looking at RAM benchmarks people need to take into account the correlation between CL and clockspeed.

To put it in perspective these are the real world latencies at each of the tested settings...

♦ 2133Mhz (16GB)CL10-10-10-10-21 (9.37ns)
♦ 2400mhz (16GB)CL10-11-11-11-21 (8.33ns)
♦ 2933mhz (16GB)CL12-14-13-13-26 (8.18ns)
♦ 3200mhz (16GB)CL14-14-14-14-28 (8.75ns)
♦ 3733mhz (16GB)CL16-17-16-16-34 (8.57ns)
♦ 4000mhz (16GB)CL16-18-17-17-36 -(memory clock 2000x2/fabric clock 1800x2/memory controller 1000x2) (8ns)
♦ 4200mhz (16GB)CL16-18-17-17-36 -(memory clock 2100x2/fabric clock 1800x2/memory controller 1100x2) (7.61ns)

Goes without saying those aren't 100% accurate as the other sub timings would've also effected things slightly but give or take a few 0.01ns they should be fairly accurate, it's also worth saying those are only real world access times from when the memory controller issues a request to when the first word is presented on the pinout, subsequent word accesses increase in latency. Also I'm not 100% sure how much is added when switching to the 2:1 divider but i suspect it adds at least one clocktick.

 

[Davedree]

3600c16 is doable. various forumers have done it on the e-dies.
of course, as always, YMMV.

I've decided now and putting it on the order, at half the price I'll risk it.

 

[tamzzy]

I've decided now and putting it on the order, at half the price I'll risk it.

e-die is where the smart money is.
as i've kept saying (like a demented parrot no less)...with >95% performance at <50% of the cost of overpriced b-die gouge ram, it's where the smart money is.
and e-die is not some cheap chinese knock-off either. it's held the memory world record previously.

 

[gerardfraser]

Dude, well done with the testing though that font is terrible to read.

LOL I used what I had.Office 2003 and you should see the font I put in the video ,where I show what part of each game was tested.You will love it.

Thanks for the replies,remember point of video is to show how it is possible to run Ram Kits and do really good in gaming at higher resolutions.It is just that simple.
Added video 4k with a different look

 

[gerardfraser]

@gerardfraser thanks for this!

could i ask, seeing as we (we as in the forum) also spec for newbies not spending thousands on a new rig...to test 3200c16 and 3000c16 as well please?
(don't really need graphs, just numbers will be fine if it makes it any easier)
would give a clearer indication whether the loss of cas latency would negatively impact performance.

Sure CL16 3000 and no difference for 3200

 

[tamzzy]

Sure CL16 3000 and no difference for 3200

thank you ever so much!
i'd buy you a pint, alas all i can but offer now is a virtual one

 

[gerardfraser]

I love beer,thanks.