Hynix CJR vs Samsung B Die

[Krit]

Best result that i can get with stable overclock was DDR4-3800 CL16-19-21-32 @1.42v. So basicley it's slightly faster than Samsung B Die when XMP is enabled for DDR4-3200 CL14 kit and DDR4-3600 CL16 kit. And maybe it's on par with DDR4-3600 CL15 samsung B die kit when XMP is enbled.

 

[oxrufiioxo]

Best result that i can get with stable overclock was DDR4-3800 CL16-19-21-32 @1.42v. So basicley it's slightly faster than Samsung B Die when XMP is enabled for DDR4-3200 CL14 kit and DDR4-3600 CL16 kit. And maybe it's on par with DDR4-3600 CL15 samsung B die kit when XMP is enbled.

3600 CL 16 Bdie seems a little faster. I'm guessing read/copy are just slower period on 1 CCX chips didn't really pay attention to any reviews on that. I know write speeds are way slower.

 

[moproblems99]

Why is @Verbatim write speeds so much lower.

 

[oxrufiioxo]

Why is @Verbatim write speeds so much lower.

1 ccx Ryzen 3000 chips suffer from half speed writes 3600-3800X.

Part of the reason I went with a 3900X although it doesn't really show up in any benchmarks gaming or otherwise I've seen as being a limiting factor.

 

[Krit]

Maybe someone with Ryzen 3600, 3700, 3800 can show numbers ? Ryzen 3900 boosts ram significantly. It's recommended to turn off PBO because that may show different result.

 

[TheLostSwede]

3600 CL 16 Bdie seems a little faster. I'm guessing read/copy are just slower period on 1 CCX chips didn't really pay attention to any reviews on that. I know write speeds are way slower.

View attachment 133707

Keep in mind, he's on a B450 board and there seems to be a slight memory disadvantage on those.

 

[oxrufiioxo]

Keep

Keep in mind, he's on a B450 board and there seems to be a slight memory disadvantage on those.

Yeah, the Benchmark in general isn't a really good representation of performance. I notice more doing CPU limited gaming benchmarks to know if what I did to my ram makes a difference or not. even with a 2080 ti in most games I've tried the gpu isn't fast enough to show a difference past 3200 14-14-14-28.


But the bottom line of this whole thread really is CJR can be really good if you get lucky... B die just takes the luck factor out of it.

 

[moproblems99]

I paid $186 for for 32GB of CJR at 3600 16 19 19 19 39. B-Die was $50 more at 3200 14 16 16 16 32 I think.

EDIT: According to DRAM Calc, I can do 3800 at 16 16 21 21 32

 

[TheLostSwede]

I paid $186 for for 32GB of CJR at 3600 16 19 19 19 39. B-Die was $50 more at 3200 14 16 16 16 32 I think.

EDIT: According to DRAM Calc, I can do 3800 at 16 16 21 21 32

I think you'll find that should be 16-21-16-21, or maybe a bit tighter in the 21's.

Yeah, the Benchmark in general isn't a really good representation of performance. I notice more doing CPU limited gaming benchmarks to know if what I did to my ram makes a difference or not. even with a 2080 ti in most games I've tried the gpu isn't fast enough to show a difference past 3200 14-14-14-28.


But the bottom line of this whole thread really is CJR can be really good if you get lucky... B die just takes the luck factor out of it.

It's not really about luck, but yes B-dies can run tighter timings, but generally costs a lot more and the question then is if you feel the extra cost is worth it or not.

 

[ECC_is_best]

What's the best that is ECC? I've only found Kingston & Samsung ECC for Ryzen so far - and there many Samsung sticks are priced differently, but it isn't obviously why.

 

[AsRock]

Think...
But have you actually tried? If I have the same RAM, although admittedly different hardware for the rest of the system and can run the RAM at 0.5V lower than you, at least it's worth a try, no?
Also, try changing the ProcODT, if you haven't, mine is at 40 something. Could be the UEFI implementation from MSI as well.

I know my Patriot Viper Steal's run hot at 3600, not figured a good way to cool them better (YET). Maybe his are getting hot too ?.

Although what's not helping now too is summer as it's around 30-32c ambient.

 

[TheLostSwede]

I know my Patriot Viper Steal's run hot at 3600, not figured a good way to cool them better (YET). Maybe his are getting hot too ?.

Although what's not helping now too is summer as it's around 30-32c ambient.

Hot? Hmmm, I have had no issues at all with mine, although I haven't used the system for four months now, as it's at home and I'm not...

 

[Assimilator]

Holy thread necromancy Batman.

 

[AsRock]

Hot? Hmmm, I have had no issues at all with mine, although I haven't used the system for four months now, as it's at home and I'm not...

Ok little over the top, getting pretty warm (about 40c), and thought f i was gaming like 3-6hrs how hot they could of gotten. Although i guess benchmarks are going make them warmer than gaming but all so having no way to cool them just in case i decided not to push my luck.

Should of really put my fluke temp sensor their but i would of been a pain to do. Shame my temp sensors broke for my fan controller many years ago.

 

[TheLostSwede]

Ok little over the top, getting pretty warm (about 40c), and thought f i was gaming like 3-6hrs how hot they could of gotten. Although i guess benchmarks are going make them warmer than gaming but all so having no way to cool them just in case i decided not to push my luck.

Should of really put my fluke temp sensor their but i would of been a pain to do. Shame my temp sensors broke for my fan controller many years ago.

Time to aim a fan that way? I have two top mounted fans as intakes, which I guess helps.

 

[AsRock]

Time to aim a fan that way? I have two top mounted fans as intakes, which I guess helps.

Yeah would like to but not really possible with a NH-D15, thankfully i only have to sticks which gives some spacing. Was thinking od some 40x40mm fan just to make sure there's no standing air.

 

[xman2007]

@AsRock You do realise this was last commented in October until the new user bumped it asking about ecc ram and then you quoted @TheLostSwede from October and started having a conversation like no time had passed

What's the best that is ECC? I've only found Kingston & Samsung ECC for Ryzen so far - and there many Samsung sticks are priced differently, but it isn't obviously why.

A new thread would of been the way to go as its not really related to the op.

Post up the links to the different ram you're looking at and maybe someone can answer your question

 

[AsRock]

@AsRock You do realise this was last commented in October until the new user bumped it asking about ecc ram and then you quoted @TheLostSwede from October and started having a conversation like no time had passed



A new thread would of been the way to go as its not really related to the op.

Post up the links to the different ram you're looking at and maybe someone can answer your question

hehe i think the forum can handle it, other wise my bad.

 

[Taraquin]

Really old tread, but did some tuning with my Micron rev E.
3733-16-19-8-11-35@1.35V 1t 525tRFC GDM-off got me to 55200 read and 67.1ns.
3733-15-20-8-11-35@1.43V 1t 520tRFC GDM-off gave me the exact same score.

On rev E tRCDRD must be high, but tRCDWR can go really low, which helps a bit. tRP can also be really low.

Compared to the dram calc fast 3733 I could lower tRCDWR from 16 to 8, tRP from 16 to 11, tRAS from 36 to 35 (21 work but gives no perf increase), tRC from 58 to 55, tRFC from 545 to 525. All in all these tightenings improved fps in certain games by around 2%, aida64 by 1ns and got a bit higher read and copy.