Overclocking B-die on Ryzen 5 2600

[VuurVOS]

Try to loosen tRCDRD i found that going under 16 gave problems at my settings

15 is enough in his case and also already proposed

 

[Deleted member 185158]

I don’t have a dog In this fight I just OC my 3600 cL16 to 3800 CL18 and literally just changing the CAS. I refuse to go down the timings rabbit hole.
Anyways AIDA64 is the “go to” test. What is your alternative?

Anything you'd might consider accurate would be downloaded from hwbot.

And you would use Benchmate on systems running W8 or newer because of RTC inaccuracies.

SuperPi 32m is a memory benchmark.
PiFast is even more accurate and a bit more sensitive than PiMod.

 

[ThrashZone]

I don’t have a dog In this fight I just OC my 3600 cL16 to 3800 CL18 and literally just changing the CAS. I refuse to go down the timings rabbit hole.
Anyways AIDA64 is the “go to” test. What is your alternative?

Hi,
Why would you raise your cas latency to 18 for 200mhz ?
You're better off with 3600c16 latency frankly.

Aida64 is nothing more than a quick latency check which you should of noticed your latency going up not down switching from c16 to c18.
Stress test is a freaking joke.

 

[INSTG8R]

Hi,
Why would you raise your cas latency to 18 for 200mhz ?
You're better off with 3600c16 latency frankly.

Aida64 is nothing more than a quick latency check which you should of noticed your latency going up not down switching from c16 to c18.
Stress test is a freaking joke.

And believe. It or not it gets lower latency. I tested it multiple times ¯\_(ツ)_/¯

 

[Deleted member 185158]

Hi,
Why would you raise your cas latency to 18 for 200mhz ?
You're better off with 3600c16 latency frankly.

Aida64 is nothing more than a quick latency check which you should of noticed your latency going up not down switching from c16 to c18.
Stress test is a freaking joke.

The way I try to describe latency....

Is nano seconds. Billionths of a second.

70 can't tell 65 to 70ns difference without leaning on a program to tell it to you.

Make your super Pi mod time faster on timings alone, no frequency change if you want challenges.

Then test that with Ycruncher.

 

[ThrashZone]

And believe. It or not it gets lower latency. I tested it multiple times ¯\_(ツ)_/¯

Hi,
You'd see better results at c16 and higher frequency
Only difference is higher dimm voltage which lower cas latency always uses more.

4000mhz with c16 timings 16-16-16-36 only price is close to 1.45v but what ever enjoy c18.

 

[INSTG8R]

Hi,
You'd see better results at c16 and higher frequency
Only difference is higher dimm voltage which lower cas latency always uses more.

4000mhz with c16 timings 16-16-16-36 only price is close to 1.45v but what ever enjoy c18.

I’ve already had the argument, tested both, CL18 3800 won. /end

 

[Deleted member 185158]

I’ve already had the argument, tested both, CL18 3800 won. /end

Bandwidth over latency, unless bum memory controller.

And that's why on AMD and high end Bdie, 4600mhz it's easy to give up gear 1. More bandwidth.

 

[ThrashZone]

I’ve already had the argument, tested both, CL18 3800 won. /end

Hi,
I'll just chalk that up as a amd memory quirk seeing amd latency is already stupid high according mostly to how aida64 translates it opposed to how it translates intel systems latency results even intel's quad channel boards results.

 

[freeagent]

My old ddr3 Intel systems have lower latency than any modern amd cpu I have seen

 

[Deleted member 185158]

My old ddr3 Intel systems have lower latency than any modern amd cpu I have seen

I can't remember what my 6400+ hit. Around 4ghz, mid to lower 30's... but I don't think I saved any screen shots of that during suicide runs.

 

[HD64G]

Those are 2 sets of settings, 1 safe and 1 fast to test. I would test the fast 1st but if not stable go for the safe one. Zen+ aren't easy to go beyond 3466 without much trial and error efforts.

 

[chrcoluk]

My 2600X couldnt get over 2667mhz 4 dimms used.

 

[VuurVOS]

My 2600X couldnt get over 2667mhz 4 dimms used.

You said it yourself. 4 dimms used. With 2 dimms it should be able to run at 3466MHz.

 

[Deleted member 185158]

View attachment 259740View attachment 259743View attachment 259743View attachment 259741
Those are 2 sets of settings, 1 safe and 1 fast to test. I would test the fast 1st but if not stable go for the safe one. Zen+ aren't easy to go beyond 3466 without much trial and error efforts.

Hey would you look at that. Found the screeny.

Fast and tight on Zen+. Not daily stable though.... but benchable.

 

[Zach_01]

Gear Down Mode. If you want to run an odd number Cas latency. Otherwise the board will default to an even number 18 opposed to 17, otherwise it does nothing.

Don't tell me you guys SERIOUSLY use shit like AIDA64 for "ACCURATE" comparisons.

If that's the case, I'll back away slowly and cautiously.

Well, I guess my crappy(?) question has got me what I didn’t care for…
I know what GDM is and what it does.
And we should all know by now that AIDA results are just indications of bandwidth/latency and not real application performance. Apps can benefit on both or more on one against the other.

I also guess now there is no straight answer

Anything you'd might consider accurate would be downloaded from hwbot.

And you would use Benchmate on systems running W8 or newer because of RTC inaccuracies.

SuperPi 32m is a memory benchmark.
PiFast is even more accurate and a bit more sensitive than PiMod.

Would those benefit from latency or bandwidth?

 

[Deleted member 185158]

Would those benefit from latency or bandwidth?

Of course.

But you don't have to lean on AIDA64. I mean it's no secret keep re-running read read read oh there's a good score! I'll show off this one!!

Let's look at a more competitive approach.

Ya'll think that tweaking is fast? Gotta fast 12th Gen, maybe a AMD 5800X3D super sport?

Beat 5m 15.339s PiMod 32m

 

[INSTG8R]

Hi,
You'd see better results at c16 and higher frequency
Only difference is higher dimm voltage which lower cas latency always uses more.

4000mhz with c16 timings 16-16-16-36 only price is close to 1.45v but what ever enjoy c18.

I mean I don't think is terrible for mediocre Hynix with mediocre timings and my voltage is 1.38v

 

[Thrawn]

Ryzen 5 2600 + Samsung B-Die user here. This setting for me is rock stable for a long time. Maybe it would be stable with 1.375V for the memory, but I wanted to be sure about that. The memory chips can handle it easily so why not? No OC on the CPU.