Ryzen DRAM Calculator by 1usmus

[Tatty_Two]

Give it a try..... gen 1.

 

[warcraft]

i did put dram current voltage to 1.36 and this is the result:
i successfully loged into windows .the question is: are these timings stable ? how can i test it ?any special program ?(sorry im newbie at this)
another thing: i put number 292 for tRFC but it jumps to 288 and tRTP jumps to 12 automaticly . why is that?

 

[Tomorrow]

Anyone here have Hynic J-Die based memory? I recently bought Kingston HyperX 3200 CL16 (2x8) and expected to get C-Die (CJR) but got J-Die (JJR) instead.

Thankfully it seems to clock as good as CJR when comparing the results. Currently running at 3733Mhz 16-19-18-36 1T in 1:1 mode with IF at 1867 on X570 Aorus Master.

There's literally no info about this J die i can find. It's fairly new on my kit. Manufactured in the middle of august in Keelung, Taiwan at 18nm.

Obviously the calclulator does not have this listed. Hopefully if the creator is monitoring this topic it will be added in the future version. In the meantime i used CJR option when calculating timings tho i can only calculate safe timings for some reason. It's a start but i can basicly tighten most things the calculator recommends.

 

[Apocalypsee]

Oh wow there is Samsung OEM RAM now which is what I use, judging by the calculator it gets very close to what I could get by manually tweaking them. I disable Gear Down mode though. Probably I could tweak it a bit more.

 

[warcraft]

Here it is Overclocked DRAM. Actually it's working at the base clock 3000Mhz - but i improved timing and got better passmark rank. i think i can get better clock speed 3200Mhz. but i don't want to mess with her ) also i didn't do any stress tests.

however i did a couple of benchmarking programs as stability test. and also have played several games such as (Control - Wolfenstein Youngblood - Deus Ex - Mankind Divided) and some more. i think its stable now. i don't agree with stress testing rams and cpus to inform stability.

 

[Apocalypsee]

Here it is Overclocked DRAM. Actually it's working at the base clock 3000Mhz - but i improved timing and got better passmark rank. i think i can get better clock speed 3200Mhz. but i don't want to mess with her ) also i didn't do any stress tests.

however i did a couple of benchmarking programs as stability test. and also have played several games such as (Control - Wolfenstein Youngblood - Deus Ex - Mankind Divided) and some more. i think its stable now. i don't agree with stress testing rams and cpus to inform stability.

I do agree that running stress test is unnecessary, I usually load the RAM to max by opening browser with multiple tabs then running games that require lots of RAM. Can be a pain if you progressing the game and it crashed, if so I just resorted to using RAM stress test for about 5-15 minutes. It should spew error within that timeframe

 

[warcraft]

Hello
i have two different Drams.
1-Crucial Ballistix Elite 8GB 3000MHz - Samsung E-Die
2-Crucial Ballistix Sport 8GB 3000MHz - Micron E-Die
and i got these timings for each one.From (Ryzen DRAM Calculator 1.6.2)
my question is which timing should i set into BIOS to get best result?

 

[TheoneandonlyMrK]

Hello
i have two different Drams.
1-Crucial Ballistix Elite 8GB 3000MHz - Samsung E-Die
2-Crucial Ballistix Sport 8GB 3000MHz - Micron E-Die
and i got these timings for each one.From (Ryzen DRAM Calculator 1.6.2)
my question is which timing should i set into BIOS to get best result?

View attachment 134601View attachment 134602

Do you have them both in the same system at the same time? As two separate bought dual channel kits?.

 

[Tomorrow]

If you have both at the same time always use the higher (worse) timings.

 

[TheoneandonlyMrK]

If you have both at the same time always use the higher (worse) timings.

I would say if you have it working consider leaving it but the higher settings might not be loose enough especially on TRC and trfc times to accommodate that many sticks, worth a try though.

 

[warcraft]

Do you have them both in the same system at the same time? As two separate bought dual channel kits?.

i have them at the same system at the time.

If you have both at the same time always use the higher (worse) timings.

that's what i thought. but tRFC number is bothering me.what is this huge deference for? both sticks are 3000Mhz from same company.
these are timings i have set- i picked (higher) value from each column . how should i know if drams are stable or not ?

i have played couple of games - control + Wolfenstein Youngblood - every thing is okay till now. any other tests should i run?

 

[TheoneandonlyMrK]

i have them at the same system at the time.


that's what i thought. but tRFC number is bothering me.what is this huge deference for? both sticks are 3000Mhz from same company.
these are timings i have set- i picked (higher) value from each column . how should i know if drams are stable or not ?

View attachment 134606
i have played couple of games - control + Wolfenstein Youngblood - every thing is okay till now. any other tests should i run?

The ryzen calc includes a stability test.
Try looser trc and trfc settings if it fails.

 

[Tomorrow]

Samsung has much tighter secondary timings than Micron or Hynix. The company that produced the stick does not matter in this case. What matters is the chip supplier. Roughly there are 3 companies that make DRAM chips: Samsung, SK Hynix and Micron. You may buy two kits from the same manufacturer like for example Crucial in you case (they are actually a subsidiary of Micron). They have the same speeds, same price. Same looks and perhaphs even the same XMP timings.

Because XMP only sets the voltage, frequency and primary timings people often think it's ok to use these kits together because they appear to be the same. But they are not the same. Generally you never mix and match two different kits together. That's just asking for trouble.

I had that experince on DDR3 days where i bought one G.Skill kit. And later bought a second G.Skill kit that even had the same product code and was identical by all accounts. It was not until i installed both kits on the motherboard that i found that the newer kit actually had worse secondary timings. And even setting both kits to use those worse timings was not enough to achieve stability.

You should run memtest86 from a bootable usb stick (full 7 hour test) or run memtest64 in windows.
I would not worry about tRFC too much. Despite the high number it does not have as massive impact on performance as the number would suggest. Infact the difference between 292 and 504 may only be visible in benchmarks.

 

[warcraft]

Samsung has much tighter secondary timings than Micron or Hynix. The company that produced the stick does not matter in this case. What matters is the chip supplier. Roughly there are 3 companies that make DRAM chips: Samsung, SK Hynix and Micron. You may buy two kits from the same manufacturer like for example Crucial in you case (they are actually a subsidiary of Micron). They have the same speeds, same price. Same looks and perhaphs even the same XMP timings.

Because XMP only sets the voltage, frequency and primary timings people often think it's ok to use these kits together because they appear to be the same. But they are not the same. Generally you never mix and match two different kits together. That's just asking for trouble.

I had that experince on DDR3 days where i bought one G.Skill kit. And later bought a second G.Skill kit that even had the same product code and was identical by all accounts. It was not until i installed both kits on the motherboard that i found that the newer kit actually had worse secondary timings. And even setting both kits to use those worse timings was not enough to achieve stability.

You should run memtest86 from a bootable usb stick (full 7 hour test) or run memtest64 in windows.
I would not worry about tRFC too much. Despite the high number it does not have as massive impact on performance as the number would suggest. Infact the difference between 292 and 504 may only be visible in benchmarks.

i got it - thank you so much for your time.
thats became interesting... so what Must i set as Memory Type in Ryzen DRAM Calculator 1.6.2 program for each Dram ?

 

[Tomorrow]

The types you set before were correct. The Ballistix Sport is defenetly Micron E-Die even if for some reason the Raw card designer shows SK Hynix for some reason. If the memory kit has AES in it's module number/product code then it is guaranteed Micron E-Die.
The Ballistix Elite does seem Samsung E-Die.

Im not a memory expert but from what i know of these chips the Samsung E-Die has better secondary timings to run at stock but is not so good for overclocking, but the Micron E-Die has worse secondary timings but is much better for overclocking.

 

[warcraft]

The types you set before were correct. The Ballistix Sport is defenetly Micron E-Die even if for some reason the Raw card designer shows SK Hynix for some reason. If the memory kit has AES in it's module number/product code then it is guaranteed Micron E-Die.
The Ballistix Elite does seem Samsung E-Die.

Im not a memory expert but from what i know of these chips the Samsung E-Die has better secondary timings to run at stock but is not so good for overclocking, but the Micron E-Die has worse secondary timings but is much better for overclocking.

very appreciated to you

 

[damric]

Love your program. Requesting some insight on Hynix D 17nm.

 

[R3jn1971]

Anyone here have Hynic J-Die based memory? I recently bought Kingston HyperX 3200 CL16 (2x8) and expected to get C-Die (CJR) but got J-Die (JJR) instead.

Thankfully it seems to clock as good as CJR when comparing the results. Currently running at 3733Mhz 16-19-18-36 1T in 1:1 mode with IF at 1867 on X570 Aorus Master.

There's literally no info about this J die i can find. It's fairly new on my kit. Manufactured in the middle of august in Keelung, Taiwan at 18nm.

Obviously the calclulator does not have this listed. Hopefully if the creator is monitoring this topic it will be added in the future version. In the meantime i used CJR option when calculating timings tho i can only calculate safe timings for some reason. It's a start but i can basicly tighten most things the calculator recommends.

View attachment 132494

ad c-die sent for warranty and sent me J-die could you give me your exact timing settings so that you can reach your rating? from cpu-Z e.g. photo? and the voltage you gave on these memories ... for now so, but unfortunately 1 error pops up in memtest 5 ...

 

[Tomorrow]

ad c-die sent for warranty and sent me J-die could you give me your exact timing settings so that you can reach your rating? from cpu-Z e.g. photo? and the voltage you gave on these memories ... for now so, but unfortunately 1 error pops up in memtest 5 ...

Im using 1.45v. These are the settings DRAM Calculator recommends. The performance i get currently and the actual settings from BIOS. Hope this helps.

 

[R3jn1971]

Im using 1.45v. These are the settings DRAM Calculator recommends. The performance i get currently and the actual settings from BIOS. Hope this helps.

I have the msi b450 tomahawk max disc but I hope that even though the 3600 will reach ... and I have a question, did you test memory stability after oc? Eg TestMem5 from this Russian publisher? because it's very important how you can run the test if you have any errors

 

[Tomorrow]

I have the msi b450 tomahawk max disc but I hope that even though the 3600 will reach ... and I have a question, did you test memory stability after oc? Eg TestMem5 from this Russian publisher? because it's very important how you can run the test if you have any errors

I used Memtest x86 from a bootable USB stick. It ran for 60 minutes and completed almost 2 passes. No errors were reported so i considered it stable. So far i have not noticed any problems using the computer for a month. I believe these settings are also somewhat motherboard and CPU dependant. For example my 3800X does not go over 1867 IF clock so i have to limit the memory to 3733Mhz too. If this was not the case this kit would do over 4000Mhz easily.

 

[R3jn1971]

Memtest 86 not good for ddr4... 3733 is good for infinity fabric max next go to low timings 20=>19 its my thanks for you help 3200 to 3733 very good

I used Memtest x86 from a bootable USB stick. It ran for 60 minutes and completed almost 2 passes. No errors were reported so i considered it stable. So far i have not noticed any problems using the computer for a month. I believe these settings are also somewhat motherboard and CPU dependant. For example my 3800X does not go over 1867 IF clock so i have to limit the memory to 3733Mhz too. If this was not the case this kit would do over 4000Mhz easily.

infinity fabric ends at this setting, therefore, for ryzenów 3600 memories are recommended because 3733 are not sold at 4000, the asychnorical setting is already set and then the performance drops ... it would be worth fighting for lower timing cl 16 here is ok but those that have 20 e.g. go down to 19 or 18

 

[Tomorrow]

Memtest 86 not good for ddr4... 3733 is good for infinity fabric max next go to low timings 20=>19 its my thanks for you help 3200 to 3733 very good


infinity fabric ends at this setting, therefore, for ryzenów 3600 memories are recommended because 3733 are not sold at 4000, the asychnorical setting is already set and then the performance drops ... it would be worth fighting for lower timing cl 16 here is ok but those that have 20 e.g. go down to 19 or 18

Perhaps. I have not had time to test with other programs. I did test 16-19-18-36 but it was not stable in memtest although it was stable in windows. 16-19-19-36 was also not stable. 16-20-20-36 is stable from what i can determine. Atleast at 3733Mhz.

 

[R3jn1971]

Ok 1.45 v is goot for 24/7?? Norma this ram 1.25 xmp 1. 35

 

[Tomorrow]

1.45v is perfectly fine. DDR4 does not get very hot.