8 PACK MEMORY RANGE GROWING: SAY HELLO TO 8 PACK RIPPED EDITION & 32GB KITS!!!

MrPils said:
Gaming is pretty good at testing for infinity fabric stability - any audio weirdness or usb weirdness is most of the time down to that. If you want something you can turn on and walk away from then Realbench is pretty good too, an hour or two of that will give you a close enough indication of your fabric stability that you can tweak the voltage over time and in day to day use for stability if required. Watch for WHEA errors either in your event log or by using hwinfo, most of the time these show up well before any crash and will get you very close to your stable IF voltage before you need to spend much time testing at all.

I would first give yourself an idea of where your fabric speed will end up when you get your build done by doing some gentle testing, don't try and find your final stable fabric voltage until you're at 1:1 with your memory. If you are manually clocking your CPU do that before you do any kind of thorough stability test on your memory or fabric, but if you're going to use PB2 or PBO then you can skip straight to fabric/memory. Once you have a mostly stable fabric clock (no whea errors, pass 20 mins or so of realbench) you can start raising memory speed towards it while keeping your memory timings loose enough that double your fabric speed should be well within the capabilities of your memory. Once you have reached 1:1 you can start tweaking memory timings - at this stage if you hit instability you want to add memory voltage first to see if that resolves the problem. You may find you need a small bump more fabric voltage along the way as your memory becomes faster, but mostly you'll be adding memory volts. As a general rule a test fail means more memory volts whereas whea errors mean more fabric voltage - as you close on in your final voltages and your best memory timings though this distinction can become more blurred.
Click to expand...

Hi MrPils.

I have tried a lot of FCLKs on my system the last couple of weeks but I have a very strange issue:

On my newish 5950 build I can do FCLK at 2000Mhz but NOT 1900Mhz no matter what settings, voltages etc I have tried in BIOS ... For testing I have memory at stock 2133MT/s Jedec (and UCLK at 1066Mhz) to rule them out of the equation. Trying 1900 FCLK has always resulted in NO POST but 2000 FCLK boots into windows just fine allthough with a few WHEAs piling up during use. (My current daily settings with tight 14,13,13,14,26 timing at 1866 FCLK is stable as a rock with ALL voltages set to Auto in BIOS except MEM which is at 1.5v (set by D:O:H:C)

I am aware there is work going on by AMD optimizing AGESA to make FCLK stable over 1800Mhz but simply cannot understand why 2000 FCLK is sort of working and 1900 FLCK is completely nogo.. And yes no matter what FCLK setting I can only get my RAM working at 3733MT/S which also is a pain as I bought them for the 1900 KHZ sweetspot :D

Hope you have some insights, ideas or pointers I can chase as I really want to run 1:1:1 @ 1900Mhz to get my Gskills to work at published 3800MT/s (my ASUS MB is on the G-SKill QVL for my quad 8Gb sticks F4-3800C14Q-32GZTN )

EDIT: (Mem can do 2000Mhz/4000MT/s easily with only two sticks on mobo with 2000:2000:2000 FCLK:MCLK:UCLK but not been able to do it in quads yet but thats another story :) - makes NO difference for the completely unobtainable 1900Mhz FCLK though)

Thanks in advance

Cheers

C
 
Last edited:
GregI said:
I was reading a review of the G. Skill 2x16GB 4000 kit published yesterday.

I knew abit about the PCB layout for single rank but they gave some pretty interesting info for dual rank PCB layout: https://www.overclockers.com/g-skill-trident-z-royal-series-32gb-ddr4-4000-review/

"The SPD read also shows us that this PCB is running the B1 layout, which we confirmed with a physical inspection.

DDR4 PCBs are broken down into three major layout designs. The older design, called A0, has the ICs spaced out evenly on the PCB and can limit the maximum frequency achievable. The newer A1 and A2 designs (which are effectively the same) place the ICs closer together and closer to the PCB connection edge. The A2-style PCB has become the new unofficial industry standard because it allows for higher frequencies and generally better compatibility. As a result, motherboard manufacturers are now routing memory traces to coincide with the IC placement on A2-style PCBs.

Our review sample memory utilizes the modern B1-style PCB layout, which is similar to the A1/A2 style but carries the ‘B’ to signify a dual-rank format."

@Guest2 Can you attach the Taiphoon screenshot for the TeamGroup? B1 also?
Click to expand...

I bought the 2x16 3600 cl16 kit a while back but was unable to use them until I got the rest of my system, less than a week old atm

Anyways trying to get my head around the memory overclocking and I see the tab that let's you select a2/by and b0/a0 etc, was completely baffled why the software program gave me a reading of b1, this cleared it up
 
So I'm very new to this memory overclocking, I have the 16x2 2600 cl16 8pack memory

I tried the memory overclocking tool and just tried to lower the latency first, the tool recomeneded 1.36-1.4v, I've not stability tested it too much but a quick check seems 1.39 seems to work

So I've 3600mhz @ 14-14-15-14-28, brought my latency in Aida 64 down from 62.4ns to 57.7ns

Il be trying higher frequencies over the next while but thats my quick and dirty overclock
 
FireTwo said:
pump up them volts 1.45v min
samsung b die 'official' spec limit is 1.5
Click to expand...


Yeh I know 1.5 is by all accounts okay for b-die but just playing it safe atm while I get used to 1-overclocking ram, and 2 - overclocking amd in general (been on sandybridge for a good few years so its abit new to me)
 
Hi All,

So this is where I've settled on clocking the 32G 3200 CL14 kit. The IC's on this do not appear to be as good as the 16G 3200 CL kit I have but I don't think this is a bad effort. These are tested via various tests: 1) Memtest86 (which some say is useless) its quick and will tell you if you are even close or not = 4 hours here. 2) Testmem 5 actually used a couple of configs but 1usmus 4 hour stable.

Realbench too as its a system wide kinda test. Good hour in here while watching eventviewer.

It took a bit of time with this kit, they seem to be harder to get right compared to my 16g kit (results for that kit are in this thread.

trdwr and twrrd would not go any lower than the values here - none post.
trfc any lower and funky stuff started happening - post was fine but the system just did not feel right.
trcdrd just a no boot any lower.

For me, BIOS 3.80 was a no go for RAM OC. This probably resulted in two days lost trying to get stable.

AIDA Results are:
R= 58530
W= 58147
C= 56427
L= 63.2 - thought that was good?

uc
 
Last edited:
Doco said:
Yeh I know 1.5 is by all accounts okay for b-die but just playing it safe atm while I get used to 1-overclocking ram, and 2 - overclocking amd in general (been on sandybridge for a good few years so its abit new to me)
Click to expand...
1.5V here. 4 hours with no errors in memtest5 with anta extreme1 config. 3800 / 1900 if / CL14
Anyone know why the anta extreme1 config changes the memtest details to symbols/jibberish? Standard and some other configs appear fine
 
Guest2 said:
1.5V here. 4 hours with no errors in memtest5 with anta extreme1 config. 3800 / 1900 if / CL14
Anyone know why the anta extreme1 config changes the memtest details to symbols/jibberish? Standard and some other configs appear fine
Click to expand...

That's sweet. I've ordered the 32GB 3200C14 kit, and hopefully will be able to achieve the same as you... Still waiting for CPU though :rolleyes:

Which mobo have you got?
 
Guest2 said:
That’s with the 3600C16 kit (2 x 16GB)
Motherboard is the MSI MEG Unify x570
Click to expand...

I know you've got the 3600C16, but results should be the same with 3400C14 kit. I've got a x570 Tomahawk, which is very similar to the Unify for overclocking, so hopefully I'll also get good results.
 
Stu said:
I know you've got the 3600C16, but results should be the same with 3400C14 kit. I've got a x570 Tomahawk, which is very similar to the Unify for overclocking, so hopefully I'll also get good results.
Click to expand...
Are, good stuff. Yeh, i've read the 3600CL16 and 3400CL14 are similar as is the x570 Tomahawk and Unify
 
Hi All,

Further to my post above. I've been playing a little more. RAM at 1.5v Soc at 1.050 VIOD 1.025 and VCCD 950.

Various stability tests run inc:

Realbech - as its more of 'real' world type of a test = overnight.
MemTest86 = 4 hours
TestMem5 1usmus v3 = 4 hours

Does anyone have any further suggestions as to stability testing?

@MrPils I know you are a busy man with work and family but if you have a spar 5 min I would gratefully receive and feedback you have. And, again, I need to thank you for your efforts thus far in this thread. Without your input, I would have given up with my 16g kit and certainly wouldn't of bothered trying to improve the new 32g kit.

uc


uc


uc
 
@Supernaut91 Yes they are dual rank. I wouldn't bother upgrading in your scenario. Going to dual rank is not worth the price if you already have 16GB of good memory. The kit at £199 makes sense for people doing a complete pc upgrade or didn't have comparable RAM before.

I was actually looking at a really interesting Crucial SKU that has flew under the radar a little: BL2K32G36C16U4B

2x32GB. This would ensure dual rank and at £275 it's not much more money to double your RAM (especially if the £199 sale for 32GB doesn't come back on).
 
You must log in or register to reply here.
Back
Top Bottom