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

Space Monkey · 24 May 2020 at 16:45

GregI said:
That's an interesting score but I feel like it's bugged or something. The G.Skill C15 kit I have was 100 and 4600 C16. You would assume 100% would be a maximum score.

What's the screenshot for the Main tab Fast preset looking like? I wonder if it will also recommend 14-14-15-28 at 1.45v

Not sure if this is what you are after? It's set to 3600 Fast but will give me figures for RAM speeds a lot higher?

GregI · 24 May 2020 at 16:52

Oh yeah I had that problem with 1.7.3 too

Yeah that screenshot is great. It seems this calculator gives these values to every good B Die kit.

Out of curiosity btw are you able to run XMP at 1T with Gear down mode off?

Space Monkey · 24 May 2020 at 16:54

GregI said:
Oh yeah I had that problem with 1.7.3 too

Yeah that screenshot is great. It seems this calculator gives these values to every good B Die kit.

Out of curiosity btw are you able to run XMP at 1T with Gear down mode off?

Where would I confirm that? I'm running XMP at 1T but not sure about the Gear Down Mode setting in use?

GregI · 24 May 2020 at 17:12

It should be in the BIOS. For Aorus it should be under "Ddr frequency and timings > controller config > gear down mode set to disabled"

Also someone did some interesting overclocks with this set: https://community.hwbot.org/topic/196168-teamgroup-argb-ddr4-3600c14-a-first-look/

Safe: "After a quick test on the XMP values, it quickly became clear that there was still room for improvement in the clock and timing. Based on this, I made the first tests with a maximum of 1.5v, which is still acceptable for daily use at Samsung.

As expected, the XMP on DDR4-3600 ran smoothly with CR1T, so I set out to tackle bigger challenges. DDR4-4000 C15-15-15-38 CR1 ran through Memtest86 and all test benchmarks like SuperPi 32M, Geekbench 3, Cinebench R11.5 also ran with tightened sub-timings. At Cas 15 I finally reached stable DDR4-4133, but after that I had to switch to Cas 16 and CR2T to get ahead. Amazingly, the set values from C16-16-16-38 to DDR4400 were stable, at 1.5 volts, and even DDR4-4500 booted but could no longer be brought into Windows.

In the end, I managed to push the absolute limit of my storage controller. With the DDR4-4800 C18-18-18-42 CR2T with auto-subtimings, I was able to complete a complete benchmark course. As a brief interim conclusion, it can be said that the DDR4-3600 ARGB was able to distance itself from the DDR4-4500 8Pack Edition that I tested a while ago on the same platform, a surprising result."

Crazy aka don't do this: For benchmark use on platforms like UL (ex Futuremark) or Hwbot, DDR4-4000c12 and higher have been standard for years and I therefore also tested these settings.

DDR-4000 C12 was not a challenge, DD4-4133 C12-12-12-28 ran with 1T and 1.98V for Geekbench 3 and other comparable tests, SuperPi 32m also ran with 1.99V.

Since DDR-4600C14 and higher are now part of the bench routine of enthusiasts and professional overclockers in the bench circus with A2-PCB layout memory, I naturally also made a short test on these settings. DDR4-4800 C14-14-14-28 and C14-13-13-28 ran stable through multiple tests at 1.98-1.99 volts.

Space Monkey · 24 May 2020 at 17:19

GregI said:
It should be in the BIOS. For Aorus it should be under "Ddr frequency and timings > controller config > gear down mode set to disabled"

Also someone did some interesting overclocks with this set: https://community.hwbot.org/topic/196168-teamgroup-argb-ddr4-3600c14-a-first-look/

That's some amazing overclocking there with these sticks! So there is a good chance the figures quoted on my first pic are correct?

I'm in the middle of downloading at the moment so I'll have a look in the BIOS later and report back

GregI · 24 May 2020 at 17:26

More than likely they are correct, just not at good voltages <1.6v.

Space Monkey · 24 May 2020 at 19:20

GregI said:
More than likely they are correct, just not at good voltages <1.6v.

According to Ryzen Master I'm running 1T with GDM enabled, so what is the implication of this?

GregI · 24 May 2020 at 20:07

Just a very small latency penalty. Apparently the guy said in the review that the RAM can run true 1T on Intel at XMP but Ryzen IMC might struggle to do this. You can test it if you want and if you can, happy days. If not go back to gear down mode on.

Troezar · 24 May 2020 at 20:07

Space Monkey said:
According to Ryzen Master I'm running 1T with GDM enabled, so what is the implication of this?

I think that means it will run at 2T if it thinks it's required. AFAIK GDM disabled will run at a fixed 1T.

Space Monkey · 24 May 2020 at 20:34

Ah right, so I'm already running XMP/1T/GDM Enabled so that's fine.

I remember reading on AMD setups that with GDM Enabled if you're running an odd numbered CAS then it'll default to an even number, so CAS 15 becomes CAS 16 but even stays even?

Troezar · 24 May 2020 at 20:36

Space Monkey said:
Ah right, so I'm already running XMP/1T/GDM Enabled so that's fine.

I remember reading on AMD setups that with GDM Enabled if you're running an odd numbered CAS then it'll default to an even number, so CAS 15 becomes CAS 16 but even stays even?

I've heard something similar. I may try disabling GDM and see what happens.

Skynet5 · 24 May 2020 at 21:45

Space Monkey said:
Ah right, so I'm already running XMP/1T/GDM Enabled so that's fine.

I remember reading on AMD setups that with GDM Enabled if you're running an odd numbered CAS then it'll default to an even number, so CAS 15 becomes CAS 16 but even stays even?

What are your dram calc scores for easy and default?

Space Monkey · 24 May 2020 at 21:52

lnoton said:
What are your dram calc scores for easy and default?

Do you need all of them or just the basics?

If I set 3600 & Safe it gives me C16

Skynet5 · 24 May 2020 at 22:01

Space Monkey said:
Do you need all of them or just the basics?

If I set 3600 & Safe it gives me C16

Nah. The built in benchmark. Two modes easy and default.

Space Monkey · 24 May 2020 at 22:15

lnoton said:
Nah. The built in benchmark. Two modes easy and default.

Ah ok, I'll have to run that tomorrow as getting some kip now

Is there any other settings I need to select before running the benchmark?

Skynet5 · 24 May 2020 at 22:22

Space Monkey said:
Ah ok, I'll have to run that tomorrow as getting some kip now

Is there any other settings I need to select before running the benchmark?

Nope. Just select easy or default and run.

A pop up appears. Don't touch machine and it'll disappear and run the test.

opethdisciple · 24 May 2020 at 22:42

Still puzzled why my score at default in the dram calc benchmark is so poor 245.

And if it even matters....

From what I can tell it's not important. AIDA64 gives me a latency of 66ns and I have gained 1 flop when running the intel burntest. It's not much but at least it shows an increase in performance from 95 to 96 flops.

opethdisciple · 24 May 2020 at 22:47

Space Monkey said:
Not sure if this is what you are after? It's set to 3600 Fast but will give me figures for RAM speeds a lot higher?

Version 1.7.3 of the calc will give you different timings.

Skynet5 · 24 May 2020 at 22:56

opethdisciple said:
Still puzzled why my score at default in the dram calc benchmark is so poor 245.

And if it even matters....

From what I can tell it's not important. AIDA64 gives me a latency of 66ns and I have gained 1 flop when running the intel burntest. It's not much but at least it shows an increase in performance from 95 to 96 flops.

My understanding was that the number of cores should make no difference, otherwise the test would be pretty rubbish as its a memory test not a cpu test. Buy maybe not.

That might be a reason? I got a random 208 yesterday. I'm normally a bit under 210. I think the new chipset drivers helped!

opethdisciple · 24 May 2020 at 22:58

lnoton said:
My understanding was that the number of cores should make no difference, otherwise the test would be pretty rubbish as its a memory test not a cpu test. Buy maybe not.

That might be a reason? I got a random 208 yesterday. I'm normally a bit under 210. I think the new chipset drivers helped!

I'm uptodate on all drivers.

The things I haven't applied from the calculator are the VDDG and VDDP voltages. I've left them at stock.

I've been using the PC over the last 10 days and gaming on it so it's stable.

---

I just tried applying the 'rec' voltages for vddg and vddp and it made no difference what's so ever to my membench score.

The thing is the test takes 240% divides it by the number of CPU threads and then runs the test. So it makes sense that a CPU with more threads finishes the test faster.

When I run the test each thread does 15% of the task. 15 * 16 = 240

I reckon your 3900x each thread is only doing 10% of the task meaning the job finishes 33% faster.
10 * 24 = 240