Share your AIDA 64 cache and memory benchmark here

[LiquidTrance]

Appears you have done quite a bit of testing.

I am at 60/61 and 6/6 @ 16-16-16-34 2T 4000MHz.

This kit will not train above 4000MHz, even though it is 4000MHz XMP2. Not on the QVL, probably lucky it boots at all...

I spent 1 year learning from scratch, first overclocking desktop for me. See those are decent rtls for c16 4000. You can lower latency further by using MemTweak it software to lower tXP to 6 and PPD to 0 or by adjusting them in the bios if they are visible. You must enable real time memory training in the bios first before using memtweak it or the changes don't stick. You also have to use memtweak it every reboot to make those changes unless you have some sort of auto script that runs every reboot or something like that.

If you want to train anything else higher than 4000 on that kit, you may have to manually enter everything in. Might even have to guess the rtls/iols.

 

[Arctucas]

I have tXP at 4, currently.

Tried 15-16-16-34 and RTLs went to 58/59. Will not train at 15-15-15.

I guess I should be satisfied with a C19 kit getting those timings.

 

[LiquidTrance]

I have tXP at 4, currently.

Tried 15-16-16-34 and RTLs went to 58/59. Will not train at 15-15-15.

I guess I should be satisfied with a C19 kit getting those timings.

is it c19 4000 @ 1.35v with +50mv for each cas reduction? so from c19 to c16 1.5v vdimm total? If that is the case, you would need 1.55v for 15-15-15-32(in theory) or 1.55v for c16/4100/4133(in theory) but there is no guarantee those dimms will do it on your board. Never messed with that kit before. Have you tried clocking up at all at xmp cas to something like 4400 or 4500?

Setting ppd to zero will net 1ns lower latency, maybe even more

 

[maksover]

 

[Arctucas]

is it c19 4000 @ 1.35v with +50mv for each cas reduction? so from c19 to c16 1.5v vdimm total? If that is the case, you would need 1.55v for 15-15-15-32(in theory) or 1.55v for c16/4100/4133(in theory) but there is no guarantee those dimms will do it on your board. Never messed with that kit before. Have you tried clocking up at all at xmp cas to something like 4400 or 4500?

Setting ppd to zero will net 1ns lower latency, maybe even more

The kit does 16-16-16-34 @ 1.415V, 500% HCI memtest (have not tested longer yet).

Needs 1.3V vccSA and 1.25 vccIO.

Does not want to train at any frequency above 4000MHz. I have not tried really loose timings (24 or higher), but that would offset any frequency gains, so what would be the point?

Still need to experiment...

EDIT: the kit is G.Skill F4-4000C19D-32GTZSW.

 

[LiquidTrance]

 

[Arctucas]

View attachment 169535

Nice.

Not that I can run it, since I only have two DIMM slots, but what kit is that?

EDIT: Also, added my kit model number to my post from this morning (I was at work and did not have the number).

 

[LiquidTrance]

Nice.

Not that I can run it, since I only have two DIMM slots, but what kit is that?

It's GSkill 32gb(4x8GB) 17-17-17-37@4000 1.35v TridentZ RGB
The config is usable in games, in modernwarfare2 @ full ultra(rtx off) 1080p with a 2080ti getting 175-180 minimum fps and about 225 average fps depending on the map with gsync compatible/vsync on @ 5.05ghzCores/4.75ghzCache hyperthreading on. Average FPS goes up if I disable vsync so the gpu isn't restrained. Took 1.6v and 1.35 sa/io.

This is the graph from the map "Talsik Backlot"

 

[Arctucas]

It's GSkill 32gb(4x8GB) 17-17-17-37@4000 1.35v TridentZ RGB
The config is usable in games, in modernwarfare2 @ full ultra(rtx off) 1080p with a 2080ti getting 175-180 minimum fps and about 225 average fps depending on the map with gsync compatible/vsync on @ 5.05ghzCores/4.75ghzCache hyperthreading on. Average FPS goes up if I disable vsync so the gpu isn't restrained. Took 1.6v and 1.35 sa/io.

This is the graph from the map "Talsik Backlot"
View attachment 169544

Sorry, I do not play MW, so I have no idea what that means, but I trust what you are saying.

 

[Aksil1988]

hi guys .... i have a problem with my ram (Gskill Trident Z 3200 Mhz CL 16 18 18 38)

first of all impossible to go beyond 3600 Mhz in CL 17 I tried everything, IO and SA at 1.3 and the dram voltage at 1.45 the PC does not boot ... but even at this frequency ... I find my scores very low compared to the average ... (48.000 in Write memory only)

My built :

i9900k 5 Ghz

Maximus XI Formula

Gskill Trident z RGB 3200 Mhz CL16

Coolermaster Vanguard 1000w...

 

[mrthanhnguyen]

2x16 4000c17

 

[INSTG8R]

 

[Annihilation]

Ballistix 3200cl16. 2x16 kit. Micron e-die

 

[ncore]

 

[Zach_01]

 

[techwiz]

Is the write feature to be wrong or kit has defected? Because I'm encountering some random crash problems when XMP is on.

VENGEANCE® RGB PRO 16GB (2 x 8GB) DDR4 DRAM 3600MHz C18 AMD Ryzen Memory Kit — Black | CMW16GX4M2Z3600C18

 

[INSTG8R]

Is the write feature to be wrong or kit has defected? Because I'm encountering some random crash problems when XMP is on.

VENGEANCE® RGB PRO 16GB (2 x 8GB) DDR4 DRAM 3600MHz C18 AMD Ryzen Memory Kit — Black | CMW16GX4M2Z3600C18
View attachment 170886

My results are few up similar kit I just tightened the timings a little they are CL 18

 

[Final_Fighter]

 

[Zach_01]

Is the write feature to be wrong or kit has defected? Because I'm encountering some random crash problems when XMP is on.

VENGEANCE® RGB PRO 16GB (2 x 8GB) DDR4 DRAM 3600MHz C18 AMD Ryzen Memory Kit — Black | CMW16GX4M2Z3600C18
View attachment 170886

The scores are were they should be, given the CPU and the DRAM configuration. As a matter of fact are pretty good with those timings...

I bet you could do a little better (lower memory latency mostly) if those timings were a little tighter.
If you have any instabilities, its known that ZEN2 memory controller is pretty peaky with some memory sticks. You can try disable XMP and do all settings manually (speed, voltage, timings)

 

[LiquidTrance]

9900K@5.05-5.35 Turbo 4x8GB@3933
GSkill Trident Z RGB 4x8GB 17-17-17-37 4000mhz ram kit

 

[nicke85]

I think that many people wait to see this, pocket beast for AAA games GPD WIN MAX with LPDDR4 3733MHZ, i5-1035G7, 2TB NVME Samsung 960 Pro, 8 Inch 1280x800

 

[md2003]

Got some patriot PVS416G440C9K in hand, not the greatest but run good and stable (for higher Mhz, i.e. 4400Mhz, they need very high vccsa & io of over 1.45v, and dont know if that is safe for daily usage).

 

[LiquidTrance]

9900K@5.15GHz-5.45GHz Turbo 8C/8T 4x8GB@3933MHz 94.55%+ scaling

 

[Liviu Cojocaru]

3700X@4375mhz 1.33v ,4x8GB@3600mhz (15-19-16-36)1.38v Crucial Ballistix 3600 C16

 

[Zach_01]

R5 3600 on PB/PBO Enabled
Corsair Vengeance LPX, B-die A0, 3466MHz, 16-18-18-18-36-54 1.35V -->> @3800MHz, 16-18-18-18-36-54 1.45V