DDR1/2/3/4 Subzero: Max. MHz basics

[yosarianilives]

  On 11/16/2018 at 2:44 PM, AutisticChris said:

When is b-die cold bug?

Expand  

bdie is happiest 5-15c, below that it starts getting angry and won't do good mhz and timings. Although for max mhz runs you let timings out and run bdie on ln2.

[SparkysAdventure]

  On 11/16/2018 at 4:09 PM, yosarianilives said:

bdie is happiest 5-15c, below that it starts getting angry and won't do good mhz and timings. Although for max mhz runs you let timings out and run bdie on ln2.

Expand  

Still, when is the cold bug? Say when you're going for max MHz. I've heard anywhere from -60°C to -120°C.

[Bullant]

Sure, you can freeze B.die....benches require a specific temp....because of cb....  memory Mhz , timing and rtls can tighten as well

Will post some results of this when I have gear

[SparkysAdventure]

  On 11/18/2018 at 12:22 PM, Bullant said:

Sure, you can freeze B.die....benches require a specific temp....because of cb....  memory Mhz , timing and rtls can tighten as well

Will post some results of this when I have gear

Expand  

Looking forward to it. I haven't seen any of your killer results for a while.

[Noxinite]

Necro bump, it seems you can get around the volt-boot issues with DDR2 by doing a vmod on a motherboard to raise default voltage from 1.8V to 2.4V.

I tested in on one crappy motherboard and it seemed to work, so I'll try and move forward to testing on my UD3P.

[ground]

How much of a difference does PCB make for raw clocks on DDR3?

[websmile]

@Noxinite it is called (cold)volt boot issue because sticks don´t boot at 1,8v but need normally around 2,2v at least - you can use a volt mod but for a lot of people it is simply enough to boot with other sticks, set volts to 2,2 or above and change sticks

PCB is always a factor, but not the biggest one for raw mhz because it is more of a stability factor

[Noxinite]

  On 6/8/2019 at 7:39 PM, websmile said:

@Noxinite it is called (cold)volt boot issue because sticks don´t boot at 1,8v but need normally around 2,2v at least - you can use a volt mod but for a lot of people it is simply enough to boot with other sticks, set volts to 2,2 or above and change sticks

PCB is always a factor, but not the biggest one for raw mhz because it is more of a stability factor

Expand  

Maybe I'm impatient, but I found swapping sticks out while benching really annoying. Also, you can't really do that if you have an Ln2 pot or SS on the ram sticks.

[websmile]

The problem is that you make the board useless for some ddr2 ics and you might even get a cold boot issue on sticks that do not have this before - for cold use it is a good option though

[Noxinite]

Oh, I always put switches and trimmers on my mods so I can go back to stock easy.

Edited June 9, 2019 by Noxinite

[websmile]

My bad then, with a potentiometer this is ideal way to handle the problem ?

[dumo]

  On 11/16/2018 at 2:44 PM, AutisticChris said:

When is b-die cold bug?

Expand  

New A2 stable temp varies from 5 to -30 for 5000+ tight

Edited June 9, 2019 by dumo

[gtxx58]

I put some time on ddr2 recently, and the best way to bin ddr2 seem to be 32m spi, I set my voltage to 2,4v to not damage it, d9 is very sensitive to temps, just put better heat spreaders on it change everything, recently I manage 720 32m spi with single stick (5-5-5-15 at 2,42 bios) on air, some stick seem to cbb at -115 and cb at about -150, different result than noxinite ? maybe different pcb change cold bug

i dont no if my board do magic on spi 32m or its just the stick, i also have 1 stick with different ic doing 750+ but with loser timing 5-6-5-16

Edited February 12, 2020 by gtxx58

[ObscureParadox]

  On 2/12/2020 at 8:20 PM, gtxx58 said:

I put some time on ddr2 recently, and the best way to bin ddr2 seem to be 32m spi, I set my voltage to 2,4v to not damage it, d9 is very sensitive to temps, just put better heat spreaders on it change everything, recently I manage 720 32m spi with single stick (5-5-5-15 at 2,42 bios) on air, some stick seem to cbb at -115 and cb at about -150, different result than noxinite ? maybe different pcb change cold bug

i dont no if my board do magic on spi 32m or its just the stick, i also have 1 stick with different ic doing 750+ but with loser timing 5-6-5-16

Expand  

To be fair his result with cold bug was due to information from myself. I had about 3 sticks cold bug on my phase which does -55c ish. I've since taken a couple of sticks cold and both of mine got full pot so I must have just been unlucky with the first 3.

For max MHz binning though I think you can just get away with doing just that on air. Set a voltage paramater and timings etc and just see where the sticks crash. Is a hell of a lot quicker than binning for 32M. Bigger problem for me though is I now need to start binning motherboards as my only two P45 boards hit hard limits on the memory controller that no voltage or timings could overcome. Again I don't know if I just got unlucky here or not but I'm not even close to hitting the limit on my 850MHz stick since it could do those speeds at 2.3v or there abouts.