Question Regarding G1 970 PCB

[LooneyJuice]

This is somewhat of a poll I guess for G1 970 owners. But, I want to ask especially the ones that bought them closest to the actual release whether they've OC'ed the memory (or not) and if so, whether they've run into trouble. 

 

The reason I'm asking is because GB (And MSI do this as well if I'm not mistaken) have opted to spread the VRAM in front and behind the PCB. Trouble is, the VRAM on the back has no TIM pads and the modules are covered by the backplate (which makes me a very worried boy). I find it concerning as well since, even though my core is less than stellar (barely breaks 1500), my memory has easily hit +650 (even +700 before artifacting) without crashing, which would offer silly gains, but I'm concerned about memory longevity. My thanks for the responses.

[don_svetlio]

If you are concerned about longevity dial back the OC on the VRAM a bit.

[racer243l]

I can only confirm, that dialing your OC back should solve it. +650 is enormous, I got mine to +400 on memory and 135Mhz on the core with an additional 60mV, so a stable boost clock of 1545Mhz. And in most games I have to dial it back even further, to like +200 on memory.

[LooneyJuice]

4 minutes ago, racer243l said:

I can only confirm, that dialing your OC back should solve it. +650 is enormous, I got mine to +400 on memory and 135Mhz on the core with an additional 60mV, so a stable boost clock of 1545Mhz. And in most games I have to dial it back even further, to like +200 on memory.

I know +650 is enormous indeed, and I only ever ran it for a day to see where it'd top out. I was dumbfounded. It turned out to be able to run +700 along with the max core clock, no crashes. I finally got it to artifact at +775. Never touched that number since, been running 500/550 occasionally to make up for the lackluster core. But seeing as I mentioned, It's got mem modules on the back with no TIM pads, I'm sort of opting to stay away from touching the memory altogether. Basically, TL;DR I'm asking if people got their memory to blow up.

 

Not to mention that with my core, If i back down from the memory that much, it sort of defeats the point. First world problems I know, but 1501/8110 nets me that 10%, anything lower and it's sort of "meh". Thanks for the response though, I do indeed want some actual numbers people are running too. Especially since I can't for the life of me find any Temp measurements of the VRAM aside from some vague Thermal Imaging of the general area through the backplate.

[Lays]

If you're not messing with memory voltages it shouldn't be an issue, most likely you have Samsung memory, right?  It's very resilient in my experience. 

[LooneyJuice]

4 minutes ago, Lays said:

If you're not messing with memory voltages it shouldn't be an issue, most likely you have Samsung memory, right?  It's very resilient in my experience. 

Negative, it's actually Elpida! Yes, I laughed heartily as well. And yes, it's stock bios, stock memory voltage. Which is why I pushed it so far, 'cause it wouldn't crap out on stock voltage.

[racer243l]

58 minutes ago, LooneyJuice said:

I know +650 is enormous indeed, and I only ever ran it for a day to see where it'd top out. I was dumbfounded. It turned out to be able to run +700 along with the max core clock, no crashes. I finally got it to artifact at +775. Never touched that number since, been running 500/550 occasionally to make up for the lackluster core. But seeing as I mentioned, It's got mem modules on the back with no TIM pads, I'm sort of opting to stay away from touching the memory altogether. Basically, TL;DR I'm asking if people got their memory to blow up.

 

Not to mention that with my core, If i back down from the memory that much, it sort of defeats the point. First world problems I know, but 1501/8110 nets me that 10%, anything lower and it's sort of "meh". Thanks for the response though, I do indeed want some actual numbers people are running too. Especially since I can't for the life of me find any Temp measurements of the VRAM aside from some vague Thermal Imaging of the general area through the backplate.

Hm, I get your problem... Anyway I never heard of anybody blowing up memory modules with a stock bios. I would say, as long as you stay in the limits of the stock bios it should be fine (like additonal voltage).

[LooneyJuice]

16 minutes ago, racer243l said:

Hm, I get your problem... Anyway I never heard of anybody blowing up memory modules with a stock bios. I would say, as long as you stay in the limits of the stock bios it should be fine (like additonal voltage).

Yeah I'd never touch memory voltage. Even if all VRAM modules were on the top with TIM pads, I still wouldn't push it knowing that the bulk of the cooling is concentrated on the core. My original assumption was: If the memory responds that well to stock voltage and running it at +500 is still a clear 175-200mhz below its ceiling, does not artifact, and has 0 impact regarding crashes and instability, it should actually be a good enough and efficient sample to not fry itself by generating way too much heat. But seeing as I can't find any data to back that up, and metrics regarding VRAM are extremely rare, I wanted to sort of sample experiences on the forum. Thanks a lot for the input, much appreciated.

[racer243l]

57 minutes ago, LooneyJuice said:

Yeah I'd never touch memory voltage. Even if all VRAM modules were on the top with TIM pads, I still wouldn't push it knowing that the bulk of the cooling is concentrated on the core. My original assumption was: If the memory responds that well to stock voltage and running it at +500 is still a clear 175-200mhz below its ceiling, does not artifact, and has 0 impact regarding crashes and instability, it should actually be a good enough and efficient sample to not fry itself by generating way too much heat. But seeing as I can't find any data to back that up, and metrics regarding VRAM are extremely rare, I wanted to sort of sample experiences on the forum. Thanks a lot for the input, much appreciated.

No problem, always glad to help