OC question

[Gbaby1008]

I recently got a radeon x700 pro pci e, it comes with an overclocking program that will automatically find the max core and memory. i can get the core to 450 stable and cool but when ever i try and oc the memory even by as little as ten after a few minutes of gaming the screen starts turning wierd colors then little squars repeat over it then it freezes. it has me stumped.

btw stock core is 425 stock mem is 432

thx

 

[Killrose]

Bad news if you keep it up. You could try ramsinks and maybe get stable performance of 10MHz over, but you will more than likely kill the card if you persist

From your description, you lost the overclocking lottery.

 

[Gbaby1008]

Originally posted by: Killrose
Bad news if you keep it up. You could try ramsinks and maybe get stable performance of 10MHz over, but you will more than likely kill the card if you persist

From your description, you lost the overclocking lottery.

so if i lay off the mem should and just oc the core is that ok like no bad news for a while anyway

 

[mindless1]

It's very rare for a card to only have < 10 Mhz memory margin, I'm wondering if your "400W" PSU is a generic that has just about ran out of magic smoke.

 

[Gbaby1008]

Ya its a generic and i've had it for about 2 years so thats what i'm leaning towards. I guess its off to newegg for me

I was also wondering if its safe to go past the 'max core' that the program sais as long as its stable?

 

[mindless1]

Increasing GPU higher, higher, and even higher, is not in itself a problem. Here are the potential issues:

More heat- 'sink ever-closer to inadequate and if 'sink fan were to fail (a commmon problem with thin/small/cheap stock fans on many cards... dunno about yours) then GPu is that much closer to a critical temp

More current and caps- if your card has electrolytic capacitors their lifespan is directly proportional to their temp. If you touch-test your cards and find them more than barely warm, it's a fair bet your card won't have a full life, but which day/week/month it will fail, I cannot answer.

More current and fets - modern parts are smaller and handle more current but that means lower margins for temp. In many higher-end cards you'll now find voltage regulation circuit components have 'sinks on them, while lesser cards (using leser current per same regulation components) do not. If your o'c increases the current substantially (it's roughly linear if the GPU core voltage stays the same) then you approach the threshold for these parts too.

Overclocking programs mostly just check for errors. If your card produced one error in 500 trillian you might never notice it during gaming and benefit from the performance boost. On the other hand it might kick you out to the desktop or other interferance with your gaming which can be most frustrating. It doesn't tell you anything about how long a card run will at a give speed over the long term. Take a nVidia TI4200 as an example. When the clock speeds increased a little more there was obvious increase in caps and other power supply circuitry on the card to make it substantially larger and more expensive, only to help accomdate the additional power used.

So, I guess I give a mixed answer- you can completely ignore what anything says is our max core as it doesn't really relate to card lifespan, and yet these programs might even overaggressively estimate the max that is stable in any particular game since the game may repetitively stress a certain part of the GPU.

What you might want to do is benchmark it with the core at stock and raised some to see what the difference is. Often the memory is a bottleneck and raising core gains so little it's not worthwhile if the memory isn't raised a lot too.