Help with overclocking my r9 290's Safely?

Hello everyone! Decided I wanted to see what my 290's were capable of overclocking wise. They're Powercolor PCS+ models. Stock clocks of 1040 core and 1350 memory. At stock, I'm scoring 7915 on firemark extreme with xfire enabled. Bumping core to 1100mhz and power target to 15, leaving everything else alone, I score 8027 on the same benchmark. Couple questions before I really start pushing. What voltage am I look at when looking at GPU-Z sensors page? What voltage should I stay under? I'm really looking to squeeze as much as safely possible out of these cards so help is appreciated!

Safely, rock 1.25v and raise the power limit to the max.

EarthDog said:

Safely, rock 1.25v and raise the power limit to the max.

Click to expand...

ok so 100% power limit. what part of the sensors page am I looking at to not go over 1.25v? VDDC?



If not, what tool do you use to monitor? I can't seem to find a voltage field in afterburner

Afterburner should have it. If not, use vddc in gpuz.

You may have to add voltage to the Afterburner graph in Settings.

Ok well my card apparently does NOT like going beyond 1110 mhz on the core. Even 1150mhz with +100 voltage gives me artifacts when benching. I manage to get 1100mhz core and 1550mhz memory with maxed out power target and stock voltage of +50. A bit of a disappointment. Thought I'd get more out of these cards. Oh well. Can't have every card be like my stock gtx 670's going from 915mhz to 1215 mhz without a hitch . Thanks guys.

Btw, voltage never went beyond 1.21v even when core voltage slider was maxed out, all voltage and safety stuff disabled in MSI AB. This mean that the card just gets unstable because of the binning of the chip itself rather than needing more voltage? Because the bench never crashed, just started getting flashes and bright lights.

Considering it has a 947 reference clock, 1100mhz is a ~16% improvement over the stock clocks. Stock memory is 1250, 1550mhz is a whopping 24% improvement. Thats a great boost in performance, albeit our FX chips limit them in my testing. 1025mhz was where I started seeing diminishing returns, 1100 provided zero improvement over 1050 in fps of games and benchmarking [heaven, r15, valley, 3d mark, game benchmarks] (albeit firestrike scores went up a bit more, but not a lot).


1550mhz, Im guessing thats hynix memory, my elpida crap doesnt go above 1350! Food for thought

All in all, I would consider those numbers pretty solid. I've "seen" these 1250+ 290's but I think they are far and few between.

bob4933 said:

Considering it has a 947 reference clock, 1100mhz is a ~16% improvement over the stock clocks. Stock memory is 1250, 1550mhz is a whopping 24% improvement. Thats a great boost in performance, albeit our FX chips limit them in my testing. 1025mhz was where I started seeing diminishing returns, 1100 provided zero improvement over 1050 in fps of games and benchmarking [heaven, r15, valley, 3d mark, game benchmarks] (albeit firestrike scores went up a bit more, but not a lot).


1550mhz, Im guessing thats hynix memory, my elpida crap doesnt go above 1350! Food for thought

All in all, I would consider those numbers pretty solid. I've "seen" these 1250+ 290's but I think they are far and few between.

Click to expand...

I think you nailed it on the head here with the perspective of using a reference 947mhz. I wonder if OP realized that the card is a boost from reference. So a boost on a boost there is only so much give. He could've got reference cards and water cooled them probably reached 1200ish.