Sign in with
Sign up | Sign in
Your question

Maximum Overclock for Sapphire R9 270X Dual X?

Tags:
  • Overclocking
  • GPUs
  • Sapphire
  • Memory
Last response: in Overclocking
Share
August 25, 2014 2:15:38 PM

The default clocks of my R9 270X are 1070mhz GPU and 1400mhz memory. It's been stable at 1120mhz GPU and 1520mhz memory for some time but today I tried to push it more and managed to get 1125mhz GPU and 1550mhz GPU. My power limit has been increased by 20% and GPU fan is at the default 40%. Overclocking the memory to 1560mhz results in my PC messing up in Battlefield 4 even though it passes 3DMark, Tomb Raider benchmark and GPU-Z stress test. It took me a few reboots to get my PC to boot back to Windows as well so I could change the overclock. 1550mhz memory appears to be okay though. The max temp the GPU reaches is 57-58 degrees. The GPU fan is set to 40% speed.

Simpler term:

GPU clock from 1070mhz to 1125mhz
memory clock from 1400mhz to 1550mhz
Power limit increased to 20%
Fan speed at default 40%
Stability tested through 3Dmark, GPU-Z built in stress test, Tomb Raider benchmark and Battlefield 4 multiplayer.

Is there any way I can make my GPU overclock further? I don't think my voltage is unlocked. My GPU drivers are the AMD Catalyst 14.6 Beta drivers.

More about : maximum overclock sapphire 270x dual

a b K Overclocking
August 25, 2014 3:03:28 PM

You can raise the voltage with MSI afterburner and EVGA precision, even if the voltage is locked in the GPU's BIOS. It will only allow the maximum safe voltage. Technically, the voltage is still locked even though you can raise it slightly in one of those OC programs. Core clock has a larger benefit in gaming than memory, so I would leave the memory at +0-50 offset (+300 core clock will yield better FPS than +150 core and +150 memory). Also it is easier to find max overclocks if you OC the core and memory separately. I would manually set the fans to 100% or the max allowed by the BIOS if you don't mind noise. OC'ing isn't a "one size fits all" kind of thing. Every GPU, even if the same model, will perform differently when OC'ed. My GTX 770 can hit +200 core clock offset on the stock voltages, which not all cards can do. Keep increasing core clock by +25 and stress testing it, and if it's unstable, MSI afterburner or EVGA precision should automatically undo the changes before the system crashes. Once you find the max OC for the core clock, set it back to default and repeat the same steps for the memory clock. Once you find the max OC for both, work on combining them, favoring core clock over memory (maybe +25 on the core while only +10 on memory). There will be a certain point where either increasing voltage any further causes temps to sky-rocket, or you need to dramatically increase voltages to even get +20 more MHz out of the GPU. High voltages can cause damage/kill the VRMs on the card.

Depending on what cooler than card has on it, you should be able to get +150 to +200 on the core clock and probably limit the memory clock to +50 to +75. Just watch the GPU temps and VRM temps if the card has VRM temperature sensors.

I learned this just yesterday: most games can only handle a certain OC on a certain card, before the game will crash. I think the reason for this is the game's graphic API that uses DirectX is expecting a certain response from that specific GPU model when rendering, and when an OC it too high, there's a timing dependent variable in the game engine that your overclocking caused something to come back too fast, too slow, not in order, or something similar to that.
m
0
l
August 26, 2014 1:07:59 AM

So how many mV of voltage should I increase at a time before attempting to increase the GPU overclock further? I downloaded the Sapphire TriXX utility and it says VDDC (GPU voltage?) is at 1225 which I think means 1.225. I pressed the + on it once and it jumped to 1226. Should I do 5 or 10 more mV of voltage at once or what?
m
0
l
Related resources
a b K Overclocking
August 26, 2014 9:04:44 AM

Yes, VDDC is the voltage. Keep raising the clock speeds until the drivers crash. They should automatically restart. That's when you know you need to increase voltage. I increase by 10mv once it becomes unstable. Don't go over 1.300 volts though. At that point you're .1 volt over stock and the VRMs are probably getting pretty toasty. I wouldn't keep an OC 24/7 if it's above stock voltages, I only leave it over-volted for shorter gaming sessions, but that's my own opinion. Btw, the Sapphire R9 270x Dual-X is factory overclocked, but their stock voltage is higher than it needs to be. You can try under-volting it to maintain stock clocks, but it would allow it to run cooler and use less power.
m
0
l
August 26, 2014 6:55:41 PM

It took a little while but I managed to boost the core clock to 1170mhz while the memory clock stayed at 1550mhz. This makes my GPU's core clock +100 over my GPU's factory clocks, and +150 over the AMD factory R9 270X. Memory clock is +150 over the stock memory. I increased the voltage incrementally from 1.225 to 1.230 (+ 0.005mV) which somehow enabled me to boost the core clock up 55mhz, but maybe it's because Sapphire increased the voltage more than necessary as you said. The maximum temps I get on load through 3dmark are 68 degrees, though it normally stays around the low 60s.

I think I've reached the memory clock limit as pushing the clock to 1560mhz results in very heavy artifacting in 3dmark within the first couple of seconds. I attempted to raise the GPU core clock to 1180mhz at 1.240mV but it appears to make my PC crash. I'm thinking this is the max I can go. I'm not sure whether increased voltage will eliminate artefacts or not. I'm at 1.230mV atm which gives me room to increase voltage but I don't think it would be a good idea to raise it close to 1.300mV.

GPU clock = 1170mhz (+100mhz)
Memory clock = 1550mhz (+150mhz)
Pixel fillrate = 37.4 GPixel's
Texture fillrate = 93.6 GTexel's
Bandwidth = 198.4 GB/s (6.4Ghz/6400mhz effective memory speed)

The GPU drivers haven't crashed; the only thing that happened is very heavy artefacts or my PC force rebooting. Should I leave the voltage where it is or can I push it a bit further? GPU core clock of 1200mhz would be a good goal to reach if possible.
m
0
l
a b K Overclocking
August 26, 2014 7:43:12 PM

It's odd your computer crashes overclocking your GPU too high, if I set my core clock to +300, the drivers just crash and restart. Maybe your PSU can't handle it? The artifacts you're seeing are caused by too high of a memory clock, if I set my memory clock to +800 the screen turns a gray color and I need to reboot to fix it. You can raise the voltages higher... but increasing it won't correct artifacts. I don't think you will be able to get any more performance out it even by increasing voltage to 1.3 volts. For those clock speeds, 1.230 volts seems a little high, maybe your card just doesn't OC as well as others. But +150 on Rx 200 series cards sounds about right, you can try going a little higher, but I would rather have my GPU last longer, be more stable, and consume less power than gain an extra 1 fps in a game.
m
0
l
August 26, 2014 7:51:51 PM

If I reduce the memory clock to 1450mhz or something would I be able to possibly increase core clock further?
m
0
l
a b K Overclocking
August 26, 2014 7:53:42 PM

You probably can, but I can't say how much because I don't have that card.
m
0
l
a b K Overclocking
August 26, 2014 8:08:05 PM

the memory overclock will show a performance increase in some synthetic benchmarks like unigine valley. but, in real world gaming on a single 1080p/1440p monitor the memory overclock will give you virtually no fps increase. save the stability for the core clock.... see how far you can push your core with no memory overclock.

i would use msi ab and unigine valley for initial testing, then a good hour or so of gaming for final stability testing. crysis 3 is probably still the go to game for system punishing and stability testing, some think it can fish out unstable cpu overclocks that are intel burn test/prime95 stable. but i do hear the tomb raider can fish out gpu instability too.
m
0
l
August 26, 2014 8:34:06 PM

It appears that I can only get the core clock to reach 1170mhz even if I leave the memory clock at 1400mhz. I'm assuming that 1170mhz core and 1550mhz memory is the highest I can go. I haven't attempted to increase the voltage further than 1.240mV with the core. Now it seems that sometimes when I try and modify the core clock from a fresh boot via Sapphire TriXX from 1170 to 1180mhz my PC freezes (I try to modify it again right after my PC boots from a crash), the display blanks and my fans speed up in my PC to max. Seeing as 1.230mV is apparently high voltage (when 1.225mV is factory) I guess I shouldn't risk things and leave things there.

I don't have anything running in the background at startup apart from Folding @ Home which is always left at idle.
m
0
l
a b K Overclocking
August 26, 2014 10:19:26 PM

Icaraeus said:
It appears that I can only get the core clock to reach 1170mhz even if I leave the memory clock at 1400mhz. I'm assuming that 1170mhz core and 1550mhz memory is the highest I can go. I haven't attempted to increase the voltage further than 1.240mV with the core. Now it seems that sometimes when I try and modify the core clock from a fresh boot via Sapphire TriXX from 1170 to 1180mhz my PC freezes (I try to modify it again right after my PC boots from a crash), the display blanks and my fans speed up in my PC to max. Seeing as 1.230mV is apparently high voltage (when 1.225mV is factory) I guess I shouldn't risk things and leave things there.

I don't have anything running in the background at startup apart from Folding @ Home which is always left at idle.


i wouldn't worry about the core voltage as long as your temps are ok. not sure about your particular card if you can read vrm temps, you really dont want them over 85c, if you can read them. usually vrm temps are a little bit higher than the core temps depending on the card.
m
0
l
August 26, 2014 10:46:48 PM

My PC doesn't appear to display any VRM temps of my GPU. Sapphire Trixx, MSI Afterburner, CPU ID HWMonitor and CCC don't appear to have it. Is there any software that may show VRM temps?
m
0
l
a b K Overclocking
August 26, 2014 11:25:46 PM

If your GPU had VRM temperature sensors, one of those programs would've displayed it. You can try GPU-Z, but I doubt that will work. That voltage isn't high enough to cause VRM issues on your card anyways.
m
0
l
a b K Overclocking
August 27, 2014 9:32:09 AM

what are your maximum gpu temps when running the 1.24v@1170mhz? or whatever clocks volts you are running
m
0
l
August 27, 2014 2:02:07 PM

The GPU gets to 68 degrees at 1170mhz/1550mhz @ 1.23mV.
m
0
l
!