Disappointing FuryX overclocking

[fullinfusion]

I feel your pain @xkm1948

I just dumped my Fury X today and had my MSI TFIV 290 in the storage bin.. I lucked out and found a mint PCS 290X and man does it clock...

What temps were you getting? My XFX Fury X would hit about 70c @ 1200mhz but I didn't screw with the memory.. The system didn't like the memory being messed with but what ever the fury is good if you leave them alone but like me I cant lol..

Sure loving twin R9's again.... I'm back as Johnny said haha!

And A bitch thread on trixx is on the way in 3,2,1 Bamm!

 

[RealNeil]

This AIO cooler is not doing a great job in terms of keeping the FuryX temp under control

Does it have twin fans for push/pull?

 

[xkm1948]

Does it have twin fans for push/pull?

Nope. Stock FuryX only have one 120mm radiator with a gentle typhoon fan.

 

[RealNeil]

So would two high output/low noise fans such as a pair of Cougar Vortex PWM fans in Push/Pull control temps a little better?

 

[fullinfusion]

So would two high output/low noise fans such as a pair of Cougar Vortex PWM fans in Push/Pull control temps a little better?

A tiny bit but don't expect the world

 

[RealNeil]

I would think that it may be worth a try.
I added them to my Cooler Master Seidon AIO (AMD FX-9590, 220W CPU) and had a good reduction in temps.

 

[vega22]

the cooling is designed to keep it in check at stock clocks, not overvolted.

once they let them make custom pcb, uprated vrm and cooling i'm sure oc results will change.

 

[PCGamerDR]

I tried 100% fixed fan speed. It was horrible. At the same time I can finally get to 1150/575 stable through FSE looping. So temperature is the key. This AIO cooler is not doing a great job in terms of keeping the FuryX temp under control.

I will look into custom water cooling if I have the time in the future. For now a 4.2G 5820K and a mildly overclocked FuryX is more than enough to meet my needs.

Well now that you have a max core/mem clocks, do further testing while decreasing core volt, it should lower temps without ramping up the fans.

 

[xkm1948]

Well now that you have a max core/mem clocks, do further testing while decreasing core volt, it should lower temps without ramping up the fans.

+54mv at 1150/575 will instant crash the system.

Now you mentioned it, I wonder if it is safe to use +72mv over prolonged period of gaming.

 

[Cybrnook2002]

Biggest rival will be heat. PM me if you go into the market for a block. I may be able to help you.

 

[xkm1948]

Biggest rival will be heat. PM me if you go into the market for a block. I may be able to help you.

Too much of a hassle to go customized water cooling. I opted for Noctua D15 just to avoid all the problem of AIO coolers. I may look into re-pasting the VRM heat sink though.

 

[Jism]

I think the Fury is an excellent oc'er, but you need a stable bios for that. Second; some guy managed to get 1450 MHz GPU and 1000MHz HBM. Pretty amazing results if you ask me.

I think regular Fury cards have a hardware lock or something, 72mv is'nt a huge increase for a steady OC as well.

http://wccftech.com/asus-radeon-r9-...y-unlocked-fury-features-1-tbs-bandwidth-ln2/

I managed todo a 100% OC on a Radeon 2600 Pro GPU, with almost 2.0V on the GPU and a clockspeed of over 1250MHz. I resoldered all caps on the card completely with bigger ones, upgraded it's voltage supply and kept it going at -25

 

[Cybrnook2002]

Too much of a hassle to go customized water cooling. I opted for Noctua D15 just to avoid all the problem of AIO coolers. I may look into re-pasting the VRM heat sink though.

Custom loop really isn't that bad at all. Only pain is if you don't put a point in your loop to drain it when you want to change something (If you do, your good to go). Custom you can have a much larger radiator and manually adjust flow rate to your liking. And push/pull to your hearts desire.

Off topic, I just picked up a nano from tigerdirect yesterday. After boa cash back deal, it will come out to $320. I plan on throwing that in the wifes pc, but will see what kind of clock I can get out of it first.

 

[GhostRyder]

Custom loop really isn't that bad at all. Only pain is if you don't put a point in your loop to drain it when you want to change something (If you do, your good to go). Custom you can have a much larger radiator and manually adjust flow rate to your liking. And push/pull to your hearts desire.

Off topic, I just picked up a nano from tigerdirect yesterday. After boa cash back deal, it will come out to $320. I plan on throwing that in the wifes pc, but will see what kind of clock I can get out of it first.

How the heck did you do that??? Heck I would pick up a two or three for that price!

 

[Cybrnook2002]

For whatever reason they had the visiontek nano for $403 yesterday. Then with boa, I get %20 off tiger direct (maximum $100).

They have an msi fury x (with star wars) for 558. However, cross your fingers that nothing goes wrong with the order. TD custom service is quite literally to die for.

 

[Kanan]

There is the possibility that Fiji is similar to Maxwell. Maxwell scales terribly with added voltage unless it's sub zero. Something to do with the extra heat causing problems for stability. It needs cold (less 'vibration') states to operate. Fiji might be running at the same levels, therefore voltage increases might not help without substantial exotic cooling.

No, Fiji under LN2 achieved a max overclock of about 1500 with some user on a Asus Fury Strixx card (Maxwell max OC under LN2 is over 2000 MHz), that was modded to have a OC bios + 4096 shaders active, because that board has way better OC potential, due to the better PCB compared to reference AMD PCB. In the same thread the user achieved HBM 1 GHz OC, which resulted in 1 TB/s bandwidth btw. That was more of a "wow" factor I think.

The problem with Fiji or all AMD cards is, they are compute/game cards, not specialized to gaming like Maxwell, and therefore way more complex than Maxwell - that's why the OC is so limited I guess. They are best compared to Kepler, which isn't as good overclock as Maxwell too with max clocks of about 1200-1300 MHz under air which is way below Maxwell. They are same generation btw. - but Maxwell is 1 generation ahead, I don't see GCN 1.2 as a whole new architecture, just as a incremental upgrade, but Maxwell surely is.

 

[Jism]

If you want speed, just crossfire the damn thing. Double the fun in general, call it a day.

 

[PCGamerDR]

+54mv at 1150/575 will instant crash the system.

Now you mentioned it, I wonder if it is safe to use +72mv over prolonged period of gaming.

Try something between 54 and 60 maybe ?

And as for a game try GTAV, it has been my favorite game so far to test both GPU and CPU OCs and actually a few days ago i found out that GTAV stresses my CPU more than AIDA64 does(i find this extremely funny :"D...), max aida temps being 56*C(stable, no throttle) and gtav max temps being 62*C(stable, with throttle) that friggin game made me do more volt tweaking for my cpu and i found my sweet spot thanks to it and i love it!!!, temps now down to 58*C on GTAV and 53*C on aida .

 

[xkm1948]

I did some more tweaking around. It seems the VRM is bottle-necking the OC. By sticking with 0% power limit increase my overclock gets a lot more stable. This is pretty weird. All the overclocking guides so far have suggested sliding the power limiter all the way up. If I do that along with +54mv at 1175/575 I will crash my system in the first test of FSE. However if I keep power limiter at stock the test finishes just fine. The final result did not show any GPU core clocking throttle. I am guessing that the VRM is very sensitive to temperature rise. By sliding up the power limiter the VRM will overheat shortly with volt assisted overclocking, thus crash my system

 

[EarthDog]

They certainly aren't garbage, but they are pretty damn far from an overclockers dream as some AMD dude claimed.

Not sure how many times I posted this here at TPU, but here goes again...

When he talked about it being an overclockers dream, he was talking in the context of the cooler and the amount of heat it could dissipate.

All the overclocking guides so far have suggested sliding the power limiter all the way up. If I do that along with +54mv at 1175/575 I will crash my system in the first test of FSE.

Do that and don't just jack up the voltage.

 

[xkm1948]

Also the placement of radiator hose is important. According the official installation guide, it should be mounted so that the hoses come out from the bottom of the radiator. I have been installing it wrong this whole time. Just switched the orientation. One thing I noticed instantly is the damned bubble noise from the pump is gone!

 

[xkm1948]

Not sure how many times I posted this here at TPU, but here goes again...

When he talked about it being an overclockers dream, he was talking in the context of the cooler and the amount of heat it could dissipate.

Do that and don't just jack up the voltage.

Dude, I have to do this. I know Lisa Su looks like a total dude. But she is actually a WOMAN. So please stop calling her "he."

Yeah she did say it should be an overclocker's dream.

 

[EarthDog]

No, I meant someone else... the engineer. HE is the one that I THOUGHT said that (I was sitting right there in the second row). I also had a deep dive with HIM (the engineer, not HER the CEO) who stated that.

 

[Fluffmeister]

Not sure how many times I posted this here at TPU, but here goes again...

When he talked about it being an overclockers dream, he was talking in the context of the cooler and the amount of heat it could dissipate.

“You'll be able to overclock this thing like no tomorrow,” AMD CTO Joe Macri says.

 

[EarthDog]

“You'll be able to overclock this thing like no tomorrow,” AMD CTO Joe Macri says.

Yep. Joe Marci. Thanks!

But that is taken out of context. I have it in my notes somewhere, LOL!