Associate
- Joined
- 29 Jun 2016
- Posts
- 529
Hi All,
I have recently bought a Sapphire Fury Tri-X 2nd hand. What an absolutely fantastic card! I can really see why these are still in demand. The card hasn't breached 60C, or made a whisper for that matter, where my old R9 290 Tri-X regularly hit 70C and made a racket.
A few photos:
It came unlocked with 4096 SPs, which is quite tasty, but it looks to have a max stable overclock around 1070/545 (+0mV). I thought the max overclock might be limited by the unlocked SPs, but it appears to be the same for 3574 SPs.
I was following the advice of AMDMatt about reading the resistor states using AIDA64 and found that at the stock 1000MHz, DPM7 has a voltage of ~1.18.
What was weird though, was that the DPM7 state voltage changes as I use MSI Afterburner to overclock the card. DPM7 got as high as 1.25V at 1050MHz w/ 4096 SPs and ~1.2 at 1050MHz w/ 3574 SPs. Some "clever" logic there?
One thing I noticed was that my overclock instability came about under certain circumstances: When I first run Firestrike/Timespy, the GPU Core clock jumps to the target frequency, but the core voltage is often several steps (DPM3-6) lower - sometimes exactly the value of a DPM state, sometimes a random value between states. I haven't observed instability at higher clocks (1110/545) when the card ran at the DPM7 voltage + offset.
Is this normal behaviour, or have I been sold a slightly borked card?
Not really surprising the core isn't stable trying to run 1110MHz at <1.18V...
Volt mod time?
Oh, I "measured" the voltage/core/etc using HWInfo, MSI Afterburner and GPU-Z.
Apprecaite any input!
I have recently bought a Sapphire Fury Tri-X 2nd hand. What an absolutely fantastic card! I can really see why these are still in demand. The card hasn't breached 60C, or made a whisper for that matter, where my old R9 290 Tri-X regularly hit 70C and made a racket.
A few photos:
I was following the advice of AMDMatt about reading the resistor states using AIDA64 and found that at the stock 1000MHz, DPM7 has a voltage of ~1.18.
What was weird though, was that the DPM7 state voltage changes as I use MSI Afterburner to overclock the card. DPM7 got as high as 1.25V at 1050MHz w/ 4096 SPs and ~1.2 at 1050MHz w/ 3574 SPs. Some "clever" logic there?
One thing I noticed was that my overclock instability came about under certain circumstances: When I first run Firestrike/Timespy, the GPU Core clock jumps to the target frequency, but the core voltage is often several steps (DPM3-6) lower - sometimes exactly the value of a DPM state, sometimes a random value between states. I haven't observed instability at higher clocks (1110/545) when the card ran at the DPM7 voltage + offset.
Is this normal behaviour, or have I been sold a slightly borked card?
Not really surprising the core isn't stable trying to run 1110MHz at <1.18V...
Volt mod time?
Oh, I "measured" the voltage/core/etc using HWInfo, MSI Afterburner and GPU-Z.
Apprecaite any input!
Last edited:
