Overclock.net › Forums › Graphics Cards › AMD/ATI › [Official] Polaris Owners Club
New Posts  All Forums:Forum Nav:

[Official] Polaris Owners Club - Page 58

post #571 of 4133
So is there a review of rx 470 compared to r9 380 clock vs clock?
post #572 of 4133
I decided to try to get 1445 stable again. I was able to do firestike ultra at 1248 and I was able to do heaven at that voltage. Before I had to do 1268 to do 1445. Not sure what's changed other than some settings on ram and processor. What do you guys think? 1248 safe for my beleaguered pci-express power distribution? Not that I have a real need for extra performance right now. Games run fine. Though I do drop to 50fps at times with vsync on in Rise of the Tomb raider. (That's what prompted this. ) Thoughts? last time my motherboard started burning was at 1.268 in firestrike ultra stress test. So I'm only 20mv away from that. I was at 1445 then too. I just hate to leave mhz on the table.

Here's what I'm currently running:

4790 @ 1.453volts 5.0Ghz 4.2Ghz cache
gskil trident x ddr3 1600 at 2200 10,11,11,24,1t @ 1.65v I can run 2400 but the timing is bad and I lose points in 3dmark compared to 2200.
rx480 1445 @ 1.248v & 2210 auto vmem

http://www.3dmark.com/3dm/14130756?
Edit
http://www.3dmark.com/spy/277815
end edit








Unigine Valley Benchmark 1.0







FPS:
60.2
Score:
2521
Min FPS:
30.9
Max FPS:
115.9

System






Platform:
Windows 8 (build 9200) 64bit
CPU model:
Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz (3998MHz) x4
GPU model:
Radeon (TM) RX 480 Graphics 16.200.1035.1001 (4095MB) x1

Settings






Render:
Direct3D11
Mode:
1920x1080 8xAA fullscreen
Preset
Extreme HD

Powered by UNIGINE Engine


Edited by greytoad - 8/14/16 at 8:31pm
post #573 of 4133
Honestly, pushing really hard for that last 25-50mhz does not seem to do much for performance. I was seeing performance regression past a certain point.

Truthfully its best left on the table. Its nice to bench at the bleeding edge, but for 24/7 usage ~1400 should not make much difference vs ~1450.
Edited by KarathKasun - 8/14/16 at 8:50pm
μRyzen
(12 items)
 
Mini Box
(4 items)
 
 
CPUMotherboardGraphicsRAM
Ryzen R5 1400 MSI B350M Gaming Pro Zotac GTX 670 4GB G.SKILL FORTIS Series 8GB (2 x 4GB) 
Hard DriveCoolingOSOS
WD Green 3tb Wraith Stealth Windows 10 Debian 8.7 
MonitorKeyboardPowerMouse
ViewSonic VX-2257-8 Chinese backlit mechanical Kingwin 850w Chinese laser optical 
CPUMotherboardGraphicsRAM
Athlon 5350 Asus AM1I-A EVGA GTX 750 Ti SC 2x4GB DDR 3 1333 
  hide details  
Reply
μRyzen
(12 items)
 
Mini Box
(4 items)
 
 
CPUMotherboardGraphicsRAM
Ryzen R5 1400 MSI B350M Gaming Pro Zotac GTX 670 4GB G.SKILL FORTIS Series 8GB (2 x 4GB) 
Hard DriveCoolingOSOS
WD Green 3tb Wraith Stealth Windows 10 Debian 8.7 
MonitorKeyboardPowerMouse
ViewSonic VX-2257-8 Chinese backlit mechanical Kingwin 850w Chinese laser optical 
CPUMotherboardGraphicsRAM
Athlon 5350 Asus AM1I-A EVGA GTX 750 Ti SC 2x4GB DDR 3 1333 
  hide details  
Reply
post #574 of 4133
Quote:
Originally Posted by KarathKasun View Post

Honestly, pushing really hard for that last 25-50mhz does not seem to do much for performance. I was seeing performance regression past a certain point.

Truthfully its best left on the table. Its nice to bench at the bleeding edge, but for 24/7 usage ~1400 should not make much difference vs ~1450.

Totally agree In my case more than 1.420 MHz its a increible voltage brutality to get stability so, more than 1400 mhz deppends too much of the asic, and how much cooler are your VRM.-

Im now with 1420 and 1.23 volts and everything its very cold.-
post #575 of 4133
I'm seeing a slight uptick in performance. I'm only going up 1 percent on my overclock so I wouldn't expect more than that. that was 35 or 36 points higher in firestrike ultra and 6 points higher in valley (only 2 points more if you count my run at 5.1ghz.) It was about 30 points higher in TimeSpy. Interestingly right now I seem to have the top score for people with both the 4790 and only 1 rx480 in timespy. I don't see how that's possible. Maybe I"m reading the chart wrong. People with newer cpus and higher overclock bench runs are taking all the top spots of course.

You're right about that not being much of an improvement. It feels safe-ish because people are saying 1.2-1.25 is safe. I'm in the high end of that range. Here's what hwinfo looks like running Heaven at that:




vdroop is sending the VDDC much lower than the VID setting of course. I'll play with it and monitor power. I'm not throttling but I'm probably running into power limit. I'll do a run with gpuz.
post #576 of 4133
I'll sleep on it. You're probably right. I have a replacement board already here but I don't have time to re-do my loop right now.
post #577 of 4133
do to a math error on my part I am actually running a VID of 1240.5mv That's 28mv under the last time my motherboard smoked. No idea why it's stable with so much less voltage at 1445 speed now. Maybe it's burned in? It makes no sense. I couldn't pass heaven let alone 3dmark at that voltage before. I can kinda get why things improved with my ram. I'm running a newer firmware on my motherboard than when I initially tested the ram 2 years ago. The video card. I've never had a video card improve it's overclocking ability before.
post #578 of 4133
Quote:
Originally Posted by greytoad View Post

do to a math error on my part I am actually running a VID of 1240.5mv That's 28mv under the last time my motherboard smoked. No idea why it's stable with so much less voltage at 1445 speed now. Maybe it's burned in? It makes no sense. I couldn't pass heaven let alone 3dmark at that voltage before. I can kinda get why things improved with my ram. I'm running a newer firmware on my motherboard than when I initially tested the ram 2 years ago. The video card. I've never had a video card improve it's overclocking ability before.

Warning: theory ahead!

FinFet processes have a much more significant change of properties as the product ages. This is because you now have three types of conductance changes occurring rather than one for each transistor. You can have an actual skin effect changed (traveling of electrons along the perimeter), a normal-path change (where the resistance to electron flow changes where the electrons normally flow), and you can even have new path formation (electrons finding a way to flow through a new path in the transistor - think of lightning).

Only the normal path change applies to more traditional transistors, though a few new paths can form, they are much less common.

It is this reason why AMD has invested in anti-aging voltage control technologies for their FinFet GPUs - the amount of voltage needed will change over time (not always for the better).
post #579 of 4133
Interesting. The only other thing that's changed is the driver version. I was running 16.7.2 before and I'm running 16.8.1 now. There were VDDC and vdroop setting changes between those drivers it seemed. VDDC output is much more stable now. Maybe that helped a bit. Vdroop can be a problem, but it shouldn't cause that much a difference in stability. When 16.8.1 first came out I checked if I could run stable at a lower voltage since the VDDC was more stable. It couldn't. So I don't know You're theory is the only plausible one left.

I guess there was one other change. I wiped my drive and reloaded once since then.

edit: I decided to step back to 1440. While it didn't burn in 3dmark firestrike ultra stress test I could smell it a tiny bit. gpu only power usage went to 265 watts. Which means the real power usage is probably bumping close to 300 watt absolute insane limit for the reference card.
Edited by greytoad - 8/15/16 at 8:16am
post #580 of 4133
I don't think it's very safe to judge your overclock based on if the motherboard starts to smell burnt or not.
Power Bac G5
(6 items)
 
  
CPUGraphicsRAMHard Drive
Intel i7 6950X EVGA GTX 1080 Ti GSkill 32GB 4133Mhz Samsung 960 Pro 1TB 
CoolingMonitor
Corsair H110i ASUS PG278Q 
  hide details  
Reply
Power Bac G5
(6 items)
 
  
CPUGraphicsRAMHard Drive
Intel i7 6950X EVGA GTX 1080 Ti GSkill 32GB 4133Mhz Samsung 960 Pro 1TB 
CoolingMonitor
Corsair H110i ASUS PG278Q 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD/ATI
Overclock.net › Forums › Graphics Cards › AMD/ATI › [Official] Polaris Owners Club