As if the 290x is capable of overclocking at all. 15% under air or 16% under water according to hwbot which seem way higher than actual OCs.
V8 engines are not balanced, only V6 and V12/W12
It should be obvious why. Nvidia has posited that this same old tired reference cooler and PCB is suddenly deserving of a $100 premium now. So it's being put to the scrutiny of that price hike and failing to meet that premium miserably.
If not for the founders blunder, the card would have been much better received. They really screwed up the whole marketing,out of character for nvidia and they are reaping the bad PR of the mistake.
Seems like you don't have either.
No. They are synced but are not sharing load per se.
Start with what is MOSFET and how it works. Then you would know the reason to have multiple phases in power delivery circuit.
First is for scalability reason. 1080 MOSFETs can push only 250amps at room temperature.
290x can handle 5x160amps=800amps at room temp but only 5x70amps@125'C (less than a half)
https://www.youtube.com/watch?v=6i6TPriNXis
So how much current can 1080 VRM push in realworld? Suddenly it becomes obvious why it is powerthrottling.
Second reason is ripple. You seem to not mention in your post how VRM phases and FETs work.
"Output ripple voltage really depends on how well the circuit has been designed (even if they are all doing the same thing). Using expensive components can help but best performance comes from the design equations themselves."
Really? If this is not about VRM phases I don't know what it is about.
First and for the most, the components quality. High quality caps and inductors are required to even dream of good ripple. You want to generate as little ripple as possible - that is why a lot of small VRM phases are better than a single beefy one.
Third is ripple suppression which is lacking on a GPUs (inductor+capacitor) compared to for example power supply. But it is tied to how much ripple both of these devices cope with and the PSU have a lot of ripple to smooth out.
For the starters, think of VRM phase as a Cylinder in a car's engine.
We had 2 cylinder engines which were running like tractors. Anyone familiar with the smooth sound of V6 or V8?
Obviously the high end AIB cards will be better, they're almost always vastly superior.
Doesn't mean the ref PCB isn't well engineered.
I'm not sure if this have been already posted or not.
So, freqs falls to base clock after 20 minutes of load... what a joke.
:thumbsup:
I think some people defending the FE cards clearly cannot separate GP104 from GTX1070/1080 FE video cards. I think GP104 is a great little GPU ASIC, but 1080 FE is an overpriced engineering failure that shouldn't be purchased by 98% of PC gamers or so. 1920 CC 1070 will beat a stock 980Ti while using at least 100W less power, have superior feature set and more VRAM. Another great little GPU that will need AIB cards to max its full potential. Once AIBs release cards that can reliably hit 2.2-2.3Ghz overclocks, the FE cards are dead in the water.
Good for you. I hope you have well paying job. Were you in any way involved in designing this board?What if I told told you that my profession is an electrical and electronic engineer.
Would you like to continue the discussion on this topic further because to me clearly you just want to disagree regardless.
Plus your not going to convince anyone that your an expert in the area after an afternoon session on the internet about VRMs and power supplies on google.
FuryX was an overclockers dream - when looking at the pcb and cooling solution compared to 1080fe.
1080fe is an overclockers dream too! But that dream is a Nightmare, because your GP104 is chained, closed in cage, which is again cuffed in chains and dropped deep in the ocean.
So, a Fury X cant be overclocked but this okay because the PCB was designed for overclocking.
A GTX1080 can be overclocked 15%+ but this is bad because... because!
Fury OC's 10% on average, why do you keep posting this non sense? The 1080's "premium" cooler can't even run boost clocks let alone OCing without throttling.
Wait a second? A 15% overclocking on "premium" PCB with falling to base clock after 5 mins of gameplay? YEAH, premium design by nVAnd the premium cooler and the premium PCB allows for 15% overclocking - twice as good.
Haha, 10%. Good joke. :thumbsup:
It is between 5% to 7%.
And the premium cooler and the premium PCB allows for 15% overclocking - twice as good.
The maximum achievable 1,886 MHz, the GeForce GTX 1080 Founders Edition are more of a theoretical value: In games you never see this clock. There, the highest ever measured frequency was during the tests 1,835 MHz, but was also the clock after a few seconds history. In demanding titles 1.785 MHz has finally placed as the highest, realistic clock out.
Single fastest card in the world by a large margin. "Mid range".
Is this shilling or have I missed something?
Thanks for the laugh
http://forums.anandtech.com/showpost.php?p=38247456&postcount=153
http://www.computerbase.de/2016-05/geforce-gtx-1080-test/6/
Only a few posts up show proof that the "premium" PCB and cooler can't even handle the boost clocks let alone OCing.
Can't even handle gaming for 20 minutes @ 100% fan. Guess you'll enjoy your new benchmarking card since it's apparently not built for gaming.
Haha, 10%. Good joke.
It is between 5% to 7%.
Doom @ 2000MHz: https://www.youtube.com/watch?v=UsbJXzqtdb0
That is a 24% overclock over base clock. Looks just fine. :thumbsup:
Hmm, it reminds me something.. What happens when a card reaches a temp limit of 83°C? Bam! A base clock and Dmitry is well aware of this.. so he stopped recording a gameplay after 82°C.Yep, and he stops playing once it hit 82c
@ ~2:30 it was @ 76c
@ ~3:00 it was @ 77c
@ ~3:30 78c...
...
@ ~6:30 - 81c
@ ~7:00 - 82c
@ ~8:00 he stops playing
Wow whole 5min 30 secs of gameplay, amazing! Good thing he didn't test for 20 minutes in an actual case like computerbase and others that noticed throttling issues.
Seems like you don't have either.
No. They are synced but are not sharing load per se.
Start with what is MOSFET and how it works. Then you would know the reason to have multiple phases in power delivery circuit.
First is for scalability reason. 1080 MOSFETs can push only 250amps at room temperature.
290x can handle 5x160amps=800amps at room temp but only 5x70amps@125'C (less than a half)
https://www.youtube.com/watch?v=6i6TPriNXis
So how much current can 1080 VRM push in realworld? Suddenly it becomes obvious why it is powerthrottling.
Second reason is ripple. You seem to not mention in your post how VRM phases and FETs work.
"Output ripple voltage really depends on how well the circuit has been designed (even if they are all doing the same thing). Using expensive components can help but best performance comes from the design equations themselves."
Really? If this is not about VRM phases I don't know what it is about.
First and for the most, the components quality. High quality caps and inductors are required to even dream of good ripple. You want to generate as little ripple as possible - that is why a lot of small VRM phases are better than a single beefy one.
Third is ripple suppression which is lacking on a GPUs (inductor+capacitor) compared to for example power supply. But it is tied to how much ripple both of these devices cope with and the PSU have a lot of ripple to smooth out.
For the starters, think of VRM phase as a Cylinder in a car's engine.
We had 2 cylinder engines which were running like tractors. Anyone familiar with the smooth sound of V6 or V8?
The PCB might be well-engineered but the card is limited by TDP headroom via the 8-pin connector or NV's Greenlight. Even though the connector in theory can draw more power, NV is clearly not allowing for it to happen as Overclockers.ru reported 220W at 2000mhz boost. Not that it matters because the reference blower cannot sustain advertised boost clocks in a highly popular Fractal Design R5. It may not be a PCB failure, but it's a videocard SKU engineering failure. TPU found the same.
Why don't you want to admit the facts?
This card cannot maintain advertised specs after minute of gaming, but we were shown sub-70C with a 2.1Ghz overclock? Why shouldn't the PC community call out such marketing lies?
The house AC unit broken yesterday and the thermometer showed 83F (28.3C) ambient.
Right now I have i7 6700K + R9 295X2 1018mhz + R9 390 @ 1150mhz, with the 3 GPUs running at 100% usage 24/7.
All in the same case
R9 295X2 GPU1 = 74C (-40mv undervolt)
R9 290X GPU2 = 71C (-40mv undervolt)
Open air cooled R9 390 OC = 81C (VRM 1 = 72C, VRM2 = 82C)
Kill-a-Watt shows 863W from the wall. Under 28-29C ambient conditions, I wouldn't be surprised if GTX1080 Tri-SLI would throttle to 1550-1600mhz boost max at reasonable noise levels, while using <500W of total system power.
NV should have designed a $599 reference blower card and a $699 AIO CLC reference card. Instead, they added a $100 premium for marketing aluminum shroud and early launch date availability.
The fact that a certain stubborn section of the PC community keeps defending the Blower Myth in 2016 is an insult to the PC Enthusiast community. Unfortunately the PC gamers before me ruined it by repeating this lie for decades and now it's taking another 1-2 decades to reverse their lies and reveal the truth that blowers were never superior for 98% of PC gamers. What's mind-blowing is that in light of real world tests that prove that blowers are outdated technology for most PC gamers, they are still being defended and with a $70-100 premium too. It's insanity that after you apply "aluminum shroud" lipstick to inferior tech, it's suddenly premium.
It's only going to get worse for a 250W TDP GP102 card. It's going to require > 50 dBA operation to ensure that a 230-250W TDP 1080Ti doesn't thermal throttle, while the best AIB 1080Ti cards will manage sub-30 dBA with higher GPU boost clocks.
Reminds me of people who were adamant that you still needed to cover the whole heat shield with thermal paste after 2010 or so. Acting like it was back in the Athlon days and worrying about cracked cores.
Besides the cooler being inadequate at keeping a <200W card cool, which obviously was under engineered, the PCB appears to be bare bone basic too. Thus the singe 8p vs. 2x 6p that is typically used in a more robust design. It looks like lots of corners have been cut. Except for price, of course. Add that it's a "backward looking" design more suited for a dying DX11 than current API's and nVidia's propensity to stop optimizing as soon as a product goes EOL. Anyone who buys this card knowing what they are getting I have no sympathy for.
The writing is on the wall. These cards are going to crumble with DX12. $700 for a cheaply made outdated design.
What malarkey is this? I didn't know I needed sympathy from AMD supporters when I want to spend my money.
People on this forum need to worry more about themselves. Woof. So many AMD posters with their crocodile tears. Keep it to yourself.