XFX GTR RX 480 running 1500 core / 2100 memory on AIR.

CSI_PC said:
It DOES adjust the VRM loads because of the phases, and that is shown in the earlier charts where it is 50/50 to the latest where they no longer are.
Case in point the July article you link even says this as a section: 'Reconfiguring the power phase controller'
It means more cycle supported for 3 phases and less for the other 3 phases, putting more strain on one group and why you would be limited what you can do outside of reference spec.

If it was simple, they would also had tested OC and compared that as well like they did with reference power, but they did not, nor could AMD distribute much more between mainboard phases and auxiliary.

Before the update reference for 4k Metro Last Light is 79W for PCIe slot, but 4% OC took that to 90W to 95W and real sustained clocks of 1250MHz and 1.05V to 1.06V.
The 16.7.1 reduced the 79W by 8% to 71W and increased by 12% 6-pin (adding more pressure to 3 phases of VRM but acceptable in reference spec), and with compatibility on reduced by 12% to 67W (very close to PCIe SiG spec for 12V at 66W) and increased by 5% 6-pin , none of this tested OC.

Well I learnt something new,I must admit I thought the driver needed to enable the phase power distribution because when used it adds pressure to 3 phases for any models using this circuit-architecture design, and possibly why no sites (only 3 with the right gear and methodology) who can accurately measure the watts has tested this situation with OC (maybe AMD request *shrug*) - interesting that both PCper and Tom's Hardware did not do an OC test when testing the AMD solution even though they did for the launch review and analysis on its behaviour.

Easier to read from Tom's Hardware but similar trend as PCPer:
01-Power-Consumption_w_727.png



Well this raises even more why need to see the XFX measured by either Tom's Hardware/PCPer/Hardware.fr, need to see how the phase distribution behaves in normal XFX GTR custom clocks and then OC, still as I say the mainboard is still going to be at least 100W if OC to 1.3V and we need to see how much additional demand it creates for the 3 other phases.
Of course that comes down to whether Buildzoid made a mistake or not when doing his breakdown and identifying the phase split as 50/50 for the XFX.

BTW my initial post that has everyone in uproar regarding watching out for behaviour on mainboard power only says
Context by half is GPU core phases not necessarily watts although they match.
Anyway to re-iterate again my post relates more to running beyond AMD's standard 480 spec meaning OC (such as the 4% and much higher).

The 1.3V related posts are more about being realistic on real-world 24/7 clock-voltages with reasons explained multiple times now including also why it is no longer great for Nvidia and Pascal compared to Maxwell.
Cheers

Edit:
Here is an image showing how the phases are split 50/50 between mainboard and auxiliary for the 'reference' launch 480.
rx480-phases.jpg

Cheers
Click to expand...
Ok let me try this again.

Here is your first post here in its entirety:
You did notice the 1.3V?
Good luck running that voltage all the time and not suffering a failure longer term, the power leakage-energy waste-draw must be very high (need one of the 4 sites that has the lab type measurements to accurately report this)
The absolute ceiling so far is around 1.4V-1.45V for Polaris according to some of the best extreme OCers in terms of voltage tolerances and that is with LN2 and only for benchmark world record type attempts, usual suggestion is a max of 1.2V to 1.25V (usually water and ensuring power stage is fully cooled).

This generation of 14/16nm are far from the 28nm in terms of the long term voltage abuse they can take, and even the short term; combination of density-silicon die thermal-electical characteristics-die size.
Case in point Maxwell could volt to 1.5V, now with Pascal its voltage tolerance with LN2 and only for benchmark world record type attempts is around 1.3V
The voltage tolerance is a bit higher with Samsung, but that is still only around what I mention in the 1st paragraph.

But it does show how good the XFX cooling system is, albeit having to run at 100%.
Click to expand...

This is why everyone is at your throat. You started in with the negativity right from the start. The last line has a little positive although seems retracted with saying at 100% fan speed. At any rate as I have to keep hammering into the Anti-AMD brigade in the AMD CPU forum section, This is where we should be free from the constant negative AMD bashing. Even here no one was posting the results are indicative of all XFX 480s. You just kinda chimed in on your own and even after the push back you just keep on digging. That hole is getting real deep, not sure if you can get out.
 
JustReason said:
Ok let me try this again.

Here is your first post here in its entirety:


This is why everyone is at your throat. You started in with the negativity right from the start. The last line has a little positive although seems retracted with saying at 100% fan speed. At any rate as I have to keep hammering into the Anti-AMD brigade in the AMD CPU forum section, This is where we should be free from the constant negative AMD bashing. Even here no one was posting the results are indicative of all XFX 480s. You just kinda chimed in on your own and even after the push back you just keep on digging. That hole is getting real deep, not sure if you can get out.
Click to expand...
I was specifically talking about those who are debating the PCIe mainboard situation not the 1.3V, did I quote the 1.3V as the 'uproar' situation, no.
You guys are being seriously defensive and towards AMD for some reason, because you are not willing to discuss facts that balance out the 1.3V as several posts were intimitating this is potentially normal behaviour for a custom OC XFX.
I do not get this issue when balancing out Nvidia details in their topics, nor when I created the topic back in July on extreme well known OCer reached 1500MHz OC 480 on water (all safeguards removed) with its caveats, I mean what is wrong pointing out the OC had to be done with a 100% fan in this thread that you now seem to have an issue with.
On another forum I know quite a few are sending back the XFX because the fan is too loud even without such a fan profile.
These are facts that need to be considered, because many here are looking for an OC they can use 24/7 and not just benchmarking-world records.
Cheers
 
Last edited:
Thinking back to the July OC on water.
One difference between the XFX and OC water one is the voltage improvement for XFX.
The 'reference' watercooled card required 1.35V for 1500MHz, and now with XFX it is 1.3V.
Can be for a range of reasons.
Cheers
 
thesmokingman said:
The interesting thing is that these high clocking 480s are only now just matching 290x's from two years ago, albeit they are doing it with much less heat and fan fare but require 200mhz more clock.
Click to expand...

The RX-480 is usually faster than the 290x at 1080p, roles are reversed at 4k. 1440p they are pretty even.

Bageland2000 said:
Yeah, the temps on that are about the same as my 290 running on a custom water loop. Almost want to side-grade just to cool my loop down a bit....
Click to expand...

I just did this not too long ago. It does run a bit cooler, but tthe difference isn't very much, I lost maybe 2c water temps at full load.

CSI_PC said:
You did notice the 1.3V?
Good luck running that voltage all the time and not suffering a failure longer term, the power leakage-energy waste-draw must be very high (need one of the 4 sites that has the lab type measurements to accurately report this)
The absolute ceiling so far is around 1.4V-1.45V for Polaris according to some of the best extreme OCers in terms of voltage tolerances and that is with LN2 and only for benchmark world record type attempts, usual suggestion is a max of 1.2V to 1.25V (usually water and ensuring power stage is fully cooled).

This generation of 14/16nm are far from the 28nm in terms of the long term voltage abuse they can take, and even the short term; combination of density-silicon die thermal-electical characteristics-die size.
Case in point Maxwell could volt to 1.5V, now with Pascal its voltage tolerance with LN2 and only for benchmark world record type attempts is around 1.3V
The voltage tolerance is a bit higher with Samsung, but that is still only around what I mention in the 1st paragraph.

But it does show how good the XFX cooling system is, albeit having to run at 100%.
Cheers
Click to expand...

the absolute ceiling is 1.5 volts? lol nah 1.2-1.25 volts being max suggested? nah

To correct your statement, 2.0 volts is the max under ln2, and 1.3-1.35 volts is the max I would personally run with air/water cooling. The Polaris GPU looses most of its efficiency past 1.2 volts, there is no real reason to go beyond 1.3 volts as your power consumption would jump though the roof and ultimately be limited by board power aka throttling. I will note that pushing 2.0 volts to polaris even under ln2 is rather pointless, as every card I've tried to push past 1.6 volts or so doesn't gain anything in terms of clock speed just more heat.

you could go higher than 1.3-1.35 volts under water, and really good air cooling, but at that point its nearly impossible to prevent a relatively stock rx-480 from throttling due to powerlimit, which requires hard modding and such to prevent. I wouldn't worry about failure unless you plan on keeping the card for 5+ years.
 
ebduncan said:
the absolute ceiling is 1.5 volts? lol nah 1.2-1.25 volts being max suggested? nah

To correct your statement, 2.0 volts is the max under ln2, and 1.3-1.35 volts is the max I would personally run with air/water cooling. The Polaris GPU looses most of its efficiency past 1.2 volts, there is no real reason to go beyond 1.3 volts as your power consumption would jump though the roof and ultimately be limited by board power aka throttling. I will note that pushing 2.0 volts to polaris even under ln2 is rather pointless, as every card I've tried to push past 1.6 volts or so doesn't gain anything in terms of clock speed just more heat.

you could go higher than 1.3-1.35 volts under water, and really good air cooling, but at that point its nearly impossible to prevent a relatively stock rx-480 from throttling due to powerlimit, which requires hard modding and such to prevent. I wouldn't worry about failure unless you plan on keeping the card for 5+ years.
Click to expand...
Power consumption has nothing to do with extreme V in my context, the ceiling is a mixture of hitting the performance wall, thermal spots on the GPU, and silicon-node voltage sensitivity-threshold.

And the absolute ceiling (there will be some leeway as you know due to variance between dies) as I mentioned tested on LN2 so far is just a bit above 1.4V for Polaris 480 and just about 1.35V for Pascal (although the extreme OCers mention even with safeguards off and for benchmarking maybe better closer to 1.3V).
These are extreme international OCers who removed all safeguards and configure PWM controller directly while also using LN2.

Where are you getting 2V has been achieved on Maxwell 2 (again this is not fully relevant to Polaris/Pascal as the node shrink greatly affects voltage sensitivity/density/thermals/voltage characteristics/etc)?
I need to double check but last time average LN2 was between 1.5V and 1.6V for Maxwell 2, let alone 2V and context when talking about being stable enough to run complete benchmarks for competition or critically 3D loads (nor talking about gaming).

Edit:
Elmor reached 1.66V (2200MHz) some time ago with Maxwell 2 GPU, and these guys do not care if the GPU lasts only 1-3months or less as they are going for world records.
Cheers
 
Last edited:
ebduncan said:
To correct your statement, 2.0 volts is the max under ln2, and 1.3-1.35 volts is the max I would personally run with air/water cooling.
Click to expand...
Just checked one of the guides written by those international OCers, and they state the:
You can go as high as 1.7-1.85V with this kind of modification, so it’s suitable for extreme overclocking.
But beware, safe voltages even on LN2 are rather around 1.5-1.6ish range
Click to expand...
This is Maxwell 2 using LN2 with all safeguards removed and heavily modded, semantic wise we could have different POV on whether ceiling should be the safe limits when temporarily set-used for benchmarking with LN2, or all out break the GPU and go for world records above 1.6V, also this is on 28nm node though and not 14/16nm which so far for Pascal has the equivalent of the 1.5V to 1.6V being 1.3V to 1.35V for LN2.
But that is digressing from the thread.
Cheers
 
Last edited:
Just a reference for comparison with the reference card modded by dancop and der8auer with LN2 at 1600MHz and then 1700MHz.
LN2: 1600MHz, 1.41V, 173.2W (this is not the full tdp just the calculation for the GPU core)
LN2: 1700MHz, 1.45V, 205W (caveat as above)
Water: 1480MHz, 1.35V, 147.5W (caveat as above) - downclocked from 1500MHz as it had a better Fire Strike score with 1480MHz, also noted that above 1.35V thermal temps were getting too much internally for the GPU, watercooled temp was 52c to 60c.

Buildzoid with XFX watercooled, 1500MHz (by TRIXX as GPU-Z has no reading for some reason and same version as der8auer, curious why), 1.43V, 211W (came caveat as above).
Part of the difference compared to LN2 is the result of thermals and power leakage/energy waste.
Cheers
 
Last edited:
Came to thread excited by XFX overclocking. Left thread sad after crazy anti-OC person ranted about voltage and watts. Here I am drawing quadruple the power from my XFX GPUs and people complaining about 14nm smh. It's like no ones overclocked their PCIe bus before. Older motherboards had 4pin power right to the slot for epic overclocks.
 
why the fuck has this thread turned into PCI-e voltage draw again. WTF enough with that shit already. This is a custom card, push the limit, overclock the f out of it, who really cares.
 
CSI_PC said:
Power consumption has nothing to do with extreme V in my context, the ceiling is a mixture of hitting the performance wall, thermal spots on the GPU, and silicon-node voltage sensitivity-threshold.

And the absolute ceiling (there will be some leeway as you know due to variance between dies) as I mentioned tested on LN2 so far is just a bit above 1.4V for Polaris 480 and just about 1.35V for Pascal (although the extreme OCers mention even with safeguards off and for benchmarking maybe better closer to 1.3V).
These are extreme international OCers who removed all safeguards and configure PWM controller directly while also using LN2.

Where are you getting 2V has been achieved on Maxwell 2 (again this is not fully relevant to Polaris/Pascal as the node shrink greatly affects voltage sensitivity/density/thermals/voltage characteristics/etc)?
I need to double check but last time average LN2 was between 1.5V and 1.6V for Maxwell 2, let alone 2V and context when talking about being stable enough to run complete benchmarks for competition or critically 3D loads (nor talking about gaming).

Edit:
Elmor reached 1.66V (2200MHz) some time ago with Maxwell 2 GPU, and these guys do not care if the GPU lasts only 1-3months or less as they are going for world records.
Cheers
Click to expand...

You are running off on tangents. I did not mention maxwell or pascal. I mentioned Polaris. ONE GPU. The 2 volt number I referenced was from Hawaii aka the 290x/390x. I've pushed polaris to 2 volts as well, Like I mentioned before I didn't gain anything past 1.6 volts. On stock cards, going past 1.35 volts under air or water, will likely result in going past the board TDP limit causing the card to throttle clock speeds.
 
XFX RX 480 GTR 8GB XXX Edition DD would that be a upgrade from my GTX 970 strix ? I'm playing titanfall 2 at 2560x1080
 
i'm about to pull the trigger on this card
it's on sale on amazon
 
ebduncan said:
You are running off on tangents. I did not mention maxwell or pascal. I mentioned Polaris. ONE GPU. The 2 volt number I referenced was from Hawaii aka the 290x/390x. I've pushed polaris to 2 volts as well, Like I mentioned before I didn't gain anything past 1.6 volts. On stock cards, going past 1.35 volts under air or water, will likely result in going past the board TDP limit causing the card to throttle clock speeds.
Click to expand...
We both are.
I mentioned maxwell/pascal to give you a clear indicator on difference between the silicon nodes, also they have far more extensive international OCers to get a better indication of this trend, but it will also apply to AMD as the node shrink has implications as I mentioned, I may try to dig out one of the EE degree course notes on this subject but why digress even more as I can clearly see not everyone is interested.
You also seemed to ignored where I commented about the difference of safe and breaking level voltage on LN2, a guide written by one of the top international OCers who was an engineer at one of the IHVs,
You can go as high as 1.7-1.85V with this kind of modification, so it’s suitable for extreme overclocking.
But beware, safe voltages even on LN2 are rather around 1.5-1.6ish range
Click to expand...
While this is Maxwell it is still applicable albeit with different levels for Hawaii and Fiji - context here is 28nm.

Thought I misread that you said you managed 2V on Polaris with LN2 but looks like I did not.
I am really curious how you managed 2V on Polaris and with what stable frequency under benchmark condition, I have not heard of any international clockers able to do that yet on LN2, would expect it to be info around just like Pascal hitting 3000MHz (some done that but not posted much about it as it is very limited to R&D with restricted benchmark records).
Especially as I have seen the performance spec envelope (voltage-frequency-performance) of the silicon from TSMC and Samsung.

But that is great result considering the LN2 I have seen so far from those of various/previous record holders and their own results, one of which I posted earlier with 1700MHz at 1.45V for a heavily modded reference 480 on LN2.
Cheers
 
Last edited:
123Lanoix said:
XFX RX 480 GTR 8GB XXX Edition DD would that be a upgrade from my GTX 970 strix ? I'm playing titanfall 2 at 2560x1080
Click to expand...

nope they are about equal and side grades... your upgrade paths start at GTX 1070 and above..
 
i was looking at titanfall 2 benchs on youtube and the rx 480 is beating the 1060 easily and got much better frame rates than my gtx 970. found a better deal
SAPPHIRE NITRO+ Radeon RX 480 100406NT+8GOCL 8GB 256-Bit GDDR5 PCI
for $249 - 15$ rebate = $234
 
123Lanoix said:
XFX RX 480 GTR 8GB XXX Edition DD would that be a upgrade from my GTX 970 strix ? I'm playing titanfall 2 at 2560x1080
Click to expand...

Is it an upgrade from a GTX 970? Yes, without a doubt it is. In games that can utilize more than 3.5GB of VRAM it will help your frame rate tremendously! You will be one happy camper! It will blow the darn doors off a GTX 970. A lot of the newer AAA titles can utilize the amount of memory on a RX 480. As you said the TitanFall 2 benches don't lie.

If game is less demanding and the game would normally run fine on your GTX 970, then it's not a huge upgrade. It will still be a tier faster, but not blow your socks away faster. Still an upgrade though.

A nicer upgrade would be a GTX 1070. That would be a full 2 tiers faster than your current card, 1 tier faster than a RX 480, and it has the memory capacity to handle demanding games. Look for one on Jet.com if it fits your budget. Jet is a WalMart owned company.

If a GTX 1070 is too expensive, then a Fury X could come into play at around a $300 price point.

Hope that helps. In other words you have plenty of options. ;)
 
^^I'd rather get the nitro+ Lanoix described over a Fury, easy. 8gb vram, hw hevc encode and decode, hdmi 2.0, 1/3 less power draw. There's more to it imo.
 
iam trying to stay around $250 270. the 1070 scares me because of the VRM issue with EVGA cards also the card is really expensive. is sapphire nitro a ok card ? or should i go with the more expensive XFX ?
 
123Lanoix said:
i was looking at titanfall 2 benchs on youtube and the rx 480 is beating the 1060 easily and got much better frame rates than my gtx 970. found a better deal
SAPPHIRE NITRO+ Radeon RX 480 100406NT+8GOCL 8GB 256-Bit GDDR5 PCI
for $249 - 15$ rebate = $234
Click to expand...

That's a really nice card also. It seems that the XFX cards are really built nicely. I'd lean towards that if price isn't a consideration. But for $30 cheaper you can't go wrong with the Sapphire. ;)
 
123Lanoix said:
iam trying to stay around $250 270. the 1070 scares me because of the VRM issue with EVGA cards also the card is really expensive. is sapphire nitro a ok card ? or should i go with the more expensive XFX ?
Click to expand...

Nitro+ and XFX GTR OC are about the same quality. How far you can clock depends on the silicon lottery so buy either and pray for 85% asic!
 
yeah i think iam gonna go with the sapphire they are both clocked similary but it's alot cheaper $234 after$15 rebate vs $264 after $10 rebate. the sapphire is clocked higher at 1342mhz vs xfx 1338mhz
 
123Lanoix said:
iam trying to stay around $250 270. the 1070 scares me because of the VRM issue with EVGA cards also the card is really expensive. is sapphire nitro a ok card ? or should i go with the more expensive XFX ?
Click to expand...

EVGA VRAM issue is a EVGA only issue. There is another issue with Micron memory, but the individual manufacturers put out a bios update that you can flash your cards with. Basically the Micron memory will do strange things like flash checkerboard images until the bios is flashed. Some people aren't comfortable with flashing a bios. The GTX 1070 is a nicer card if it could fit your budget. :)

Congrats on the Sapphire. ;)
 
ok i pulled the trigger on the Sapphire Nitro
RX 480 8 GB@ newegg sweet black friday early sale
 
i just can't justify $400 or 375$ right now. i just need a little bump in performance. I can throw my GTX 970 in my 2nd gaming machine. I will be putting up a old MSI R9 280x on ebay.
 
i was really in the market for a 1060 6gb or rx 480 8gb but the titanfall 2 benchs and doom benchs swayed me. i guess battlefield 1 also uses DX 12
 
123Lanoix said:
i just can't justify $400 or 375$ right now. i just need a little bump in performance. I can throw my GTX 970 in my 2nd gaming machine. I will be putting up a old MSI R9 280x on ebay.
Click to expand...

Well you could get a used 980ti at 330 which will match a 1070 making it the ideal card at slightly over 300 bucks.
 
123Lanoix said:
ok i pulled the trigger on the Sapphire Nitro
RX 480 8 GB@ newegg sweet black friday early sale
Click to expand...

For that model just need to keep an eye on the temps and fan profile and frequency bouncing, you will need to do a little bit of tweaking to stop it throttling the clocks and at minimum increase the fans a bit.
Some notice difference by playing with the voltage and lowering it a little, if uncomfortable doing that stick with the higher fan profile and maybe a bit more power target (only if you see the clocks bouncing and not sustaining close to the card's boost clock).

Cheers
 
are the fans really loud? if i max out the power target to get the max boost speed advertise will it get too loud?
 
iam not really looking to overclock ? i dunno what the avg overclock is for this card?
 
123Lanoix said:
are the fans really loud? if i max out the power target to get the max boost speed advertise will it get too loud?
Click to expand...
The Nitro is a bit louder than some of the other models out there, my point about the power target was not to do with OC but to help sustain max boost clock, it will not need massive amount, the alternative is to lower the voltage a little but no guarantee how much that will help.
But it may not need any tweaks apart from fan, just keep an eye on the clocks and watch if they bounce around near the boost clock.
Cheers
 
what causes it to bounce around? like on my GTX 970 it basically i set the power target to the max and it pegges at 1475mhz all the time on full load
 
123Lanoix said:
what causes it to bounce around? like on my GTX 970 it basically i set the power target to the max and it pegges at 1475mhz all the time on full load
Click to expand...
Same thing that can make Nvidia FE cards do the same, temperature and power target.
The boost algorithm (controlling clocks) is looking at thermals-voltage-frequency to find optimum, the card may have an ideal voltage but also needs higher power target, very generally higher temps need slightly more voltage and power targets.
It is an ok card (especially if the price is right in comparison) but personally if you can and not too late worth trying to stretch to other 480 models, especially XFX or HIS.
Usually Sapphire is a pretty safe bet for AMD card, but this time round seems they are not up to their usual standard-spec.
Pretty controversial because there will be quite a few happy with their Sapphire Nitro, and others who are not *shrug*.

Cheers
 
argh.... i was considering XFX high end rx 480 for 275 but is it reallty worth the hassle for $30 more? what exactly is wrong with this sapphire nitro ? is it not able to keep the 1342mhz boost clock consistnetly? i don't care if it runs a t 70c or whatever
 
123Lanoix said:
argh.... i was considering XFX high end rx 480 for 275 but is it reallty worth the hassle for $30 more? what exactly is wrong with this sapphire nitro ? is it not able to keep the 1342mhz boost clock consistnetly? i don't care if it runs a t 70c or whatever
Click to expand...

Dude, why are you all wavering? I think yer responding to someone I already put on ignore and if thats the case I suggest you follow suit.
 
sorry, i guess i didn't want to get a bad card. I feel like the sapphire nitro was a really good deal for the price on newegg. I was like debating these two cards for awhile this morning
 
You must log in or register to reply here.
Back
Top