Got some time so, last nights endeavours with the 480 were a bust, not sure of the cause of the troubles but suspect driver issues. I appear to have suffered another bout of the dreaded 'stuck task(s)' phenomenon that occasionally pops up. I think I only managed to complete 3 tasks in the space of 8 - 9 hours. The odd thing is that both GPU-Z and AMD's utility reported the card as running full steam which obviously was far from the case. With not knowing the true problem I removed the card and refitted the 280X.
The 480 can now be found in this host where it is in the spare second slot and is joining a 280X in this machine. I am still running it at my preferred x3 concurrency, however it should be noted the second slot is PCI-e2 in x4 mode, as such any results will not be truly representative of the 480's ability.
I had planned on putting the card in another machine but that died overnight! Motherboard or more accurately bios chip failure... My second in recent months.
I must admit to having not seen any evidence of the scare stories. According to GPU-Z, (which I concede may only be reporting what it's told to)
reported power draw at x3 concurrency has never gone above 105 Watts whilst crunching... 90 Watts average.
My biggest issue thus far is that it is not as FAST as I hoped! And in the case of my example it is quick to semi 'lock up' running work, resulting in overly long runtimes.
Its still early days, things could improve (!) beta driver 16.6.2 is certainly buggy but also an imperfect step forward for 'Hawaii' GPU's... All we can do is wait and see at this stage and hopefully better drivers come quickly.
I must admit to having not seen any evidence of the scare stories. According to GPU-Z, (which I concede may only be reporting what it's told to)
reported power draw at x3 concurrency has never gone above 105 Watts whilst crunching... 90 Watts average.
My biggest issue thus far is that it is not as FAST as I hoped! And in the case of my example it is quick to semi 'lock up' running work, resulting in overly long runtimes.
Its still early days, things could improve (!) beta driver 16.6.2 is certainly buggy but also an imperfect step forward for 'Hawaii' GPU's... All we can do is wait and see at this stage and hopefully better drivers come quickly.
I'm in agreement here with Gav, i had expected 7970 performance.
If you run lower spec PSU / mobo i might wait until these drivers and avoid overclocking. At the wall the delta with card in and crunching and out is about 135W - and had they put an 8 pin plug in this would be a non-story.
It's a midrange GPU card price wise. I have to remind myself the 7990 were selling for 1200USD not that long ago.
According to GPU-Z, (which I concede may only be reporting what it's told to) reported power draw at x3 concurrency has never gone above 105 Watts whilst crunching... 90 Watts average.
AMD is only reporting the power draw of the chip itself to tools like GPU-Z, unlike nVidia where it's the total power draw. I've read that at about 110 W reported power consumption the total amount is about 150 - 160 W.
AMD is only reporting the power draw of the chip itself to tools like GPU-Z, unlike nVidia where it's the total power draw. I've read that at about 110 W reported power consumption the total amount is about 150 - 160 W.
MrS
I don't think it's possible to report total card power, since only some rails contain current sensing circuits (like digital PWMs), which allow monitoring of power consumed. NVIDIA might however add some constants to the power measured.
Polaris seems to be able to monitor the GPU core rail only - I can find only a single digital PWM/VR there. The second rail that was used for VDDCI/MVDDC doesn't seem to be working there.
An interesting video and a compelling argument. I tend to agree with what he says about PCI-e power limit. Without searching out and reading the full specification for such a slot, I'm inclined to say that 75 watts is a "must meet" figure in the specs but not a hard and fast power ceiling.
The same youtuber in your link also posted about using Wattman to tune the 480 in various ways, a sort of Wattman tutorial if you like and I have to admit that in my rush to get my new card thumping along it was probably me not understanding the new interface causing all my problems and the subsequent many hours per task runtimes :-)
I have now made some adjustments to fan speed, temperature settings and have maxed the power limit! The card has started motoring again. Fingers crossed I will be able to get some representative results yet... Or burn the house down!!!!
Got some time so, last nights
)
Got some time so, last nights endeavours with the 480 were a bust, not sure of the cause of the troubles but suspect driver issues. I appear to have suffered another bout of the dreaded 'stuck task(s)' phenomenon that occasionally pops up. I think I only managed to complete 3 tasks in the space of 8 - 9 hours. The odd thing is that both GPU-Z and AMD's utility reported the card as running full steam which obviously was far from the case. With not knowing the true problem I removed the card and refitted the 280X.
The 480 can now be found in this host where it is in the spare second slot and is joining a 280X in this machine. I am still running it at my preferred x3 concurrency, however it should be noted the second slot is PCI-e2 in x4 mode, as such any results will not be truly representative of the 480's ability.
I had planned on putting the card in another machine but that died overnight! Motherboard or more accurately bios chip failure... My second in recent months.
http://www.myce.com/news/4gb-
)
http://www.myce.com/news/4gb-amd-radeon-rx480-can-flashed-expensive-8gb-model-79818/
unbelievable news, congrats if its possible.
I've returned my card. The
)
I've returned my card. The excess power draw issues being discussed online have made me think it may be best to "wait and see" a little longer.
I must admit to having not
)
I must admit to having not seen any evidence of the scare stories. According to GPU-Z, (which I concede may only be reporting what it's told to)
reported power draw at x3 concurrency has never gone above 105 Watts whilst crunching... 90 Watts average.
My biggest issue thus far is that it is not as FAST as I hoped! And in the case of my example it is quick to semi 'lock up' running work, resulting in overly long runtimes.
Its still early days, things could improve (!) beta driver 16.6.2 is certainly buggy but also an imperfect step forward for 'Hawaii' GPU's... All we can do is wait and see at this stage and hopefully better drivers come quickly.
Gav.
RE: I must admit to having
)
AMD's Facebook announcement about new drivers
48 hours should see an update which may help.
I'm in agreement here with Gav, i had expected 7970 performance.
If you run lower spec PSU / mobo i might wait until these drivers and avoid overclocking. At the wall the delta with card in and crunching and out is about 135W - and had they put an 8 pin plug in this would be a non-story.
It's a midrange GPU card price wise. I have to remind myself the 7990 were selling for 1200USD not that long ago.
RE: According to GPU-Z,
)
AMD is only reporting the power draw of the chip itself to tools like GPU-Z, unlike nVidia where it's the total power draw. I've read that at about 110 W reported power consumption the total amount is about 150 - 160 W.
MrS
Scanning for our furry friends since Jan 2002
Exceeding the PCI-E power
)
Exceeding the PCI-E power limits appears to be common.
Power Problems or PCI Exaggeration?
RE: AMD is only reporting
)
I don't think it's possible to report total card power, since only some rails contain current sensing circuits (like digital PWMs), which allow monitoring of power consumed. NVIDIA might however add some constants to the power measured.
Polaris seems to be able to monitor the GPU core rail only - I can find only a single digital PWM/VR there. The second rail that was used for VDDCI/MVDDC doesn't seem to be working there.
-----
RE: Exceeding the PCI-E
)
An interesting video and a compelling argument. I tend to agree with what he says about PCI-e power limit. Without searching out and reading the full specification for such a slot, I'm inclined to say that 75 watts is a "must meet" figure in the specs but not a hard and fast power ceiling.
The same youtuber in your link also posted about using Wattman to tune the 480 in various ways, a sort of Wattman tutorial if you like and I have to admit that in my rush to get my new card thumping along it was probably me not understanding the new interface causing all my problems and the subsequent many hours per task runtimes :-)
I have now made some adjustments to fan speed, temperature settings and have maxed the power limit! The card has started motoring again. Fingers crossed I will be able to get some representative results yet... Or burn the house down!!!!
Gav.
This is not within my area of
)
This is not within my area of expertise, but I found this article and subsequent discussion on the RX 480 power draw enlightening:
http://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480
The poster Allyn Malventano in the comments area claims to be a former Navy electronics tech.