• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD VEGA confirmed for 2017 H1

Status
Not open for further replies.
I want them to release a water cooled edition, minus the dodgy whining noises the Fury X had.

Probably will not be picking one up on release anyway. Played everything I want to play now on my 1070 and nothing much on the horizon this year that I am interested in. There is Prey, but it can wait. Not even sure when Squadron 42 is coming out :o

The Nintendo Switch is not far away now anyway. That together with FIFA 17, Street Fighter V, Dead or Alive 5, Mortal Kombat X and soon Tekken 7 on the PC will keep me busy :)
 
As said before though, amd failed to provide an itx cooler for the rx470/480 and I knew they wouldn't make one even though the ref pcb is small enough.
In my elite 110 I had to make my own itx rx470, but then it was replaced by a 1060 anyway. I cannot understand why the itx cooler from both the Nano and the Sapphire tonga itx were not modified for Polaris 10, I said this leading up to and just after launch of Polaris.
 
JediFragger said:
And AMD are also well known to make the power regulation etc top-notch on reference cards :)
Click to expand...

The high and low mosfets on the ref card are pretty good, certainly amperage is overkill for a 2304 shader card. But the rx470 only has a 4 phase and it's not in my experience strong if you bypass the power limit and run over 1.25v.

;)
 
Davedree said:
The high and low mosfets on the ref card are pretty good, certainly amperage is overkill for a 2304 shader card. But the rx470 only has a 4 phase and it's not in my experience strong if you bypass the power limit and run over 1.25v.

;)
Click to expand...

I must say even the supposed power limited Nano, overclocks to 1150 and works happily at 1.3v without downclocking.

Something that cannot be said for the Pascal cards, especially the 1080, which is power limited AND throttles from 22C.
 
Panos said:
I must say even the supposed power limited Nano, overclocks to 1150 and works happily at 1.3v without downclocking.

Something that cannot be said for the Pascal cards, especially the 1080, which is power limited AND throttles from 22C.
Click to expand...

Dunno why you keep comparing those in that way - you aren't really comparing like for like there - the Nano is built on a mature 28nm process that has plenty of headroom before you encounter any significant effects of electromigration while Pascal is using a brand new FinFet process and already pushing the boundaries of the possible clock speeds and also has little margin for additional voltage before you start to have problems with electromigration due to how thin the oxide is on FF - they haven't put the hardware voltage limit lock in the core itself for lols.
 
Panos said:
I must say even the supposed power limited Nano, overclocks to 1150 and works happily at 1.3v without downclocking.

Something that cannot be said for the Pascal cards, especially the 1080, which is power limited AND throttles from 22C.
Click to expand...

With the ref Polaris pcbs you really have to balance your power limit or you won't see performance scaling even if you don't see clockspeed throttling.
For example my 470 will not scale in gpu score on firestrike if the 50% power limit was reached. Only by fooling the i2c power sensor can I see performance increase, doing so is risky though as there is no no overcurrent protection as I found out by popping a mosfet.

Comparing 28nm vs 14nm is a different scenario for voltage scaling, but If on a stock nano with stock cooler, you were too fool the i2c power sensor and used that 1.3v would the nano voltage regulation take the current load/ and would you see more performance?

For my gp106 i don't see the first -13mhz until 63c.
But haven't tried gp104 so can't really comment on your observation.
 
JediFragger said:
And AMD are also well known to make the power regulation etc top-notch on reference cards :)
Click to expand...

ofc they will make reference PCB for AIBs, the thing is, they just have to stop themselves providing the cooler for the release batch.
as for ppl worried about blower cooler, nothing stops AIBs from making their own blower cards.
 
TNA said:
I want them to release a water cooled edition, minus the dodgy whining noises the Fury X had.

Probably will not be picking one up on release anyway. Played everything I want to play now on my 1070 and nothing much on the horizon this year that I am interested in. There is Prey, but it can wait. Not even sure when Squadron 42 is coming out :o

The Nintendo Switch is not far away now anyway. That together with FIFA 17, Street Fighter V, Dead or Alive 5, Mortal Kombat X and soon Tekken 7 on the PC will keep me busy :)
Click to expand...

I'll be looking at one around release or not too long after. An AIO watercooled solution card would be quite nice but will have to see.

Would a watercooler last as long as a fan cooled card though? I'd be concerned about the pump giving out. What happened to all of the Fury X cards as they weren't on sale for long at all.
 
Davedree said:
For my gp106 i don't see the first -13mhz until 63c.
But haven't tried gp104 so can't really comment on your observation.
Click to expand...

I've seen a few 1070 and 1080 cards go from around 8-10C startup through to 80+C under normal non-overclocked GPU boost behaviour and those I've seen do not run a higher bin at <22C - though they don't all behave the same - some do drop a bin around 50C or so and another around 60C while others will go all the way to around 79C before dropping a speed bin(s) based on temperature. There are a few tests done by certain sites that appear to show a drop after ~21C but if you read the article that is because of how the data is presented from the logger at start up - its quite possible there are some cards that do run a higher bin out the box at those kind of temperatures but it doesn't seem to be the story for every Pascal card.

My 1070 for instance even on a cold morning where it boots up at say 12-13C runs 1911MHz boost and stays at 1911MHz boost until the temperature hits 79C at which point it drops down 1-2 speed bins depending on what the GPU is doing load wise.

Panos had a bad experience with a Pascal card and seems to want to rubbish them all based on it.

There is another aspect to it though in that in an overclocked situation you can get higher speeds and higher sustained speeds if you can keep the GPU below certain temperatures.
 
Last edited:
Rroff said:
Panos had a bad experience with a Pascal card and seems to want to rubbish them all based on it.
Click to expand...

I didn't had bad experience with my GTX1080. It was watercooled and overclocked to 2190 :)
However was annoyed to have it throttling at 22C and again at 32C :( When never exceed 35C while benching overclocked.

If it wasn't throttling, I could have passed the 2200 barrier.
And was my fault didn't mod it, by using the liquid metal mod on the resistor. It could have done the job.

Something that cannot be done with the Nano :( since the resistors to the vcore need to be replaced. (@Davedree)
 
You probably wouldn't have held 2200 100% of the time even then - I've not seen a 1080 yet that doesn't drop back to around 202xMHz at some point under heavy load, or slightly above that if you do some tricks to minimise power usage by the memory. So far has held true even on the cards with modified power limits though not sure if that is the case with every modded vbios - anyone claiming otherwise possibly hasn't tested it as extensively under a variety of workloads as they think.
 
old gamer said:
I think they should sell cards with no cooler on just for the watercoolers, should even be cheaper
Click to expand...

Yeah I agree on that. And give them the ability to push more power in also :D

shredgodxp said:
Probably would be yeah but just imagine the complaints from folks who buy it assuming it's ready to use.
Click to expand...

I could imagine the headlines here
"AMD is doomed, they cannot afford coolers"
or
**Do not abbreviate or disguise swearing**
 
Status
Not open for further replies.
Back
Top Bottom