Yep which is nothing major lol
-
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.
Poll: ** The AMD VEGA Thread **
- Thread starter TNA
- Start date
- Status
Yep which is nothing major lol
Soldato
- Joined
- 25 Nov 2011
- Posts
- 20,679
- Location
- The KOP
FE edition not doing to bad in Witcher 3, Get the core at 1600 locked better cooling, better drivers etc it should get better with RX Vega
That's actually nice. I've been holding off playing the expansions until I got a new card, and at 1440Mhz it wasn't far off the 1080. Had some dips here and there, but as you said; get it to a stable 1600mhz and it'll do well.
Temps were in the high 60's as well.
How much power is FE drawing? I am surprised at the low temps, wondering what is dropping the clock.
In my mind, the best case scenario for the RX is probably slightly beating the GTX 1080 whilst using more power, for the same or slightly lower cost.
Just goes to show how much better the 1080Ti is over the 1080. Pretty impressive card (the Ti). Hopefully AMD have something up their sleeve for RX Vega in the drivers etc and a competitive price (but we always get hopeful about prices for GPU's and up being dissappointed)
Exactly. We had hoped for more, but if it can get itself solidly planted between the two 1080 cards, with a competitive price, that would be a good result. Obviously we know that the manufacturers are price gouging at present however.
Last edited:
If the RX Vega card at a stable 1600mhz, with proper drivers can sit between 1080 and Ti in DX11 it'll be very good if the price is there.
Hopefully AMD can manage that; although it's surprising how cool the FE ran in that test. High 60's at 1440Mhz isn't too bad.
Yeah that's a pretty positive sign, especially since Witcher 3 is not an AMD-favouring game.
If they can get it stable at 1630 MHz (the new rumoured RX base-clock), and the RX drivers are only 10% better. It should then be a touch higher than a max-overclocked 1080 by the looks of that benchmark.
If it does come in at max-overclocked 1080 performance for £450, I'll be happy with that.
Notice that occasionally the FE was dropping weirdly to exactly 37 fps (while before and after it holding around 46fps in the same scene) which neither of the other two cards exhibited - around the same time the clocks were behaving a bit erratically as well.
Looks like they still need to do a bit of work on the drivers.
A "bit" is an understatement haha. Geometry and tiled rasterization still don't seem to be working either; and looks like the card can't keep a boost over 1440Mhz; despite well before 84 degrees core temp.
I've not really paid much attention but it looks like the cards are throttling due to power limit or something as the behaviour doesn't match temperatures. Interestingly at a guess it looks like there is some roughly correlation with intensive memory operations though that might be completely wrong.
Yeah; still lots of unknowns. AMD never sent out a detailed architecture breakdown or anything since they never seeded reviewers.
It's clear there's untapped performance though; in features, drivers, and clocks speeds. Siggraph can't come soon enough.
something for people to pick their brains over
150 Watts for memory? wow....
something for people to pick their brains over
That's interesting. Is this what AMD planned for RX Vega all along?
Saying that, it will still be price that will be a huge factor for most people who are thinking about buying one.
I see they mention 6GB of GDDR5 runs at around 76W, so 150Watts for 16GB HBM2 could be right; specially if they're having production issues and they're less than stellar samples to reach their 1.89Gbs speeds.
The FE has a 12 phase for the gpu and only 1 phase for the HBM not sure what the max Watts is for that 1 phase ..but 150 watt for the HBM on 1 phase seems strange to me ......Could be the spike drops where seeing is not the core but rather the HBM
my bad should have posted this
Last edited:
it could be that more than 1 phase goes to the HBM and that the guy doing the teardown didn't get it right.
- Status