AMD: Speculation, Rumors, and Discussion (Archive)

Status
Not open for further replies.
This is the same guy that showed the firestrike score, I posted ;) its a rx480. He posted the videos to show there was no photochopping going on.
 
The temperature on that 3dmark Firestrike run the rx480 is hitting 90 C.
Supposedly he was using a cooler not meant for the card to make sure none of the possible photos & videos he's going to take won't give out where he got the card from
 

Bah, no-one will be ever able to beat this masterpiece:

ByXz6idCAAAYjK7.png
 
If it is a 480 and it gets so hot then it's not good news.

I guess the news being good or bad depends on how fast the fan was running and how crappy job the guy did when he replaced the cooler with the one not meant for RX480. As long as it doesn't throttle and sound like jet taking off it's all good.

But I have my doubts about this being real in the 1st place.

edit: also notice temps start at very high 70C. Something is definitely wrong.
 
I guess the news being good or bad depends on how fast the fan was running and how crappy job the guy did when he replaced the cooler with the one not meant for RX480. As long as it doesn't throttle and sound like jet taking off it's all good.

But I have my doubts about this being real in the 1st place.

edit: also notice temps start at very high 70C. Something is definitely wrong.
Dynamic thermal management for FinFET-based circuits exploiting the temperature effect inversion phenomenon

I wouldn't be so sure, maybe the tests are fake, but the temperatures could be far different than what we're used to. With the smaller processes, temperature can actually effect frequency scaling. 70C may not be running hot, just the optimal temperature for performance. That also implies a really low (<0.5V) voltage. At those voltages, exceptionally dense high CU designs might be practical.
 
Dynamic thermal management for FinFET-based circuits exploiting the temperature effect inversion phenomenon

I wouldn't be so sure, maybe the tests are fake, but the temperatures could be far different than what we're used to. With the smaller processes, temperature can actually effect frequency scaling. 70C may not be running hot, just the optimal temperature for performance. That also implies a really low (<0.5V) voltage. At those voltages, exceptionally dense high CU designs might be practical.

Well we have temp data for Pascal and I don't really see a whole lot of difference in behavior compared to 28nm. Pascal is running at <40C at idle. Do you really believe RX480 is running at 70C idle?
 
Well if he screwed up something putting on the cooler that would be ok, but if he didn't, 90 degrees C is quite hot, doesn't matter if the silicon was made to take that much temperature, with thermal barriers and all and the 290x had a fairly beefy cooler if we want to look at a card with 275 watts to a card using 130 watts.
 
If the guy screwed up putting the cooler on, the card would burn itself up (or trip thermal protection and shut down altogether); although low compared to top-end graphics cards, 170ish watts is far too much to handle for a screwed-uppedly mounted cooler! :)
 
I doubt there would already be 3rd party coolers available for this "new" card, and if so whether he would also have access to it ...
Seems like this could be one of the cards that AMD is using to determine what the final clocks should be and it does run hot!
 
I doubt there would already be 3rd party coolers available for this "new" card
Why not? There is nothing inherently difficult in sub 150W & if the board is using the same hole spacing/close enough chip/mem area they might be able to just chuck on an existing cooler with a different sticker.
 
I think there will be cause if they already have AIB boards, wouldn't they send out cooler specs to other companies at the same time?
 
If the guy screwed up putting the cooler on, the card would burn itself up (or trip thermal protection and shut down altogether); although low compared to top-end graphics cards, 170ish watts is far too much to handle for a screwed-uppedly mounted cooler! :)

It's not one extreme or the other, there are various grades of cooling performance that are possible. There are various possible errors and degrees of failure in cooling. Something can be badly placed without totally stopping the system from running.

If the card beats the 980 that's great, and definitely buying. If as some rumors say overclock might get near or pass 980ti, that's even better, it would also open the possibility for it to go near or surpass the 1070 in some dx12 apps.
 
I'll consider replacing my 290X with one if it outperforms it by 10% and is indeed around 130w, but only if I can get the 8GB version for a bit less than I paid for my 290X. If it is just a cheaper lower watt GTX 980 then i'll probably have to pass - although it would be right at home in my partners mITX system. So maybe I'll be tempted to upgrade temporarily before moving it onto that system when something properly compelling comes along for the right price.

My concern is that a lot of gamers bought like me, either a 970 (majority) or a 290/290x when they were reasonably priced. If it can't outperform them by at least 20% for a nicely cheaper price then only those with older or midrange gpu's will be able go with an AMD gpu for most of this year. Yes that's a big chunk of the market, but it leaves a lot of us in the upper-mid/lower-high end range high and dry, especially with nvidia's pricing increases. There could be quite the product void.
 
I'll consider replacing my 290X with one if it outperforms it by 10% and is indeed around 130w, but only if I can get the 8GB version for a bit less than I paid for my 290X. If it is just a cheaper lower watt GTX 980 then i'll probably have to pass - although it would be right at home in my partners mITX system. So maybe I'll be tempted to upgrade temporarily before moving it onto that system when something properly compelling comes along for the right price.

My concern is that a lot of gamers bought like me, either a 970 (majority) or a 290/290x when they were reasonably priced. If it can't outperform them by at least 20% for a nicely cheaper price then only those with older or midrange gpu's will be able go with an AMD gpu for most of this year. Yes that's a big chunk of the market, but it leaves a lot of us in the upper-mid/lower-high end range high and dry, especially with nvidia's pricing increases. There could be quite the product void.

If it works in crossfire , I would get the 8 gig and use it as my main and use my 290 as the second cross fire card. Some leaks I've seen put it above the 390x in certain games
 
Status
Not open for further replies.
Back
Top