lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
You are free to check context of the whole discussion before talking about relevance of furmark.
Not to mention furmark will always be relevant as a way to keep room warm at winter.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
AMD Radeon RX 480 Supplies at Launch
- Thread starter FrgMstr
- Start date
- Status
Not to mention furmark will always be relevant as a way to keep room warm at winter.
looks like the Oc tool from AMD can only offer minus voltage regulation... I wonder if the reason is AMD new GPU doesn't like that much voltage increase beyond certain point... 
The rumor from Chiphell 1340Mhz clock oc as maximum seems to be the normal oc anyone is getting on their sample..
The rumor from Chiphell 1340Mhz clock oc as maximum seems to be the normal oc anyone is getting on their sample..
cards are designed to stay under a certain tdp now right? Reason companies has gotten smarter. I bet you amd is limited the TBP on these cards. So even if you shot up furmark it will probably throttle to stay under power limit. That's just they way they do it now. They have gotten smarter since people started killing their cards lol.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Wait, is there a source for "minus voltage" regulation? Either way, with the way the voltage curve looked on screenshot i am certain overvolting would be useless anyways because voltage scaling is present, but it's rather sad for power consumption increase associated.looks like the Oc tool from AMD can only offer minus voltage regulation... I wonder if the reason is AMD new GPU doesn't like that much voltage increase beyond certain point...
The rumor from Chiphell 1340Mhz clock oc as maximum seems to be the normal oc anyone is getting on their sample..
Im wondering... Are your Granada and Tonga examples, reference designs or AIB overclocks? That matters a lot according to you.
And that FijiXT example, does that happen to be on one of the many charts that also has the 980ti listed above 350w and he 980 above 300w? I saw a FEW charts on FuryX reviews where they all seem to be using total system power without saying as much.
too late, lol!!!!he isn't talking about that, lol, you are going on a tangent
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Well, to my knowledge Wizzard uses reference cards in his comparisons, you are free to confirm or deny that yourself. Because power consumption figure for 270x definitely aligns with reference version.
As for your Fiji question, i am not sure i understand it, TPU uses power measured at slot and connectors for their values.
They probably are just using reference I don't see why they would go so far to use AIB overclocked versions without mentioning them, cause when reviews can't match those numbers up , the reviews would call them out, and that is never a good thing.
Ah, I wasnt using the TPU charts exclusively...
I think I see why you are thinking the way you are, but I disagree.
We will see come the release of the 470/480 if their TDPs are a typical gaming draw, or a max load draw.... Considering the rumors of the 480 drawing as low as 110w of power on average, I think the listed TDPs are their max loads when not OCed.
While I can see why you are skeptical of the TDPs, I think its hardly fair to claim your assumptions as some sort of valid information without disclaimer. That said, it seems most people here pass along rumors and speculation as gospil, so bleh
I think I see why you are thinking the way you are, but I disagree.
We will see come the release of the 470/480 if their TDPs are a typical gaming draw, or a max load draw.... Considering the rumors of the 480 drawing as low as 110w of power on average, I think the listed TDPs are their max loads when not OCed.
While I can see why you are skeptical of the TDPs, I think its hardly fair to claim your assumptions as some sort of valid information without disclaimer. That said, it seems most people here pass along rumors and speculation as gospil, so bleh
Well that's the thing AMD has done stuff like this in the past with performance, frame rates, and they got a back lash from it, this time they haven't been doing that at least not what we have seen, so always give them a benefit of the doubt till the reviews come out is a good thing
TaintedSquirrel
[H]F Junkie
- Joined
- Aug 5, 2013
- Messages
- 12,707
WCCF is claiming 25x more RX 480's than 1080's are going out (citing 1 store).
Another Microcenter employee on Reddit says his store has currently only received four.
Another Microcenter employee on Reddit says his store has currently only received four.
Stereodude
2[H]4U
- Joined
- Oct 20, 2000
- Messages
- 3,285
He must be lying. He bought the other 96 to scalp on eBay.
D
Deleted member 93354
Guest
I think WCCF is full of horse manure.
But it's too early to tell. They might be slated for 1000, but only have 10 so far. But they could also get 990 unit shipments in on the day before release to general public.
But it's too early to tell. They might be slated for 1000, but only have 10 so far. But they could also get 990 unit shipments in on the day before release to general public.
DooKey
[H]F Junkie
- Joined
- Apr 25, 2001
- Messages
- 13,577
Makes sense there might be more. Smaller chip, in production sooner (supposedly), mainstream vRAM. Not sure why NV is in this conversation. Two totally different levels of consumer product.
TaintedSquirrel
[H]F Junkie
- Joined
- Aug 5, 2013
- Messages
- 12,707
True but 25x is substantial.
This looks more and more like a smarter decision for AMD. They must have known bigger dies on brand new process would suck to make chips in large quantities. It only shows Nvidia's current situation. Now people are like they are selling out, ofcourse they are given how many are available to go around. But if Nvidia was making high end in large quantities there are only so many people lined up to spend 600+ on a card. It's widely reported there is a shortage. OC uk has supplied 1000 all together to their customers for 1080s and they are reporting 1000 rx 480 in stock on day one and more on standby.
So supply seems to be plenty and doesn't look like these will be in short supply. that much is given. I don't know about 25x overall just going by one fuckin store but I am pretty confident this card will be available in decent quantities.
The last sentence, is that only received four 480's or or part of the article about four 1080s?
Thanks
funkydmunky
2[H]4U
- Joined
- Aug 28, 2008
- Messages
- 3,889
What?!? We were discussing what was high-end and mid-range. I don't give a crap when 1060 (ti?) ship. But when they do they will be mid-range.
Just like LAST GEN!
Fun
Limp Gawd
- Joined
- Dec 21, 2013
- Messages
- 344
Road trip!!25 stores worth all went to the Overland Park Kansas store
For actual purchase decision, at $200 RX480 will unlikely have any real competition, and will be significant performanc step-up at that pricepoint. Maybe I would wait for custom open-air boards, but YMMW.
But for the academical discussion here on [H], this is rather referring to Kyle's article stating, that AMD originally intended 480 to be more oomph, in later discussion clarified especially in clocks department. It was also noted by many that RX480 has effectively similar die size as 1070. And some words by Raja Koduri back in December such as "premium" might indicate aiming at least for FuryX performance.
It's all speculation until anything is backed up by proof to be honest. It just turns in to he said she said. It could all be true and it could be some what true or not true at all. Remember though both companies are in infancy of the new process.
First gpu on 14nm finfet and first on 16nm. Looks like what Nvidia is dealing with not being able to make enough 1080s but we are seeing more 1070s roll in. That just shows a few things, lot of cut down dies and probably not enough being capable of sustaining high speeds with higher shader count so cut it down and roll it on to 1070. I have seen plenty more 1070s in stock recently but 1080s are still struggle. It should all get better in time and just like AMD is probably binning chips that do higher clocks.
Polaris is getting comparably equal clock gains to nVidias news chips. Both gained around 30-35% max-clocks over last generation. Claims that it isn't getting expected clocks are flat out wrong. I want someone to explain why AMD should have gotten a higher % of gains in the clock department (when compared to NVidia), if they fully believe AMD should have higher clock...
On an 'academical discussion' level, I think the claim that Polaris was supposed to be higher to also be flat at wrong and/or a disingenuous claim. This claim is based almost entire on the assumption and/or opinion that the high end parts should come out first, and that it makes no sense for them to release mid-range without high-end... There are PLENTY of reasons for this, including waiting for 14nm yields to improve as well as higher HBM2 and gddr5x availability... Beyond that, releasing a high-end Polaris with a replacement Vega part coming 6 months later makes absolutely no sense. Vega is the performance part of this generation. Creating a high end Polaris part and then replacing it with a Vega part in 6 months, makes no sense.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Once again, you are making a statement you can't back up even with rumor mill. There is no evidence of Polaris 10 max 24/7 clocks.
There is evidence of "certain to work at" clocks (that being boost clock) being upped by about 40% for Pascal and what we know of P10 so far indicates same rise being of 20%. Go figure why some may think Polaris 10 should have landed higher.
Then we should wait for reviews right? Do you have facts claiming that AIB boards wont be able to achieve around 1500?
We both know why AMD clocked it where it clocked it. Power/Performance. May be the binned chips for higher clocks weren't enough to appeal to mass market. So they picked the best clocks, and sold it for cheap and let AIBs play around with higher clocks while claiming reference stays within certain limits.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Sure, we should wait for reviews, that i can always agree with. I am not too certain about AIB boards seeing how it did not exactly do anything notable for overclocks of r9 3xx series or Maxwell, and ofc Pascal.
Well all the previous amd boards even with stock power envelope were ok in OC. They were already pushed pretty high from factory but this card seems like they might be limiting the reference card in TDP and we would probably have to have after market cooling and 8pin to see anything above 1400.
There is someone who posted a screenshot of fire strike running at 1380. I think looking at the stock cooler you really need to play around to get close to 1400. They are probably targeting certain temp and certain power envelope.
Really? There are pascal cards that are "certain to work at" around 2.5GHz? (1.77GHz x 1.4 = 2.47GHz) Where are these mythical cards and how much are they going for?
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
I talk about Maxwell to Pascal jump. Take 980 boost clock and compare it to 1080's. And yes, i don't care if 980 never worked at boost clock at stock, it's the frequency they are certain to work at, and 1080 does not work at it's boost clock either.
Thing is amd card hold their clocks most of the time. All amd cards actually have boost clocks but the thing is they never really got focused on. Because amd called is max attainable clock or something. They sustain it most of time unless it goes too high in temps. I think the 290 did this but with launch driver update they bumped up the fan speed and it sustained it's max clock almost 100% of the time. Nvidia cards do tend to fluctuate quiet a bit especiallt pascal when overclocked. It seems to be all over the place kinda targets it's boost clocks it seems when it's heavily overclocked. I have seen a few graphs don't know exactly what website it was. It was over a 15-30 min gaming period.
Bear in mind much of the boost fluctuation with 1080 comes down to the Boost3 being too conservative with profiles; changing temp target (this is within warranty) and adjusting fan profile to 60% at 80degrees helps it to maintain that boost.
It was stupid of Nvidia to not QA the ideal parameters before shipping the Founders Edition from a persistent boost perspective, more applicable to the 1080 than 1070.
Cheers
I'm pretty sure that with Nvidia's advertised speeds boost clock is usually what you get, as a minimum. However, in extreme scenarios the GPU will downclock all the way to base clock, but never below it.
Yes but it was in context to NKD's post, that said there are a couple of examples (a couple of games and a synthetic test) that push it below even the guaranteed boost without the minor changes I mention.
Cheers
harmattan
Supreme [H]ardness
- Joined
- Feb 11, 2008
- Messages
- 5,129
Gibbo over at OCUK just confirmed there will only be reference versions available initially. Seems, as we've heard, non-ref versions not arriving until mid/late July.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
You're reaffirming his point by mentioning price.
I dare to say that perf/transistor, P10 is slower than GP104. Perf/watt should be comparable tho.
Once again, pricing is set AFTER it's known where the card lands performance-wise. Look, Kyle is not a Polaris creator (to start with, he is not Chinese
), but with what we know about P10 and GP104, it's clear that AMD could potentially expect it to land higher than it did.
Pieter3dnow
Supreme [H]ardness
- Joined
- Jul 29, 2009
- Messages
- 6,784
When did AMD start making videocards exclusively for mid range, always a flagship then mid range. Suspiciously absent right ?
It is not all bad AMD just has to adopt a new strategy and that is what they did.
AMD has to sell a boatload of these cards and if they succeed then they did well.
I don't think it's the way you say it is. The card has 32 rops, so when was the last time we saw a card that was top end has that many rop and some how they made it a midrange at the end.
I think this is due to one thing. AMD can't afford to make a high end card first, they probably didn't want to go to big ass die from the beginning they wanted to make something they could make a lot of.
I am sure amd wanted vega right now but the amount of RX 480s they have to supply to all channel is probably their priority. Plus HBM2 part was always high end they can't have have high end without hbm2 when we know that was their plan from the beginning. Sure may be they expected higher clocks from Polaris and it didn't happen straight away on the new process on lot of the chips who knows. That's just anyone's guess I am sure everyone wants higher clocks.
But vega is what we have to wait for straight and simple. What happened with Polaris doesn't really matter anymore. They have one job to do now, sell a shit load.
- Status