Discussion in 'Frontpage news' started by Hilbert Hagedoorn, May 11, 2016.
Oh... The irony.
Looks like Vega is the one to get, not that long to wait, and quite substantial performance gain over Polaris and yet a long time until Navi comes out and also not that much of a peformance jump with Navi according to the roadmap chart - Vega makes the most sense if you're gonna get an AMD card I reckon.
No it isn't, it's about the possibility that we will more forum posts paraded around as credible. Let's hope that doesn't continue.
Yeah, because of people like you. You literally posted a news thread where the source is a random forum post.
And it's kind of my whole point. We need to draw a line somewhere with the rumors on Guru3D and I think this one personally crosses the line where its just ridiculous. Even if it ends up being true. The source is a forum post where the guy literally says he won't prove it.
See this is why the naming sucks. You're talking about Polaris and Vega the same you're talking about Vega and Navi when they are not the same comparisons
Polaris and Vega are the same architecture, the same way Tonga ( 380x) and Fiji (Fury) are the same architecture .
Navi IS NOT.
Vega will not bring improvements over Polaris; it will just be a bigger die.
He's alternating between validating the rumor and calling it into question, the same way he alternates between validating the reversed naming for Vega, and calling it into question...
I did , yes.. Clearly labeled as rumor.. So every single one that has done it i making this place like wccftech.. HUH?
It is not my fault that people side grade the actually topic at hand, this is happening a lot more... So are you going to blame HH now because he is the creator of this topic..
Please show me when it was said it would not be ready until 2017, because i read nvidia and amd news every single day, and did not heard that, and a i don't believe in nothing here i just said it is not impossible and the cause is not necessarily the fright of amd or cards manufactures, this is a possibility btw.
Nvidia and amd play a chess game and i like to follow that, but i hate missinformation about rumors that people acctualy take as true, wccftech could have posted it and i would be ok, guru 3d is a serious site, or was.
Yes. I am going to blame him. And you, for posting it. Its one thing to post a credible rumor, it's another to post something with literally no basis, something clearly said by a troll.
Guys. come on. No more.
I will start handing out more infractions
1070, you're right, Hilberts post only sates "1080", but it also states "higher-end to enthusiast segment" which I would imagine, also encompasses the 1070. Perhaps I'm wrong here, but I don't recall the GTX 970 (which is being replaced by the 1070) was "ever" classified as "mainstream", to which the post clearly states, that's all AMD should be launching.
Ummm, no. That's cherry picking, and expecting everyone to totally ignore the 1st paragraph. Edit: The 1st paragraph is more based in fact, according to many accounts in this thread, stating that 2017 was "supposed" to be the real release timeframe. The October part is the real "speculation." Either way, the fact that AMD doesn;t have an answer for Nvidia's soon to be released, higher-end to enthusiast segment, is the real disappointment.
I get some of you don't like "some" of this News post, but I don't think Hilbert has a reputation for knowingly and willingly re-posting fud (with exception to April 1st). Some of you might want to give him the benefit of the doubt.
LOL, thanks anyway.
I'm using the roadmap graph where they show Performance Per Watt increases vs timescale of release on the x-axis - by definition this is a very good way of seeing the performance potential of a card regardless of whether they are considered the same architecture or not. So, I still think that makes my post valid - Vega seems to have a good perf/watt jump over Polaris, whereas Navi is a long way out & not much of a perf/watt jump over Vega (proportionally). (I do find the quality of forum posts & discussion on Guru3D seems to have diminished recently, a lot of name calling as well as irrational & hastily composed posts that seems to make any proper discussion difficult, hard to see the wood from the trees in this forum today!)
Ah. You're using the fancy graph with the starry background and no numbers at all on the axes.
Vega has HBM, Polaris doesn't.
Polaris may have to sacrifice efficiency for higher clocks.
They are the same uarch.
Polaris was revised to 2xPerf/W
please note double performance/w of CURRENT MAINSTREAM OFFERINGS
It's not a 'fancy graph' at all, it's very simple to understand even without specific numbers listed - it's showing relative performance between the differently named architectures (or whatever you prefer to call them) - it's just that simple. I don't really need to get bogged down in the detail of your terminology to understand that graph. If you're saying the graph is wrong, then that's different, but I don't think AMD would just spout misleading untrue nonsense onto that graph. I'll leave it at that on this particular discussion today, I was originally just commenting that Vega seems like the one to buy as it's not far away and shows a good proportional increase in Perf/Watt.
Yes, AMD absolutely would put a bunch of BS on a graph. Starting with the fact that 2.5x Perf/Watt for Polaris is WRONG it is 2x.
Polaris brings double performance/watt COMPARED TO 28NM MID-RANGE GPUS.
The fact that the graph has no values on the axes makes it impossible to compare Polaris to Vega unless you literally start counting pixels and call the distance between Polaris and '28nm' GPUs 2x
You don't need to count Pixels, you can see the relative gaps even without a ruler. OK, then I'll give you that the 2.5x perf/watt for Polaris is wrong if that's the case, I suppose that could cast some doubt on usefulness of that graph now, but if that will change the relative performance gaps between the different 'named architectures' I don't know. Either way I'll stick with my idea that Vega is the one to get for now given that Navi is a long way out & not showing much of a performance/Watt increase over Vega. (Broke my promise about ending my discussion on this, see if I can keep it now!).
Vega is the one to get because it replaces the high-end. Any performance/w it gains over Polaris will be due to HBM (and the associated IMC) and/or reduced clocks to hit the perf/watt goals
Tonga and Fiji are generally considered as same, but Tonga has UVD 5.0 and Fiji/Carrizo have UVD 6.0.
While I generally do not disagree with naming certain chips as part of 1st, 2nd, ... generation of GCN, AMD always releases GPU with everything what was ready and working at given time.
Many changes are often not publicly known. Like HD 7790 having TrueAudio HW, but it is disabled, and next refresh using exactly same untweaked chip has it enabled.
So, while AMD may be ready to release Vega shortly after Polaris, there may and likely are some differences.
In the end Polaris was sampled last year and Vega (rumor) last month.
The block I'm interested in is the gcn block, uvd won't affect perf/w in games
Other than hbm what do you expect will affect perf/w between Polaris and Vega?
Clocks should be the only thing