I would say it is not that clear cut because Intel's normal release pattern became FUBAR with Broadwell where many analysts wanted them to just write it off and skip straight to Skylake even if meant say 5-6months after the launch date of Broadwell, however Intel tried to make a product cycle...
I think we are possbly at cross communication/perspectives.
The situation with GDDR5X mirrors that of HBM2 IMO with regards to what was expected, what was actually available, what is definitely coming that is closer to the standards spec ceiling.
Cheers
Well you could say they are overclocking then because the V100 has HBM at 1.8Gbps.
So why now and not with P100, simplist answer is that Samsung could only do 1.6Gbps at launch, these days that rating-tolerance-yield is probably the cheapest HBM2 memory and why at that in Vega56.
But maybe this...
Same way GDDR5X could only do 10Gb/s at launch, now up to 12Gb/s, indicators 14Gb/s at end of year.
Fair to say there was no die above 10Gb/s at launch for GDDR5X in same way your context of HBM2.
Cheers
Because the HBM2 spec is to 2Gbps, and the manufacturers were initially saying they could do this before reality set in.
Same way GDDR5X is rated to 16Gb/s and memory manufacturers were confident this was going to be achievable early, but probably only see 14Gb/s by end of year mass production...
If that was the case Nvidia could had run their HBM2 comfortably at 2Gbps rather than 1.6Gbps at launch or 1.75Gbps like now.
Not all the manufacturered dies have the same tolerances and higher spec.
In theory then people should be able to OC their GDDR5x to 1.4Gbps no problems and HBM2 memory...
Not sure about that tbh because I read reports to get the higher speed the spec needs higher voltage, does not mean it is the same 1.6Gbps spec rated die as the 1.75Gbps or higher ones.
Same way GDDR4/GDDR5 and GDDR5X needs higher voltage as the speed increases, but does not mean they are all...
I would just wait on the monitor until it becomes clearer what is happening with HDR in the gaming world for PCs, meaning Q4 should give a better indicator.
Unless your comfortable to buy VRR monitor now and possibly replace in 8-12 months if HDR does take off.
Just saying because HDR has been...
Part of that comes back to my point in the FS article that you need trained particpants to have a more critical and correlated/aligned response, especially for subtle perception and preference differences.
Same way you will not get a clear consensus the level and which AA is needed to enjoy the...
Which GPU were you using with the 144hz monitor?
I appreciate one can tweak their game settings but just curious if you had to lower them much, maybe comes back to game played *shrug*.
Cheers
I just went back and checked.
Seems the codes must be used with the same model when Nvidia combines them in bundle but also hints of it being more restrictive, so still does not really stop miners although restricts more the game codes viability in that scenario...
Does AMD now enforce those games to be registered to the card/owner like Nvidia (not sure but I thought there was grumbling Nvidia was going to or has done this)?
If not then miners can still buy the game package and just sell the games on albeit not at full price to some of those key services...
I think it is more mining related because the GTX1080 custom cards are still a lot cheaper than the Vega 64.
Also could be argued the Vega56 would force the 1070 to be cheaper as it is competitive purely from an fps perspective.
Cheers
Could you link any of those other reviews with said games please.
I did a quick look and not sure where I can find any reviews using 8x MSAA beyond Dirt4.
Cheers
Well it also happens in Polaris like I mentioned and he shows in the vid.
So it is not just Vega or Fiji.
You might as well say it is a bug/GPU design limit with most of GCN then if people really feel something is untoward beyond Dirt4 and 8x MSAA in general
But seriously and mentioning more...
Not sure that much can be read into it, because to get that much more performance out of the engine relative to the Nvidia cards obviously the game has some hefty optimisation, and maybe that optimisation starts to drop after 4x MSAA due to the internal engine design.
I mean 17-20% more from the...
He also shows that a FuryX is actually faster than a GTX1070 with no AA in that game by around 6-7% and the 1060 vs 580 where the 580 is faster by 17-20%.
He also then goes to another site that has Dirt4 running 4x MSAA and at that setting the Fury X is in front of a 1070 and Vega56 is over 20%...
Ah my bad,
I looked at his profile and says 1080 while now looking back his OP says 980 doh.
But I know others had 100% fan speed with Pascal outside of mining malware albeit in early days of Pascal.
Cheers
So sound like it will be a very personal thing where it is fine for some but not others.
Good to know, probably means best suggestion for anyone going this route without experience like quite a few is to try both at 45-65Hz (fps) and also say 100-144Hz (fps) before making a decision on what they...
I think the variable refresh monitors and what is acceptable is going to be a very personal thing, maybe some happy with lower refresh rates and others not going by some owners on this forum *shrug*.
Makes it even harder to reach a concensus with that being the case.
Only way to know for sure is...
That is interesting because I remember reading one RX Vega review commenting about their thermal temps for the HBM2 memory.
Needs a lot more information but will be interesting to see if their heat behaviour comes back to the HBM stack height and being the packaging that is lower.
Cheers
Or he could give up the Freesync monitor and wait and buy a GSync from one of the better retailers when they have them on sale now and again (although needs to be careful selecting a Gsync monitor as many are not worth their price)....
But he is thinking that he will not be replacing Vega with...
Any idea how much network traffic is generated by the malware miners?
I assume looking at network traffic may also help to prove it is happening and actually that.
Just throwing out another one to consider, didn't some Pascal cards have 100% fan related bugs but I cannot remember if they came...
Some reviews were also hinting that the Vega 'nano' is also binned, which would be another plus point in its favour if true (they might just be speculating based upon performance envelope analysis).
Cheers
One thing to watch out for; Gameworks is now fully DX12 integrated and that will make it much better and viable; previously it would had been a nightmare to integrate Gameworks well into engines designed for both DX11 and DX12, also as it is a 'bolt-on' suite it really needs excellent dev-code...
I am not sure about that because in AAA games AMD has closed the gap with some of the notable ones out there even in DX11, DX12 ones depend if it is a new engine and with Nvidia engineers engaged with the studios.
I feel the console strategy has paid off to some extent as they seem to be getting...
Where do I talk about reference blowers?
If your going down that route (though is model meaning blower vs custom card not GPU die that is my context) well then complain about sites and people talking about the need to undervolt Vega and Polaris to give them greater relevance...
I feel you...
That has nothing to do with my post or context and based upon 'information' that again is speculation.
What we do know is the relative performance/envelope of Volta and also the product strategy Nvidia used with Pascal that is a mirror of Volta in many ways.
Cheers
It is being taken out of context, having to mention this in another thread.
In those financial calls Nvidia NEVER talk about products or segments that are not yet announced/launched.
Case in point financial call Q1 2016 in May, they briefly talk about GP104 (it was announced the week before)...
Need to consider how is the HBCC going to help when one has system RAM-VRAM-cache as the primary data buffers on consumer PC rather than HPC?
If AMD has hooked it somehow heavily into the VRAM and is now a compromise unless the drivers can use it properly with said software well then someone...
Yes and nothing in there tells you anything about future strategy, all he says is Pascal is still strong for the foreseeable future (a neutral response to keep consumer interest still in Pascal) but that has no relevance about Volta and if they launch GV104 before end of year or even when Volta...
You have a 1080ti for the performance, and probably had to go that route because waiting for the upgrade cycle from AMD so you were not getting the value out of your Freesync.
But with that experience it is probably a coin toss in terms of value because you may find yourself back in the...
The good news is that messing around with voltages will make it quieter, but then one can do that with both manufacturers although it really is needed for AMD products.
That said the general consumer I doubt will mess around with undervolting and testing to a point to find stable setup with said...
This is directed to AMD rather than HardOCP:
Sorry just watched Rogue One for nteenth time tonight (yeah meme from Return of the Jedi I think) :)
Also the test had to rely upon one monitor with a natural refresh of 60Hz and overclocked while the other was native 100Hz.
TBH in that context it...
This is been seriously taken out of context by the various tech sites.
Same tech sites that were then caught out when 1080 and 1070 launched.....
People and those sites were making assumptions about Pascal Geforce because he did not mention it at CES last year with the focus on automobile and...
But you included Doom that is using proprietary low-level extensions and why AMD has better performance by a fair margin (been shown by various reviewers Async Compute is not the primary performance gain in Vulkan Doom for AMD).
One could argue that is a biased game, Nvidia has only recently...
One needs to separate die and HBM in this context.
The GV102 is also a heavy lifter HPC GPU and is a primary Tesla card over £5k, with 24GB GDDR5X.
In fact GV102 is Nvidia's HPC FP32 card by a fair margin, and also DL card for massive performance inferencing (int8).
The HBM is more niche for...