![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Async compute gets 30% increase in performance. Maxwell doesn't support async.
- Thread starter StormClaw
- Start date
Cr4ckm0nk3y
Gawd
- Joined
- Jul 30, 2009
- Messages
- 847
That has been Hallock and Huddy's MO for a while now. Huddy's continued poor behavior and rhetoric is what put me off AMD, especially this gen.
From their interviews and social media anytime NV does anything that could be construed as better than AMD or faster than AMD both just spout off. Huddy on Gameworks was just sad. Like a spoiled child who didn't win.
I apologize to anyone I offending or yelled at during this whole ordeal.
I apologize to the HardOCP staff as well.
Now I can spend more time with my wife.
Dude you brought up a good point and theory no need to apologize for that, who knows nV might not have done anything till much later to address the issues?
Also the good thing is reviewers will be aware of this so reviews will be better.
An inaccurate point and an obvious FUD campaign.
AMD should focus on making better products and less time trying to run a smear campaign against the competition. It just makes them look bad and people clearly don't want to support their unethical way of doing business.
This is what happens when the information isn't readily available, I can't say Mahigan purposeful started that because no one really knew anything about what was going on, even at B3D after all those tests there was still confusion.
I can say it was very bad for Oxide to open up their mouth is such a manner before consulting nV about potential issues and fixes, and AMD's reps going ape sh*t over that, it was a tag team gangbang lol.
Factum
2[H]4U
- Joined
- Dec 24, 2014
- Messages
- 2,455
I apologize to anyone I offending or yelled at during this whole ordeal.
I apologize to the HardOCP staff as well.
Now I can spend more time with my wife.
That is all you learned?
So the next time you feel like you don't understand something...you'll do it all over again
Ffs...look, learn...adjust behaviour.
There is an army of angry fanboys, just waiting for the type of hype-Fuel you gave them!
Sorry just dosn't cut it IMHO
That is all you learned?
So the next time you feel like you don't understand something...you'll do it all over again
Ffs...look, learn...adjust behaviour.
There is an army of angry fanboys, just waiting for the type of hype-Fuel you gave them!
Sorry just dosn't cut it IMHO
Umm I didn't apologize because I was wrong. I apologized because I treated people wrong.
Let me copy paste the same response I posted somewhere else...
You do realize that the lack of Asynchronous Compute wasn't my theory right? That's something which disproved my theory...
The developer at Oxide claimed that, "afaik, Maxwell doesn't support Asynchronous Compute".
I made a big deal out of it in order to get nVIDIA to acknowledge the issue. It has been my goal from day 1, if you follow my many postings on overclock.net.
It remains to be seen what the final outcome will be. But if what David Kanter has stated is true... don't expect a miracle under high Asynchronous Compute workloads because at that point my theory is valid.
Oxide's Ashes of the Singularity makes mild use of Asynchronous Compute (Glare and Lighting). So no, my theory wasn't FUD.
See here: https://www.youtube.com/watch?v=tTVeZlwn9W8&feature=youtu.be&t=1h21m35s
Go to 1:18:00
Or read here: http://www.dsogaming.com/news/oculu...-is-best-on-amd-nvidia-possibly-catastrophic/
Swallow your anger.
I swear, I don't know who's worse... the Red fans or the Green fans
Factum
2[H]4U
- Joined
- Dec 24, 2014
- Messages
- 2,455
Before that post I simply took you for a misguided guy, with a superficial knowlegde of DirectX...
Now I think you are a real troll.
You took the time to create users on several forums, post misleading "conclusion" based on very thin data...and keep at it for days.
What do you do when you find you you were mistaken like a broken muppet?
A little "Ohh...I see.. well...sorry I guess...BUT NOT MY FAULT!"
You keep your copy/paste shite for yourself...you have poluted enough.
You're in for a REAL surprise...
5 days ago, Mahigan played his hand when discussing whether he would buy a 980Ti or not (you can see the post I'm talking about on OCN): http://www.overclock.net/t/1569897/...ularity-dx12-benchmarks/1380_30#post_24360610
a) Up until that point, I would ascribe Mahigan's enthusiasm as leaning purely on curiosity.
But the posts that followed (the one linked above) showed an increase in antagonism.
Now that doesn't mean I'm not thankful for the active discussion his curiosity had initially created.
b) My attention has since shifted to the words of Kollock - see his OCN profile here:
http://www.overclock.net/forums/posts/by_user/id/473364
a) Up until that point, I would ascribe Mahigan's enthusiasm as leaning purely on curiosity.
But the posts that followed (the one linked above) showed an increase in antagonism.
Now that doesn't mean I'm not thankful for the active discussion his curiosity had initially created.
b) My attention has since shifted to the words of Kollock - see his OCN profile here:
http://www.overclock.net/forums/posts/by_user/id/473364
TaintedSquirrel
[H]F Junkie
- Joined
- Aug 5, 2013
- Messages
- 12,691
He went off the rails on overclock.net for a day or two, got criticized a little bit, and then cleaned up his act. Someone who spends nearly an entire week scouring internet boards making these long posts will eventually get themselves caught up in it. I can only imagine how many times you can explain Nvidia's terrible async implementation before you would say some crazy shit, too.
It doesn't really matter if he has an agenda. I've said before I think he's an AMD employee; maybe an engineer who is familiar with Nvidia's fuckery.
It seems all of the people who are most informed on this topic, or at least the most vocal are AMD enthusiasts. Like I need the guy who owns dual 290X's telling me how terrible Nvidia is... Thanks, I get it.
Recent interviews with game devs and industry insiders citing Nvidia's troublesome performance is way more valuable than a few internet forum posts.
KickAssCop
[H]F Junkie
- Joined
- Mar 19, 2003
- Messages
- 8,330
Oh my what a laugh. AMD always so juvenile. Oxide another bunch of morons that didn't know what was going on and straight away claimed that nV does not support a feature without talking to them.
Poor AMD fans grasping at straws without actual proof. This is absolute comedy gold and a repeat of history over and over. Even AMD sponsored spokesman Mahingan posts an apology.
Just this year:
Fury is a 4k card
Fury is over clockers dream
Nano is priced at 650
A full blown smear campaign orchestrated over multiple forums
Card samples revoked
For nvidia
970 3.5 g ram issue
I think I will stick with the less unethical company that is absolutely raping the top of the line AMD products in real games out now.
Poor AMD fans grasping at straws without actual proof. This is absolute comedy gold and a repeat of history over and over. Even AMD sponsored spokesman Mahingan posts an apology.
Just this year:
Fury is a 4k card
Fury is over clockers dream
Nano is priced at 650
A full blown smear campaign orchestrated over multiple forums
Card samples revoked
For nvidia
970 3.5 g ram issue
I think I will stick with the less unethical company that is absolutely raping the top of the line AMD products in real games out now.
Agreed - if it wasn't for Kollock, I would have unsubbed that thread long ago - despite the entertainment value haha.
Its so true. Let me know when you're ready to unload that troublesome 980 Ti. PayPal ready.
5 days ago I was on my way back to Canada, emotionally compromised, leaving my wife behind because immigration Canada refused to let her in.
5 days ago I was pissed off and of course this translated into my posts. On top of that I had a ton of folks spamming me with negative comments, insults etc. I had to repeat myself around 10,000 times (exaggeration) while I had insults thrown my way.
I'm human. I apologize.
I don't work for AMD. I don't work for anyone affiliated with AMD. I've never worked for AMD and I don't have any ties to AMD. This issue is over for now. I could disappear. I'm not.
My current name, Mahigan, is Algonquin/Abenaki First Nations for Wolf. A Wolf doesn't chase after a ball. A Wolf doesn't follow anyone's orders. It's my personality to a T.
Last edited:
KickAssCop
[H]F Junkie
- Joined
- Mar 19, 2003
- Messages
- 8,330
For a guy whose wife wasn't let across the border and doesn't work for AMD, you sure had a shit load of time to run a smear campaign. I cannot understand the motivation behind it. I was curious as well, I didn't spent hours on multiple forums posting, reposting, picking up fights etc whilst my wife couldn't visit Canada. Stop with the bullshit really.
KickAssCop
[H]F Junkie
- Joined
- Mar 19, 2003
- Messages
- 8,330
He is down? I thought he was still fighting and made no apologies for spreading absolute FUD for 5 days.
It would be a smear if it weren't true. If it is true then it is journalism. My wife has been super supportive the entire time. Making me sandwiches and bringing me coffee as I worked tirelessly to obtain a response from nVIDIA.
So many folks have thanked me, compared to those who have thrown hate my way. That's why I know it was worth it.
If nVIDIA fix this issue, we all win... and that's the point. If they don't then it will force them to innovate... and that's the point. As a consumer... you will benefit.
Last edited:
KickAssCop
[H]F Junkie
- Joined
- Mar 19, 2003
- Messages
- 8,330
My patience is running out. Enjoy your campaign.
If GTX 980 Ti is a 4k card then Fury X is a 4k card.
Single GPU
http://www.computerbase.de/2015-08/...karten-von-r7-360-bis-r9-390x-im-vergleich/3/
SLI vs CF
http://www.hardware.fr/focus/111/crossfire-radeon-r9-fury-x-fiji-vs-gm200-round-2.html
http://www.computerbase.de/2015-07/radeon-fury-x-geforce-980-ti-crossfire-sli-ultra-hd-12k/2/
btw don't think that hardocp is the only site on the internet that does PC hardware testing. Their test suite is ridiculously small and heavily biased with - 3 Gameworks titles (Witcher 3, Farcy 4, Dying Light) 1 AMD GE title (BF4) and 1 neutral title (GTA V).
That was a real screw up by AMD. But thats a PR and marketing screw up. I would not call that unethical like what Nvidia did with GTX 970 where they misrepresented the specifications of a product and did nothing till they were called out by the public.
What has a product pricing got to do with ethics here ? If you don't like it don't buy it. Brent and Kyle want every new GPU launch to cause price changes and competition. Thats all fine but AMD is not in the best of financial shape to get into a price war with Nvidia. In the last few years every new product that has caused the competition to cut prices gets rave reviews (Gold rating) at hardocp. If a product fails to do that it gets average review (Silver rating). But the Fury X was hammered by hardocp in their reviews with no rating at all. There was a level of vitriol unseen in any previous hardocp review. Its as if hardocp had a score to settle with AMD and took their revenge with the Fury X review.
AMD's current products do not have the right cost structure to engage in a price war. Their GPUs are costlier to manufacture and perform lesser than Nvidia's competing GPUs. btw even a price war might not help AMD as their products are less desirable due to poor efficiency and a weaker brand. AMD know this generation is lost. If AMD try aggressive pricing Nvidia will reply and crush them. Nvidia has a lot more cash so AMD is smart not to engage in a price war. AMD knows they are supply limited on R9 Nano and will sell out at their current pricing. So why leave money on the table.
Nvidia's Maxwell architecture is very efficient. Definitely for DX11 its the better architecture. But when we see game developers like Crytek, DICE, Oxide tout AMD hardware for async compute and VR companies like Oculus VR tout AMD GCN's low latency VR features there seems to be a clue as to how Maxwell is so much more efficient. Maxwell gets a lot of its efficiency by doing quite a bit in software and thus cutting down on die area and power. We will see DX12 games starting from later this year and I would not be surprised if we see the same whole trend we saw with 780 Ti vs R9 290X where Fury X performs better than 980 Ti in DX12 games which use features like async compute.
Rightly so. TR and H don't deserve a sample for the trolling they have been doing with their Fury and Fury X reviews.
You think AMD''s acts are worse. Talk about being disillusioned. Nvidia flat out misrepresented the product specs. They did not rectify the mistake on the their own by coming clean on their own in the 4 months time they had before the public raised the issue. This is a unscruplous company of the highest order.
Want me to add more to that list of unethical actions from Nvidia.
1. Kepler performance fall - 780 Ti was undeniably the faster card when it launched. It was anywhere from 7-9% faster against a non throttling R9 290X (1 Ghz) at 1440p. Today as newer games have been added we see that gap melt away as Kepler has performed poorly in recent games in the last 12 months. There was sharp criticism for Witcher 3 performance which pushed Nvidia to fix performance in that game but for other games no such luck. If you take games released only in the last 12 months on average AMD R9 290X outright beats the GTX 780 Ti.
2. Gameworks - This is an evil beyond comparison. Here Nvidia not only harms AMD GPUs but they rubbish their own last gen Kepler GPUs with their black box code.
http://www.techspot.com/review/1000-project-cars-benchmarks/page2.html
http://www.computerbase.de/2015-05/project-cars-guide-grafikkarte-prozessor-vergleich/2/
When you see a 228 sq mm GTX 960 matching or beating a 551 sq mm GTX 780 you know Nvidia is really screwing over even their own faithful customers. In other non Gameworks games a 780 thrashes 960 by 30-35%.
http://www.techspot.com/review/979-battlefield-hardline-benchmarks/page3.html
And you still think Nvidia is more ethical. Ethics and Nvidia cannot be in the same sentence. Thats how far apart they are.
You mean you will stick to your favourite company. Hey if you have a preference admit it. But don't fool around telling Nvidia is more ethical than AMD.
Last edited:
Nvidia, quite wisely, doesn't respond to trolls.
What some people seem to be missing is that even if nvidia's implementation of async compute is sub-optimal compared to the competition -- which is completely feasible, as good engineering is all about compromises after all -- at this moment it remains largely unknown just how much of a role async compute is going to play in typical games on typical PCs.
All we know right now is that, at least for AotS, nvidia currently has the best performing card and is leading the benchmark pack.
If you want to work tirelessly and keep companies honest, why don't you call up AMD and demand they fix their trainwreck DX11 drivers?
Ok I can't believe you're making these comparisons. Let me point out what's wrong:
So out of 5 things you listed, ALL of them are pre-release marketing bullshit and noise, with the only real point being "card samples revoked". None of the other 4 points affects informed buyers assuming they read actual reviews before buying. The performance of the cards has not changed since the initial reviews. TPU even has a piece just on how (badly) Fury X OC's so there's that too.
As for refusing card samples, yes I agree it's bad practice, but again unless the "favored review sites" completely fake their results, it doesn't affect any informed buyer. Plus sites refused samples can always buy their own sample and review it, much like what KitGuru did with the Fury X after they were refused a sample. I'm almost certain [H]ard will run a review even if just to give a big "fuck you" to AMD.
Yes I get it, you bought the card based on the reviews, and the performance hasn't changed since the initial reviews. Sure, but people don't realize the opportunity cost in not coming clean until 4 months after release. See I bought my 970s a month after launch. In between Oct and Jan, AMD ran a firesale on their cards, then we had the Black Friday and Christmas specials, and I missed all of that. Then I decided to watercool my 970s literally 3 days before the big revelation, so I'm stuck with $300 in watercooling gear. Can I at least return my 970s for a refund? Nope! You know the biggest kicker for me is, had this info been conveyed even 2 weeks after launch, I still would've been within the return window. But of course coming clean only 4 months later pretty much guarantees there won't be a massive return wave. Brilliant isn't it?
I'm really not sure why you think AMD's pre-release bullshit to nVidia's 4-month-post-release bullshit is even remotely comparable in terms of the magnitude of impact and potential opportunity cost.
Last edited:
Sounds more like you are rattled that someone came along with posts that upset the orthodoxy that nvidia was superior in all things. Turns out that the current cards are weaker at one of the defining performance enhancing features of dx12, and you can't stand anyone pointing that out.
KickAssCop
[H]F Junkie
- Joined
- Mar 19, 2003
- Messages
- 8,330
I have no issues if my 980 tis do not perform in direct x 12 games. I will upgrade to whichever card does perform. I bought 980 tis for current games since I don't believe in future proofing non sense.
However, the amount of FUD I have read over the past week is comedic to me with AMD fans always grasping at the next sight of hope in terms of features and performance. We have all been there and with recent deception by AMD marketing and poor levels of support for current games I have no faith in them. If time proves me wrong so be it. I will be happy to install their next fastest card.
However, the amount of FUD I have read over the past week is comedic to me with AMD fans always grasping at the next sight of hope in terms of features and performance. We have all been there and with recent deception by AMD marketing and poor levels of support for current games I have no faith in them. If time proves me wrong so be it. I will be happy to install their next fastest card.
KickAssCop
[H]F Junkie
- Joined
- Mar 19, 2003
- Messages
- 8,330
By same virtue none of the buyers who bought 970s suddenly got less performance than they saw at release. I was pissed as much as the next guy and hence I put it under bad practices in this year by nvidia.
you realize it was 970 users like goldentiger (who was a strong Nvidia supporter on hardforums and ocn) and other such Nvidia users who raised the issue of inconsistent frametimes and stuttering when the last 0.5 GB was accessed especially in SLI and at high res. How pathetic of you to make that statement that the 970 users got the same performance as at launch. If you have to find out actual specifications of a product by playing games and testing and running other benchmarks thats disgusting. To top it all you have to then ask the GPU vendor why you are seeing performance issues and then their CEO tells you that its a new feature. Disgusting is the word I have for Nvidia's behaviour .
That is not how everyone else who buys a GPU looks at it. Most people don't buy a GPU every 1-2 years. I would go so far as to say that the people who do that are a minority. Since you don't believe in future proofing does not mean everyone else also does the same. There are many people who don't change a GPU for 3-5 years. In fact there are lots of people still running GTX 580/GTX 680 and HD 6970/HD 7970. Fermi was undoubtedly a much more forward looking architecture than VLIW. GTX 580 dominates HD 6970 in DX11 tesselation and compute performance. Today we know that HD 7970 was the more forward looking architecture/product than GTX 680. Most people don't want to see their GPU purchase show its age within a year. People like you are an exception as you are always on the bleeding edge.
Look who is talking about poor support for current games. Did you know your favourite company's last gen flagship 780 Ti which used to perform better than R9 290X now regularly fails to even match it in the latest titles.
I think this illustrates the difference between AMD fans and NVIDIA fans.
For me it's just a hobby, for you and others it seems to go off the rails a bit.
Factum
2[H]4U
- Joined
- Dec 24, 2014
- Messages
- 2,455
I think this illustrates the difference between AMD fans and NVIDIA fans.
For me it's just a hobby, for you and others it seems to go off the rails a bit.
Two major problems with his behaviour:
1. Creating multiple users, spreading false information based on ignorance and a misguided sense of that NVIDIA should provide the information he wants...when he wants it.
2. Not taking responsibilty for the FUD he spread.
Now watch him work just as tirelessly in order to admend this, watch him make an equal amount of posts on all the forums he has users on now declaring he was off the rails...oh wait...the muppet still dosn't think he did something wrong
Dude I specifically addressed the "you bought the card based on the reviews, and the performance hasn't changed since the initial reviews" bit, and talked about why the impact can extend beyond just performance. It's all in my post above.
I think this illustrates the difference between AMD fans and NVIDIA fans.
For me it's just a hobby, for you and others it seems to go off the rails a bit.
Your posting habits would suggest otherwise, unless you mean your hobby is to throw dirt at AMD every chance you get, then yeah I agree.
Last edited:
TheOCerIceMan
n00b
- Joined
- Jun 8, 2015
- Messages
- 14
I couldn't help but to comment in this thread that you guys are hilarious. In the end we're all going to die and video cards will outlive us. How about that for a reality check
ManofGod
[H]F Junkie
- Joined
- Oct 4, 2007
- Messages
- 12,864
I think this illustrates the difference between AMD fans and NVIDIA fans.
For me it's just a hobby, for you and others it seems to go off the rails a bit.
You are right, one just slings mud (Nvidia), the other works at getting us good information that we can learn things about(AMD). However, he is not a fan of either company and I appreciate the time he puts into these things.
You, on the other hand....
Are you trying to justify your $1300 dollar purchase? AMD has their issues and I prefer their products but lets get real, future proofing is not non sense, especially for folks who do not have money like that to just throw around. AMD fans now what reality is and now what they prefer to use, well Nvidia folks seem to go bat crap crazy over anyone saying anything positive about AMD.
Me, I may just buy this game to support a company that does not cave to Nvidia's pressure tactics.
Polluted? He is the only one that actually has provided anything of real substance here. Almost everyone else here has essentially done what you have done, shoot the messenger since it does not agree with your world view of Nvidia perfection.
How about you take the time to do something you care about instead of just flinging mud?
Last edited:
ManofGod
[H]F Junkie
- Joined
- Oct 4, 2007
- Messages
- 12,864
Unless I am misunderstanding, software async is not really the same thing as supporting hardware async. It would truly be interesting to see Nvidia respond but, if it does not fit their narrative they develop, I do not think that will happen. Oh, and before someone comes in a yells AMD FANBOY!, yes, I prefer AMD but, I am aware of their screw ups and it gets really frustrating as well. (Does not change what NVidia does though. Bumpgate? nForce3 gate?)
Last edited:
There is no such thing as only software async as it is not viable in a real time application, what are the queues being set up by the CPU? And then the GPU puts the compute instructions into the graphics pipeline, what is the latency going to be if that happens. I can see where you can have a second queue being lined up by the CPU and the GPU is doing the first queue, that makes sense.
http://www.ertl.jp/~shinpei/papers/icpads13.pdf
Just look at how the latency is enormous, we aren't talking microseconds here we are talking milliseconds.
There is much to consider to do "software" async, the DMA, bandwidth from the motherboard bus, bandwidth of IO on both the GPU and CPU, IO latency on both sides.
Last edited:
Lol...
What's true Async, then? Compute and graphics being done at the same time? Please give us your definition.
I have to admit, you have some guts to come in here without any form of evidence, tells of a pretty senior member of Beyond3D he's a noob. I salute you.