Intel AMD Agreement.

[soccerballtux]

Its not good for any company's business if this doesn't get renewed.

That said, I never fully understood why Intel so staunchly refused to cross-license the necessary IP with Nvidia.

Its not like Nvidia had (or has) the resources necessary to create an Intel-killer x86 CPU, and Intel sure could have used some of Nvidia's graphics IP.

-because Nvidia actually has a clue about marketing
-because it would hurt AMD more than it would hurt Intel, and that's something Intel can't afford
mainly #1 IMO

 

[AnandThenMan]

-because Nvidia actually has a clue about marketing
-because it would hurt AMD more than it would hurt Intel, and that's something Intel can't afford
mainly #1 IMO

Intel and Nvidia HAVE a cross licensing agreement, why are some of you refusing to acknowledge this? Direct from Intel's site:
Intel Corporation today announced that it has entered into a new comprehensive long-term patent cross license agreement with NVIDIA.

 

[soccerballtux]

Intel and Nvidia HAVE a cross licensing agreement, why are some of you refusing to acknowledge this? Direct from Intel's site:
Intel Corporation today announced that it has entered into a new comprehensive long-term patent cross license agreement with NVIDIA.

I said nothing about those things.
What I'm talking about is Nvidia's clever stuff like this

Nvidia to blame (Score:5, Interesting)
by Anonymous Coward on Thursday November 13, 2014 @05:55PM (#48382141)

Nvidia plays the game every bit as dirty as Intel. In this case, Nvidia has created something called 'GAMEWORKS'- a proprietary closed-source library of routines specifically designed to collapse the performance of games on AMD hardware (or older Nvidia hardware). Nvidia pays shills to counter information like this in forums like this one, so let me give you one example.

The best current ANTI-ALIASING is a FREE, OPEN-SOURCE collection of methods from Crytek (the people behind Crysis and the original Far Cry). Their methods run with excellent performance on older hardware, and slightly favour AMD (because AMD hardware is always more shader-powerful than Nvidia at a given class). Not good for Nvidia. So Nvidia 'invented' TXAA- an horrifically bad AA method both in appearance and 'hit' on performance- but a method that runs far better on new Nvidia hardware than it does on new AMD hardware.

Nvidia actually PAYS developers like Ubisoft to NOT include the best, SMAA methods from Crytek (remember, they are free for any publisher to use). Instead, Nvidia only allows FXAA (also Nvidia created, but lightweight on all hardware, at the cost of not being so good), MSAA (the old fashioned hardware anti-aliasing that comes with horrible restrictions), and TXAA (hated even by Nvidia fanboys because of its impact on performance). EVERYONE is asking where SMAA T2X is on Unity- but as I said, Nvidia paid Ubisoft to exclude it.

TXAA is universally loathed (even HardOCP- the elitist PC gaming site that insists on benchmarking games with every possible setting set to max, regradless of the trade-off- stated that TXAA was such an atrocity, they'd always use SMAA in preference), but for Nvidia it is the perfect model for how they seek to ruin the gaming experience of everyone, in order to synthetically make Nvidia GPUs seem 'better'.

GAMEWORKS increases the number of TXAA like performance destroyers in a modern engine (Xbox One, PS4 or PC) exponentially. Ultra slow GPU libraries to handle trivial things like particles, AI pathfinding, occlusion calculations and the like. Remember, gaming PCs and new consoles are CPU rich- with CPU performance going begging across the commonplace 4-7 cores. No serious PC gamer runs less than a 4-core i5. The consoles have 8-cores a-piece.

Nvidia literally doesn't care if bouncing ten simple particles on your screen uses 30% of your GPU performance, so long as the same effect on an AMD GPU takes 80%. Nvidia is this dirty.

Disgustingly, Epic have taken a large Nvidia pay-off to make Gameworks the EXCLUSIVE 'enhancement' library of Unreal 4 (the current most successful licensed engine), and the team behind Witcher 3 (the most anticipated open-world fantasy game ever) have agreed to ruin the performance of that game on AMD GPUs (when it is released early next year) in order to gain Nvidia funding.

Remember how a week back, more than a decade after the crime, Intel got a TINY court punishment for paying sites like Anandtech to use bent Intel benchmarks 'proving' that the putrid Intel Netburst x86 CPUs were 'better' than the vastly superior (at the time) AMD CPUs? The owner of Anandtech himself made a point of informing his readers that one core was better than two (when only AMD had gone dual core), that 64-bit was pointless joke (when AMD invented x64, long before Intel licensed the tech from AMD), and that Netburst's intent to reach 10GHz showed that only Intel had the right tech and ideas.

Nvidia no more fears punishment (in the courts or court of public opinion) than does Intel. Nvidia relies on the vicious trolling of its PR teams to hurt its opponents, and to fool the public.

For how Unity looks (far, far from remarkable), it should run at least THREE times faster on given hardware, with the most pointless settings notched down. Or, it could be THREE times better at the current framerates- and truly appear 'next-gen'. Nvidia steals our possible, doable gaming experiences to enrich itself. Just as Intel loves bloated abstracted, buggy junk like .NET on Windows, because it synthetically needs a much more expensive Intel CPU to run well.

 

[dontsh00tmesanta]

How many people have had an amd vs just regurgitating what other people think lol

 

[AnandThenMan]

I said nothing about those things.
What I'm talking about is Nvidia's clever stuff like this

What's interesting to observe is both Nvidia and Intel are well know for using these tactics (legal or not) and both companies are doing very well. AMD is not so well known for these tactics and are struggling as a company. The lesson here? Stop being the "nice guy" and cripple your competitors as much as possible by any means necessary. Intel has built an empire doing this very thing.

 

[ShintaiDK]

What's interesting to observe is both Nvidia and Intel are well know for using these tactics (legal or not) and both companies are doing very well. AMD is not so well known for these tactics and are struggling as a company. The lesson here? Stop being the "nice guy" and cripple your competitors as much as possible by any means necessary. Intel has built an empire doing this very thing.

You should look at history. Nor nVidia or Intel put the nail in AMDs coffin. They happily did that themselves 10-12 years ago. A more arrogant company couldnt be found. They delayed node progress and uarchs while dreaming on selling highend ASP products only. While they came up with one marketing BS after the other. And still today its a company that have a very hard time with the truth while the arrogance still blossom.

Its a company with a culture and a PR problem.

 

[DrMrLordX]

I said nothing about those things.
What I'm talking about is Nvidia's clever stuff like this

Cool story, informative . . . but veering into the realm of off-topicness.

This thread went a predictable way.

We strive for consistency!

You should look at history. Nor nVidia or Intel put the nail in AMDs coffin. They happily did that themselves 10-12 years ago. A more arrogant company couldnt be found. They delayed node progress and uarchs while dreaming on selling highend ASP products only. While they came up with one marketing BS after the other. And still today its a company that have a very hard time with the truth while the arrogance still blossom.

Didn't AMD always depend on others for node shrinks, even when they did have fabs? Like, IBM and suchlike? Their process R&D never was all that.

Their decision to mark up certain CPUs (notably FX chips and various X2s) is really no different than what Intel does today (and has done at several points in time). Once they had the performance crown, they charged accordingly. They sure as heck weren't going to line their pockets through volume shipping.

Its a company with a culture and a PR problem.

PR seems to be slowly improving. Slowly. I would agree with the culture thing in maybe '05 when there seemed to be a good bit of resting-on-laurels. They came out and admitted, publicly, that the Construction series of cores were a failure, and Su seems committed to steering the company through choppy waters. At the present, I don't see many problems with truth or culture at AMD.

But hey guess what, all of what I just posted may also be stretching the limits of what does or does not pertain to the topic at hand, which is the x86-64 cross licensing agreement.

Yeah they're gonna renew that, if they haven't already. No doubt. The issue of what happens when Intel is no longer required to support PCIe remains of greater interest in my opinion.

 

[SunburstLP]

What's interesting to observe is both Nvidia and Intel are well know for using these tactics (legal or not) and both companies are doing very well. AMD is not so well known for these tactics and are struggling as a company. The lesson here? Stop being the "nice guy" and cripple your competitors as much as possible by any means necessary. Intel has built an empire doing this very thing.

Sort of, except the enthusiastic market-base they've created are a very noisy population. If AMD were to try and cripple their competitors as you posit, a) they probably wouldn't have the R&D budget* to do it right, resulting in a half-effort and b) based on (a) the enthusiast community at large would throw a giant hissy-fit while forgetting about the practices green and blue have been know to engage in from time-to-time.

My take on what AMD needs to do is simply make themselves relevant again. We/everybody knows that there isn't enough money for them to have a Conroe moment.

Start small; small design wins will improve workforce morale and start seeding the capital required to increase the necessary budgets. I'm hopeful that Dr. Su and company know this. If they snag a few semi-custom wins, they can dig themselves out. The fact of the matter is that they need a small fit of innovation to build a bit of momentum. Then they can leverage that momentum and keep nurturing it into some real stability.

Otherwise, they'll just struggle along and get swallowed up by another company or private investment group. edit here: I think the cross-licensing deal is too valuable for them to just go belly-up. Just my .02.

* I'm half joking here, btw. Shout out to all the usual suspects. :whiste:

 

[ShintaiDK]

Didn't AMD always depend on others for node shrinks, even when they did have fabs? Like, IBM and suchlike? Their process R&D never was all that.

Their decision to mark up certain CPUs (notably FX chips and various X2s) is really no different than what Intel does today (and has done at several points in time). Once they had the performance crown, they charged accordingly. They sure as heck weren't going to line their pockets through volume shipping.

The node (65nm) was delayed on purpose by AMD from 2005 to late 2006 in self confidence over the K8 design. Because how on earth would Intel be able to catch up? (After seeing 65nm P4.)

AMD with Hector in the front had the idea that they could limit themselves to selling highend desktop/servers only with huge margins. While the big masterplan was to let Intel handle the 60-70% of the mainstream/value segment that wasnt "worthwhile".

 

[SunburstLP]

Yeah they're gonna renew that, if they haven't already. No doubt. The issue of what happens when Intel is no longer required to support PCIe remains of greater interest in my opinion.

The PCIe question is indeed a good one. Is Intel in a position where being ruthless with interconnect standards for system required peripherals is something they can get away with?

The scenario is interesting to think about because it's not just affecting consumers but the bottom lines of other companies. What's Asus, MSI, Gigabyte, XFX, et al going to do when told they have to make two PCBs for every SKU they want in AMD and Intel machines. I just don't see it happening. There's way too much of a possibility of a metric-truckload of ill-will from every direction for Intel. I could be wrong.