• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD likely to win Nintendo NX and continue console dominance

“Your still talking about more than one generation and cherry picking unlike chips, which isn't what they said”
Yet again you misunderstood what was said. The quote was “20-100x faster than today's models' that is precisely what they did with series 6



“so ill leave you to your little fantasies. “
So you make a number of mistakes and instead of admitting that you resort to blaming fantasies? You know what it’s best you do leave this thread with an attitude like that.
 
I read an article that misquoted them on the desktop thing, Not exactly my mistake but ok.
They made the statement in 2011, so show me an actual product from 2014 that rivalled a top end desktop product from 2011? Show me any powervr product that rivals a gtx680 or 7970. They don't, its nonsense.

But you are still skipping over an extra 2 generations of chips to make your comparison, which their statement didn't cover. And it becomes irrelevant when it is paper specs and not actual chips in production. Comparing actual device benches from one generation of powervr to the next there is absolutely not a 100x improvement. You cant just randomly start comparing a bottom end chip from one gen to the top end chip from the next, no one ever does that, its nonsense.

Even if what they technically said turns out to be "true", the way that they've said it is heavily misleading to the point it becomes totally irrelevant.

Everyone, but everyone, talks about performance increases comparing like for like chips, e.g. Same die size or same power. To suddenly ditch all of that industry standard comparison and cherry pick paper specs just makes them laughable.
 
Last edited:
I am seriously comparing clusters why is that wrong? A cluster is not a core so why are you talking about muit cores? A 6 cluster chip with 2 cores would be 12 clusters. The fact is what you said about performance is wrong.

“same with your last comment... clearly if someone says "in 3 years our SoC's will rival desktop" they are not talking desktop from the past, they are talking desktop that are concurrent with the release date of the SoC, to suggest otherwise is also ********”This is the quote your link is based on

“The Series 6 PowerVR chips (codenamed 'Rogue') are expected to equal today's desktop PC processors – and by extension exceed today's HD consoles – within three years.” Clearly you have got confused and/or misunderstood something.

Find a benchmark, or review, or statement from AMD/Nvidia comparing Fury X to a 7350, or a Titan to a 730GT, you can't because they don't. These are companies that make incredibly misleading statements and benchmarks to showcase their stuff in the best light possible.... even then they will not make such comparisons nor every put such a comparison in marketing statements.

The entire tech industry and frankly the entire world talks about products in basically the same way. You can't randomly decide to start talking about a new product being X times faster, with a completely worldwide, industry wide, business wide standard of how these things are compared... and just use a new comparison model as if it's nothing.

You compare and talk about architectures by comparing like to like, the highest end chip of one architecture to the highest end chip of another, or same architecture and same design but on different processes, etc. There is always a basis for comparison whereby the products are similar in usually at least two or three way.

Pottsey, the reason no one anywhere takes you seriously is because of this rubbish.

Can you honestly find a single AMD or Nvidia fanboy who ran about like a raving looney talking about a new generation product, claiming it was 20 times faster because they compared say a new 980 gtx to a 630gt? The most rabid of Nvidia fanboys in the entire world and I've never once seen such an unrealistic comparison nor claim? Should that not tell you how abnormal and ridiculous it is.


Again they themselves are talking about this going into LOW COST Android consoles. Nothing more, nothing less, sticking a tablet class SOC into a tiny set top box and shipping with a controller, done with almost every currently shipping ARM SOC there is already to run basic mobile games on a bigger screen. That does not make it capable of scaling up to a current gen proper console level performance. Where are you deciding that their products can scale up to desktop level. You can't magically take a low power design and add 50 clusters and have it just work, that isn't how chip making works at all.

Which precise CPU would they use for a high performance console today, which GPU would they use for that console, where is their history/experience in combining those two parts into a high end SOC/APU that they have experience making and shipping in volume?
 
Consoles makers don't want 2 chip designs, like in 360 or ps3 they want all in one like a apu and that is AMD or Intel really as they have apu's in production at the required spec's or there abouts.
 
First of all I apologies to drunkenmaster for ignoring your post and questions. I did not mean to and only just spotted your questions while looking for a link earlier in the thread.


“Exactly what console experience are you referring to for Imagination. Scaling up chips is a VERY different prospect to scaling down chips. Having memory controllers and general connections across the core capable of handling very high bandwidth can be scaled down easily. If you design something for ultra low power and low throughput, scaling it up becomes effectively impossible, you need to make something new.
To answer your question on experience PlayStation 2, Dreamcast and on a smaller scale for a gaming device PlayStation Vita. I agree scaling up chips is a very different prospect but the chips in question have been designed from the ground up to scale up to console level performance. PowerVR change the way the GPU is designed and I believe are the first to use clusters? The mix of clusters and cores get around the problems. Due to the tile approached doing different tiles the design scales up very well to consoles level. As for very high bandwidth one of the main advantages of PowerVR designs even when they were in the desktop space is they never needed very high bandwidth. A tile based deferred render needs a fraction of the bandwidth of what AMD or NVidia do while doing the same workload. They don’t need to scale bandwidth up to the high levels AMD or NVidia use.

As for CPU they could use something like a 6 core MIPS I6400 Warrior P-class 64-bit CPU’s. Or if that’s not enough another high end custom parts.


“Lastly, Imagination themselves stated quite specifically they see this latest chip as a good chip for LOW COST ANDROID CONSOLES....”
Where? All I see are consoles. Not Android Consoles. Also if that’s the case why did Nintendo approach them?


“What chip do they have that is suitable for the next gen console. In fact what SOC are PowerVR shipping today, I'm unaware of any.”
The SoC list is too large to post here and I don’t really fancy digging through the massive amount of data. They do well over 1billon units a year. They could use a 16 cluster XT family chip with or without Wizard to make the console stand out. A 16 Cluster 1 single or muti core chip with Wizard setup could to some very impressive graphics. Mixing Ray tracing with more common shaders and graphics.


“Imagination and powerVR ship and have IP for nothing at the scale of the AMD console APUs, nothing remotely close.”
I would say it’s the other way around. AMD have nothing to compete against a 16 cluster muti core Wizard setup. Anyway why can AMD make a custom part but Imagination cannot? Imagination have made custom parts for others like Sony and Apple.


“What Imagination themselves are saying this next gpu IP they have they see being used by say Samsung/Qualcomm/whoever else as part of those companies SOC and shipping in android console boxes, as many of their previous gpu ips have also been used in exactly the same way.”~
To be fair to you I wouldn’t expect you to know this as you don’t go to the meetings and I assume don’t go to the GPU trade shows. But IMG have been talking about TFLOPS range designs and consoles a lot recently. designs all the way up to compute servers.
 
You compare and talk about architectures by comparing like to like, the highest end chip of one architecture to the highest end chip of another, or same architecture and same design but on different processes, etc.
That’s just what I did. Well not chip for chip I compared core for core as that was what the statement was about. The problem with your argument is I was right. Core for core targeting comparable markets PowerVR Series6 is well beyond x20 the performance of Series 5. More so when you look at compute power. In some areas core for core Series 6 is well beyond x20 the performance of Series 5.

How am I acting like a raving looney when the statements are correct?


“Again they themselves are talking about this going into LOW COST Android consoles. Nothing more, nothing less, sticking a tablet class SOC into a tiny set top box and shipping with a controller, done with almost every currently shipping ARM SOC there is already to run basic mobile games on a bigger screen.”
Please show me where you are getting that from. They might have made some reference to low cost android consoles, but they also referenced consoles.




“Where are you deciding that their products can scale up to desktop level. You can't magically take a low power design and add 50 clusters and have it just work, that isn't how chip making works at all.”
The designs show they scale up to 16 clusters per cores then you can add extra cores. I decide that based on what’s said at the meetings. I am not talking about scaling up to Titan X level. 50 clusters is beyond the max limit per core. Up to 16 clusters per core is what its deigned around but I cannot remember the max core count right now.



“They made the statement in 2011, so show me an actual product from 2014 that rivalled a top end desktop product from 2011? Show me any powervr product that rivals a gtx680 or 7970. They don't, its nonsense.”
I will do that when you show me a statement they made saying they will rival a top end desktop produce from 2011. Even I am not that crazy to make that statement. Your right its nonsense as no statement like that was ever made.



“But you are still skipping over an extra 2 generations of chips to make your comparison,”
I am comparing the series 6 family against the series 5 family core for core which is what the original statement was about. Series 6 was built around GPU compute and is massively faster core for core then series 5. The only reason you find it heavily misleading is because you misunderstood the statement as you used a new source, based on a news source, based on a news source which lost the original text and meaning. Let me quote one of the original statements.

“Delivering the best performance in both GFLOPS/mm2 and GFLOPS/mW, PowerVR Series6 GPUs can deliver 20x or more of the performance of current generation GPU cores targeting comparable markets. This is enabled by an architecture that is around 5x more efficient than previous generations.”

http://imgtec.com/news/press-release/interim-management-statement-imagination-technologies-8/

Should we move this PowerVR talk into the recent PowerVR ray tracing thread I posted? It’s gone a bit off topic from AMD and PowerVR and Nintendo
 
Last edited:
That is what they said, they said that within 3 years their SoC would rival desktop

And now you are massively moving the goal posts, they said 20-100x, why even bother mentioning 100x if what they later relied upon was 20x

I didn't misunderstand their statement, they made a statement that was heavily exaggerated, using a system of comparison that no other chip maker uses.
And again, show me an actual in the shops device that shows a 20x or 100x improvement in frame rate going from one generation to the next.
 
Last edited:
For reference, Samsung or Qualcomm taking ARM IP and Imagination IP, pairing it together then making a chip is absolutely not like saying Imagination ship a billion SOCs, nothing at all. PowerVR ship next to no units, they do not make the CPUs in most SOCs and there is literally thousands of little bits of IP, little blocks, security amongst dozens of others, that go into a SOC besides the GPU and CPU.

Name a SOC that Imagination or PowerVR designed and shipped themselves because I can't think of one. Name a higher end CPU designed for the throughput that a higher end console would need that IMagination make through MIPS, which is effectively none. I'm under the impression even on CPU side Imagination ship nothing at all in regards to CPUs they make themselves, 99% of the MIPS cpus that ship from other companies such as Broadcom have very limited scope, they are application/device specific and don't cater to the more wide ranging use that a console/PC would bring.


The Dreamcast and PS2 CPUs are MIPS based but that is IP sold by MIPS(now part of Imagination), nothing more. Hitachi built the CPU for the dreamcast, Sony/Toshiba appear to have implemented the cpu/gpu on the PS2. I can't actually work out what the GPU(it's part of the same core) in the PS2 is based off. Either way again you're talking about companies buying a license for IP off a company and implementing the actual design, integration and working themselves.

The majority of powerVR gpus are implemented alongside ARM CPUs and merged together by third parties, Samsung, Qualcomm, many others. Imagination has zero(afaik) volume shipping products that are complete SOCs of entirely their own IP. They have ZERO direct experience in creating a console, none at all. Other parties used some of Imaginations IP for consoles they put together, they did all the R&D for and of which they would own all the implementation patents for.

You're comparing apples to oranges.

AMD designed, built and implemented every part of the PS4/Xbox One SOC, Samsung designed, built and integrated various others IP including their own, to make all their CPUs. That is experience in putting pieces together, getting them working, creating a working software stack and driver model for the entire package, taping it out, shipping what is a much more complex design, mass producing, packaging, shipping, supporting, etc.

Saying that Imagination has the same experience because they sell IP licenses to hundreds of companies who then implement it is basically ridiculous.

Arm at least actually provide IP for entire SOCs of which many companies take the license for the exact SOC and produce precisely that chip. Imagination can't even say that because there are no complete Imagination SOCs out there.
 
That is what they said, they said that within 3 years their SoC would rival desktop

And now you are massively moving the goal posts, they said 20-100x, why even bother mentioning 100x if what they later relied upon was 20x

I didn't misunderstand their statement, they made a statement that was heavily exaggerated, using a system of comparison that no other chip maker uses.
And again, show me an actual in the shops device that shows a 20x or 100x improvement in frame rate going from one generation to the next.

I have found zero statements saying 3 years their SoC would rival high end desktops. I don’t remember them ever saying that either. You need to prove that. It’s not fair for you to have a go at me or Imagination about a statement neither of us made.

andybird123 & drunkenmaster you two might owe me an apology after saying I am speaking rubbish and having a go at me for the core comparison I did. Been digging though old quotes and I found the source of the x100 numbers.

An Interview from 2011
Q: Finally, we see Apple talking about a 9x performance increase from iPad 1 to iPad 2 and benchmarking of the devices sees at least a 4x "real world" boost in GPU performance. With SGX535 as the baseline, what are your performance targets for your hardware going forward. I'm sure I read somewhere you were looking at a 100x increase within five years...
IMG: Yes we are.

Source: http://www.gamesindustry.biz/articles/digitalfoundry-powervr-tech-interview

The series 5 SGX535 was very clearly asked to be used as the baseline. They also said within 5 years. So how are goal posts moving? How has anything been heavily exaggerated? The system of comparison they made was in direct response to a question asking for that system of comparison to be used.

“And again, show me an actual in the shops device that shows a 20x or 100x improvement in frame rate going from one generation to the next”
But that is not what the statement said. No one ever said a x100 improvement in 1 generation. It was a x100 improvement over 5 years with the SGX535 as a baseline. Well I already gave you in this thread the FPS numbers with a x100 improvement with the old SGX535 chips.
 
“Arm at least actually provide IP for entire SOCs of which many companies take the license for the exact SOC and produce precisely that chip. Imagination can't even say that because there are no complete Imagination SOCs out there. “
There are complete Imagination SOCs out there. Trying to get this back on topic the fact is if Nintendo want a complete SOC or part of a SoC from Imagination then Nintendo can license the complete or part SoC design up to console level performance. If Nintendo want a custom part that can be done as well.


“Saying that Imagination has the same experience because they sell IP licenses to hundreds of companies who then implement it is basically ridiculous.”
The very fact they sell IP licenses to hundreds of companies right up to complete SoC’s and custom SoC shows they have the experience of putting designs together.
 
If you notice, your latest find references statements they'd already made. Its clear they made headline grabbing statements about a "hundred times increase", the fact you are having to find increasingly hyperbolic statements to "prove" they were right goes to show how pointless those statements are.

If you notice, you actually posted an image that shows them claiming a 700x improvement over 5 years

The comparison you made was also not of products that were available 5 years apart, but more like 7, and longer if you take an actual "on the shelf" date
 
Last edited:
A very cursory look shows that their latest GPU scales up to GT730M performance, so surely they aren't/won't be relevant in the console space for a number of years yet...

Because let's face it most people talking about consoles are talking about PS4/XBone levels of performance not some Android console.

However I would love to see ImgTec make a comeback, but it doesn't look imminent in the PC/console space according to their own marketing materials.
 
Nintendo is getting into phone games, I think they are even rumoured to be releasing a platform for it.

Yes they are Orangey. Havnt heard about a specialist device to run them though as they already have one really in the 3ds. Been saying this for years but they should just make games for other consoles. Two out of their last three home consoles have tanked in the game cube and the wii u because they try to compete with MS and sony on a graphical side rather than innovation
 
try to compete with MS and sony on a graphical side rather than innovation

Since when is the wii u competing with this gen consoles?, it is way under powered never mind hardly any 3rd party support, a lot of none gamers brought into the hype of the wii and after a while a lot of those people got bored of it.

With the shrink and HBM2 next year they could have a console out the at least matches ps4 and with their games it would sell.
 
Back
Top Bottom