News and Rumours: PS4

I spent this much time answering you for nothing. You don"t get that trinity in its desktop form may bring really few on the table va llano. Neither you noticed that amd has been pretty vocal on improvements made by trinity on the low power apus.

As for kaveri we only know it launches in 2013. Early is your specualtion.

See trinity was expected.a bit earlier and it will be sometime in Q2. For kavery as.we said we don"t but AMD has been late with almost every release whether it was cpu or apu. They also faced.availability problem. I'm saying you that for a launch in 2013 so tesging the final sku building enough stock etc. It"s a.pretty risky bet.
I'm not.saying that the.hadware would not make sense kaveri + cap verde pro would be great. But it"s.risky, really risky may AMD be late on its schedule and SONY launch could compromised, either missing the timeline or.facing availability issues, etc.

If Sony don't make.that huge bet or know AMD roadmap better than investors which would.surprise me as they can't really afford.to underwhelm them ( or they know the product will be out later.than they would have wanted, not a good news either for sony) then llano + matching architecture ( southern island) is a given vs a blend of two architectures northern/sourthern island.

And stop answering by only quoting website claims, those claim are not undisputable and it's no a proper way to discuss.
 
I spent this much time answering you for nothing. You don"t get that trinity in its desktop form may bring really few on the table va llano. Neither you noticed that amd has been pretty vocal on improvements made by trinity on the low power apus.

As for kaveri we only know it launches in 2013. Early is your specualtion.

See trinity was expected.a bit earlier and it will be sometime in Q2. For kavery as.we said we don"t but AMD has been late with almost every release whether it was cpu or apu. They also faced.availability problem. I'm saying you that for a launch in 2013 so tesging the final sku building enough stock etc. It"s a.pretty risky bet.
I'm not.saying that the.hadware would not make sense kaveri + cap verde pro would be great. But it"s.risky, really risky may AMD be late on its schedule and SONY launch could compromised, either missing the timeline or.facing availability issues, etc.

If Sony don't make.that huge bet or know AMD roadmap better than investors which would.surprise me as they can't really afford.to underwhelm them ( or they know the product will be out later.than they would have wanted, not a good news either for sony) then llano + matching architecture ( southern island) is a given vs a blend of two architectures northern/sourthern island.

And stop answering by only quoting website claims, those claim are not undisputable and it's no a proper way to discuss.
I'm a amd investor and they do not let us in on any inside info on road map or anything else BC if they did it would be against the law if they didn't tell everyone. I'm sure if Sony is working with amd they would know better than anyone else.


Like I have already said using trinity brings piledriver and works with the rumor gpu already in the dev kit.

I think we just have to disagree. I find no way in the world why sony would use your design in the final ps4. Why wait till 2013?

I think it will use a sea island gpu and kaveri. AMD has already said it "betting the company on HSA" why not show it off on the PS4?
 
Last edited by a moderator:
I just don't have much faith anymore in Sony's ability to fight against the Koreans, Apple, MS and Nintendo. I feel like Sony and many (most?) other Japanese companies are either getting outright ravaged or becoming significantly weaker out there. Sony are like a sitting goose unable to move and getting hammered from all sides. Hirai is going to have to pull some golden ducklings out of the hat soon.
 
Can someone please tell me how many flops can a supposed Trinity and Kaveri combo yield for the PS4?
Trinity we still don't know the GPU clock speed. It's supposed to be faster than in llano.

Kaveri should end around 1TFLOPS.

Just in case you don't know the calculation is pretty straight forward, in all AMD last design every SP can achieve 2 FLOPS per cycle, so for example llano has five SIMD, 80 SP per SIMD and run at 600 MHz.
That's 400SP, FLOPS calculation goes as such 400*2*600(/1000 depend on what you consider mega giga, etc.) = 480MFLOPS.

Note that it's only relevant when comparing AMD vliw5 processor. Those design are more 4+1 with a special unit for some advance calculations like trigonometry, I'm not sure what they do but it should easy enough for one to find saint read on the matter :)
SO clearly to compare to vliw4 like NI or scalar like GCN you almost need to remove 20% to this figure: that's 384MFLOPS

Actually for some people the relevant figure is half that as if multiply-add count as 2 FLOPS not all the operations are multiply add. For example when Sweeney said he needs 4TFLOPS to achieve something close to Reyes in realtime, he speaks of vanilla FLOPS (not multiply add which are way more marketing friendly x2 / still some of thess operations could be FMA).

For Kaveri AMD it self leaked that it should reach around 1TFLOPS (multiply add too), it unclear if that include the CPU. It will also include 8 compute units (if we count as packs of ALU that 5 simd for llano trinity vs 8 for kaveri), it's a meaty improvement over previous APU.
Like in CAyman there is 64 alu per CU/SIMD, so that's 512 alu.
@900MHz that's ~922MFLOPS.
THe CPU should account for around 100MFLOPS at 3GHz. So depending on whether or not they counted the CPU FLOPS the gpu could be either south of 9600MHz or south of 1GHz.

Kaveri is a massive increase in GPU power altogether, that's pretty clear. The issue is more with timeline. Will AMD deliver soon enough? Will the chip achieve it premises? for things like power consumption, what will be the yields? etc. The consideration have more to do with business and production than actual technical merit of the design which Kaveri should win hand down ;)
 
Last edited by a moderator:
At this rate, the future of your PSN content will be purely hypothetical, like what you'd be doing today if you still had your SegaNet account.
 
Trinity we still don't know the GPU clock speed. It's supposed to be faster than in llano.

Kaveri should end around 1TFLOPS.

Just in case you don't know the calculation is pretty straight forward, in all AMD last design every SP can achieve 2 FLOPS per cycle, so for example llano has five SIMD, 80 SP per SIMD and run at 600 MHz.
That's 400SP, FLOPS calculation goes as such 400*2*600(/1000 depend on what you consider mega giga, etc.) = 480MFLOPS.

Note that it's only relevant when comparing AMD vliw5 processor. Those design are more 4+1 with a special unit for some advance calculations like trigonometry, I'm not sure what they do but it should easy enough for one to find saint read on the matter :)
SO clearly to compare to vliw4 like NI or scalar like GCN you almost need to remove 20% to this figure: that's 384MFLOPS

Actually for some people the relevant figure is half that as if multiply-add count as 2 FLOPS not all the operations are multiply add. For example when Sweeney said he needs 4TFLOPS to achieve something close to Reyes in realtime, he speaks of vanilla FLOPS (not multiply add which are way more marketing friendly x2 / still some of thess operations could be FMA).

For Kaveri AMD it self leaked that it should reach around 1TFLOPS (multiply add too), it unclear if that include the CPU. It will also include 8 compute units (if we count as packs of ALU that 5 simd for llano trinity vs 8 for kaveri), it's a meaty improvement over previous APU.
Like in CAyman there is 64 alu per CU/SIMD, so that's 512 alu.
@900MHz that's ~922MFLOPS.
THe CPU should account for around 100MFLOPS at 3GHz. So depending on whether or not they counted the CPU FLOPS the gpu could be either south of 9600MHz or south of 1GHz.

Kaveri is a massive increase in GPU power altogether, that's pretty clear. The issue is more with timeline. Will AMD deliver soon enough? Will the chip achieve it premises? for things like power consumption, what will be the yields? etc. The consideration have more to do with business and production than actual technical merit of the design which Kaveri should win hand down ;)
Well I guess we're looking at something 1.5TFLOPS+ then, still not enough for 1080p Samaritan but anything is better than what's been rumored.
 
Seem like it would be an easy sell.

Anything else then what I lay out would be crappy. The other apu are not impressive at all.
It's not that Kaveri is not attractive, neither that I think that GCN is not worse it.
If you haven't read the techreport review of for example cap verde GPU read it, it appears that those gpus perform way better than just "averages" used in all the other reviews would let you think.
Those GPU using average may appear like may be too sligh improvement vs the 57xx/67xx they replace which are old design but when you dive into TR review you realize that even the hd7750 win way more than it loose to something like a gtx460.
Really potent GPU I would said "balanced" not spike in perfs either up or down it seems to strike a sweet spot.

It's more about timelines and AMD have prove anything but reliable on that matter.
-------------------------------------
UltraGPU I didn't notice that you meant really a trinity + Kaveri combo, not Trinity+ GPU or kaveri +GPU.
Trinity + Kaveri doesn't make sense, different CPU different GPUs, if SONy want to make things complicated for the dev that's the way to go. That would be weird after moving to X86.

The only valid combinations are llano + Southern Island and kaveri + GCN. I'm confident that even more exotic blend that an already exotic APU+GPU design won't happen.
 
Last edited by a moderator:
I just don't have much faith anymore in Sony's ability to fight against the Koreans, Apple, MS and Nintendo. I feel like Sony and many (most?) other Japanese companies are either getting outright ravaged or becoming significantly weaker out there. Sony are like a sitting goose unable to move and getting hammered from all sides. Hirai is going to have to pull some golden ducklings out of the hat soon.

They are in a really bad/terminal spot. They don't have the hardware anymore to beat what's coming out of other asian nations or even Japan itself be it in tv's, computers, audio, and even consumer grade camcorders now as Panasonic and Canon are beating them in reviews pretty consistently. Likewise they don't have the software skill to beat American companies like Apple or Microsoft in providing a good user experience. They are also terrible at planning/investing in future prospects. Finally they don't have a series of products that they can connect together via services like Apple and Microsoft do to demonstrate value/user experience. It's the reason I don't talk much about them anymore at all in any future thread because I don't think they have one, I expect them to become more of a software service on someone elses platform one day just like with Sega. Basically in mind mind it's over, and the ps4 will be their last hurrah unless there is some monumental change in the company. Going with that I have no idea how they can possibly launch the ps4 as a heavy loss leader anymore, I just don't see how they can possibly keep hemorrhaging money at the rate they do.
 
They are in a really bad/terminal spot. They don't have the hardware anymore to beat what's coming out of other asian nations or even Japan itself be it in tv's, computers, audio, and even consumer grade camcorders now as Panasonic and Canon are beating them in reviews pretty consistently. Likewise they don't have the software skill to beat American companies like Apple or Microsoft in providing a good user experience. They are also terrible at planning/investing in future prospects. Finally they don't have a series of products that they can connect together via services like Apple and Microsoft do to demonstrate value/user experience. It's the reason I don't talk much about them anymore at all in any future thread because I don't think they have one.

Yeah imo it's hard to see a way out of this slump for them. They don't have a clear competitive edge in anything anymore. Their products aren't bad by any means, but they aren't good enough to sell the volumes they need. I hope they can find something to turn things around and rejuvenate themselves, but when I look at their different CE products all I see is dead ends, some closer some a bit further. I feel that their current best execution would only slow the fall instead of stopping it, something big would need to happen to change that.
 
Oh man. As if things didnt look bad now this:
http://www.gametrailers.com/side-mi...to-6-4-billion-reportedly-cutting-10000-jobs/

This is bound to affect Sony's strategy and spending on the PS4.

:???:

I wonder what would they even spend? There is no PS3 level spending needed to make a console. PS3 was a huge fuckup. They get the tech from AMD and that´s it. Such tech that can feasible fit in a console that can be sold for $399 at launch and will projected to be profitable in a few years

They will be in a better position if they launch at the same time as MS
 
Wrt to the question "what can they do?" May be a early 2013 launch (and not a world wide one) could gave them a nice lead start, that's if MS doesn't launch in 2013.
Other than that there in a tough spot for sure.
 
Guys sony is not broke. They still have $10billion plus in the bank. The job loses are really double counting companies they have sold.

Amd was never that problem in the timeline. It was gf. Its the reason they split.

Time line works perfect for a late holiday 2013 release. Sea island will launch late this year.
It's not that Kaveri is not attractive, neither that I think that GCN is not worse it.
If you haven't read the techreport review of for example cap verde GPU read it, it appears that those gpus perform way better than just "averages" used in all the other reviews would let you think.
Those GPU using average may appear like may be too sligh improvement vs the 57xx/67xx they replace which are old design but when you dive into TR review you realize that even the hd7750 win way more than it loose to something like a gtx460.
Really potent GPU I would said "balanced" not spike in perfs either up or down it seems to strike a sweet spot.

It's more about timelines and AMD have prove anything but reliable on that matter.
-------------------------------------
UltraGPU I didn't notice that you meant really a trinity + Kaveri combo, not Trinity+ GPU or kaveri +GPU.
Trinity + Kaveri doesn't make sense, different CPU different GPUs, if SONy want to make things complicated for the dev that's the way to go. That would be weird after moving to X86.

The only valid combinations are llano + Southern Island and kaveri + GCN. I'm confident that even more exotic blend that an already exotic APU+GPU design won't happen.
 
Last edited by a moderator:
Oh man. As if things didnt look bad now this:
http://www.gametrailers.com/side-mi...to-6-4-billion-reportedly-cutting-10000-jobs/

This is bound to affect Sony's strategy and spending on the PS4.

:???:

Or this is the outcome of Sony's strategy and spending on the PS4.. Like Stringer before him, Hirai is very keen on rationalising the company, a 6% cut from the budget to offset the R&D being done elsewhere. The majority seems to have come from offshoots in the chemical industry etc.
 
I wonder what would they even spend? There is no PS3 level spending needed to make a console. PS3 was a huge fuckup. They get the tech from AMD and that´s it. Such tech that can feasible fit in a console that can be sold for $399 at launch and will projected to be profitable in a few years

They will be in a better position if they launch at the same time as MS

There was an interview done recently with their CTO in which the budget for the PS3 was sort of discussed and the cost of the PS4 was delicately touched upon. The CELL cost them $400million alone in the PS3, and the PS4 is their first $1billion project.

I've been trying to find the interview, it was in a EE magazine I think.
 
Back
Top