Support NeoGAF

[SonGoku]

Yeah, while Jaguar was doodoo, even the fastest Intel single core at standard clocks is only ~6x more powerful without overclocking. For one core, it's hard to find much faster anywhere without blowing up your power budget.

So what do you make of those claims "bigger CPU jump in PlayStation history" bigger than ps2 to ps3.

 

[Negotiator]

Don't get me wrong, its a huge increase i guess i was led to believe the gap over jaguar would be 2 gens worth (10X+)

Something like this perhaps?

Sony PS4 review: T3's official PlayStation 4 review

Despite the PS4 being superseded by more powerful systems, the PlayStation 4 remains a top draw gaming console thanks to a vast and varied games library

www.t3.com

* 4 core (2 core pairs) 3.2 GHz AMD x86 (Steamroller)
* aggregate, 10x PS3 PPU performance

Yes, PPU was that slow.

 

[LordOfChaos]

So what do you make of those claims "bigger CPU jump in PlayStation history" bigger than ps2 to ps3.

Honestly I didn't really know anyone was making a claim like that. Those 10x jumps are going to be incredibly hard to come by from here on out, barring exotic materials.

 

[SonGoku]

Something like this perhaps?

Sony PS4 review: T3's official PlayStation 4 review

Despite the PS4 being superseded by more powerful systems, the PlayStation 4 remains a top draw gaming console thanks to a vast and varied games library

www.t3.com

* 4 core (2 core pairs) 3.2 GHz AMD x86 (Steamroller)
* aggregate, 10x PS3 PPU performance

Yes, PPU was that slow.

I read ZEN2 had higher performance (floating point) than the SPEs which is crazy

Honestly I didn't really know anyone was making a claim like that

would you agree that its a bigger jump than ps2 to ps3?

 

[ethomaz]

So what do you make of those claims "bigger CPU jump in PlayStation history" bigger than ps2 to ps3.

I think you are making some mistakes.

A CPU IPC improves from a gen to another just 10-20%... if you reach something like 30% air more it talked like a miracle.

And yes Jaguar to Zen in consoles is probably the biggest jump in CPU ever... we are talking about over 200% in IPC plus double the clock.

 

[LordOfChaos]

I read ZEN2 had higher performance (floating point) than the SPEs which is crazy

would you agree that its a bigger jump than ps2 to ps3?

Both were Weird Sony era, but I'd have to say no, this was more worlds apart

Edit: There's some mistakes here (Cell clock speed), but even if you scale it down, the differences are much bigger

 

[SonGoku]

A CPU IPC improves from a gen to another just 10-20%... if you reach something like 30% air more it talked like a miracle.

I meant console gens, sorry I should had been more specific.

And yes Jaguar to Zen in consoles is probably the biggest jump in CPU ever... we are talking about over 200% in IPC plus double the clock.

Bigger than PS1->PS2 & PS2 ->PS3?

 

[ethomaz]

An IPC increase for AMD and Intel... over 50% happened twice in the last 15 years.

 

[ethomaz]

I meant console gens, sorry I should had been more specific.

Bigger than PS1->PS2 & PS2 ->PS3?

I believe yes... most because the Jaguar last gen was too weak so barely being a jump over Cell.

The Emotion Engine is way more stronger than these high level specs charts says... it had 2 Vector Units that help in graphics too just like Cell.

Remember what Kitase said.

Kitase on Emotion Engine for PS2: "PS2 really was a very high performance graphics machine"

Ken Kutaragi is the main man behind PlayStation. He was the one who developed PlayStation 1 to 3. He is known for his crazy ambitions and talks in

gearnuke.com

“That would be the fact that the PS2 was such a great platform. Of course, the PS3 and PS Vita have better specs but with the Emotion Engine at its core, the PS2 really was a very high performance graphics machine.”

When they ported games from PS2 to PS3 and Vita they had a big bottleneck... CPU.

Sony took years to create a PS2 emulator because the same reason... CPU.

 

[demigod]

So 5x over jaguar best case scenario? Thought the gap would be higher.

Didn’t someone claim that it’s only going to be 2x stronger in the old thread?

 

[LordOfChaos]

And yes Jaguar to Zen in consoles is probably the biggest jump in CPU ever... we are talking about over 200% in IPC plus double the clock.

I'm not sure how you're saying that, 2x IPC, 2x clock speed, some extra magic and it's 5x the single threaded performance and about the same ratio multithreaded.

Emotion Engine - Cell went from 10 million to 235 million transistors, 6.2Gflops in floating point to over 200.

The PS1 before it was a 33MHz core with a 5KB cache...

Both of those were at least over one order of magnitude.

“That would be the fact that the PS2 was such a great platform. Of course, the PS3 and PS Vita have better specs but with the Emotion Engine at its core, the PS2 really was a very high performance graphics machine.”

When they ported games from PS2 to PS3 and Vita they had a big bottleneck... CPU.

The VPUs are included in the total chip floating point afaik. And were those ports even trying to use SPUs at that point? Most everyone hit early CPU bottlenecks because the PPU was easier to use.


Besides, if you tailor to any weird platform like the EE or Cell, there's going to be edge cases where a more powerful system will stumble on a straight port.

 

[SonGoku]

Didn’t someone claim that it’s only going to be 2x stronger in the old thread?

That would imply that either IPC or clock speed stays the same, which wouldn't make sense

 

[ethomaz]

I'm not sure how you're saying that, 2x IPC, 2x clock speed, some extra magic and it's 5x the single threaded performance and about the same ratio multithreaded.

Emotion Engine - Cell went from 10 million to 235 million transistors, 6.2Gflops in floating point to over 200.

The PS1 before it was a 33MHz core with a 5KB cache...

Both of those were at least over one order of magnitude.



The VPUs are included in the total chip floating point afaik. And were those ports even trying to use SPUs at that point? Most everyone hit early CPU bottlenecks because the PPU was easier to use.

That means nothing in terms of performance... like I said the jump Jaguar to Zen is terrific... Emotion Engine to Cell was not in the same ballpark and Cell to PS4 was barely a upgrade lol

 

[LordOfChaos]

That means nothing in terms of performance... like I said the jump Jaguar to Zen is terrific... Emotion Engine to Cell was not in the same ballpark and Cell to PS4 was barely a upgrade lol

So how about PS1 to Emotion Engine? I'm very hard pressed to believe this is the biggest jump in PS history when the 90s were known for CPUs being out of date by an order of magnitude in years, now that's very hard to come by.

This is looking like a 5x gain, PS1 to PS2 at minimum was an order of magnitude.

You're right that the spec sheets aren't everything but they aren't nothing, and the same way, CPUs don't spend all their time running peak SIMD floating point in games all day, Jaguar would be far better than Cell in anything branchy.

Just to give another example...

A P4 1.6Ghz is what needed in PC to match performance of Emotion Engine without the use of the 2 Vector Units.

Now compare that with Cell.

If you tailor to any weird platform like the EE or Cell, there's going to be edge cases where a more powerful system will stumble on a straight port. The Wii took a while to emulate well on far more powerful PC CPUs, no one thinks it was powerful in its own right.

 

[ethomaz]

Just to give another example...

A P4 1.6Ghz is what needed in PC to match performance of Emotion Engine without the use of the 2 Vector Units.

Now compare that with Cell.

 

[ethomaz]

Let’s make a table...

- PS1 CPU matches Pentium MMX (160-200Mhz)
- PS2 CPU matches Pentium 4 1.6Ghz
- PS3’s Cell matches Phenom X3 (2.5Ghz?)

Now find the improvement in performance from Pentium MMX to Pentium 4 to Phenom X3... you will be surprised.

PS. That is not the exactly matches but around what I believe the PC at time needed to run the same game at similar quality.

 

[LordOfChaos]

Let’s make a table...

- PS1 CPU matches Pentium MMX (160-200Mhz)
- PS2 CPU matches Pentium 4 1.6Ghz
- PS3’s Cell matches Phenom X3 (2.5Ghz?)

Now find the improvement in performance from Pentium MMX to Pentium 4 to Phenom X3... you will be surprised.

PS. That is not the exactly matches but around what I believe the PC at time needed to run the same game at similar quality.

Linpack:
Pentium MMX @ 200MHz - 23
Pentium 4 @ 1.7GHz -382

Linpack Benchmark Results - Roy Longbottom's PC benchmark Collection

Roy Longbottom's PC benchmark Collection - Linpack Benchmark Results On PCs

www.roylongbottom.org.uk

Soooo....

I said at least an order of magnitude and that's a 16x gain in a real test. How was that not a bigger jump than 5?



Besides that, how did you match these CPUs back? What is needed from a CPU to emulate a system (at the same playing performance) is far greater than the original CPU. You need a faster still CPU to match a 700MHz PowerPC G3 from the Wii, yet no one thought it was as powerful as the PC CPUs needed to play it. Was the MIPS R3000A really doing 6x per clock what a Pentium was?

 

[ethomaz]

Linpack:
Pentium MMX @ 200MHz - 23
Pentium 4 @ 1.7GHz -382

Linpack Benchmark Results - Roy Longbottom's PC benchmark Collection

Roy Longbottom's PC benchmark Collection - Linpack Benchmark Results On PCs

www.roylongbottom.org.uk

Soooo....

I said at least an order of magnitude and that's a 16x gain in a real test.

Besides that, how did you match these CPUs back? What is needed from a CPU to emulate a system (at the same playing performance) is far greater than the original CPU. You need a faster still CPU to match a 700MHz PowerPC G3 from the Wii, yet no one thought it was as powerful as the PC CPUs needed to play it. Was the MIPS R3000A really doing 6x per clock what a Pentium was?

?

Ryzen R7 1800X do 220k in the benchmark lol

 

[LordOfChaos]

?

Ryzen R7 1800X do 220k in the benchmark... it is way over 16x over Jaguar lol

Where do you see a Jaguar result for Linpack?

The Athlon 5150 is Jaguar at 1.6GHz, I see no benchmarks showing beyond 16x gains? Or even close to them

ASUS All Series vs AMD Qogir - Geekbench

 

[CyberPanda]

Yeah, the moment he says "NVidia" is where I say "bullshit".

The dreamcastguy has been known to be a pathological liar.

 

[TLZ]

The dreamcastguy has been known to be a pathological liar.

Always felt something's off with him.

 

[psorcerer]

Interesting thought though that for doing a serial CPU tasks, the most expensive supercomputer in the world won't do much better than a humble gamers rig.

That's why the whole discussion about "cpu power" is so stupid.
Modern gaming machine does not need CPU. At all. If the software (game) is designed correctly, and optimized correctly.
But most of the modern games (multiplatform) are built by 1000s of people in sweat shops where nobody cares about design or optimization. That's why they whine 24/7 about Jaguar, their game code which should use maybe 1% of the frame time suddenly uses 50%. Because level designers, game designers and such are usually not engineers, and even Lua is "too hard" for them. And then it's too late to address it.

 

[Ar¢tos]

That means nothing in terms of performance... like I said the jump Jaguar to Zen is terrific... Emotion Engine to Cell was not in the same ballpark and Cell to PS4 was barely a upgrade lol

You can't compare directly a SIMD oriented CPU with a MIMD oriented one just by the flop count.
Its like comparing a bullet train to a F1 car simply by saying one is a bit faster than the other.

 

[ANIMAL1975]

"possibly hit 8k natively
has 14 teraflops of cpu power inside this
16 gigs of ram
256 Go SSD
PS5 = GTX 1660
...."

 

[Aceofspades]

Blast from the past:
DF: Orbis(PS4) vs Durango (X1) Specs analysis

Notice how Rich tried his hardest to mitigate the obvious power advantage of PS4.

So does the GPU difference translate into as large an advantage as it sounds? VGleaks' Orbis spec, again derived from platform holder documentation, suggests that four of these CUs are reserved for Compute functions, conceivably bringing the PlayStation's raw advantage down from 50 per cent to just over 16. However, while Compute is often used for elements like physics calculations, there's nothing to stop coders hiving off specific graphics features to this hardware - Just Cause 2, for example, used NVIDIA's own Compute solution, CUDA, for enhanced water effects, while a core element of Battlefield 3 - the deferred shading solution that power its beautiful lighting - is handled via DirectX 11 Compute shader code.

Apparently more ROPs in PS4 in overkill and not needed, 16 in Xone is sufficient for 1080p gaming Lol

Other information has also come to light offering up a further Orbis advantage: the Sony hardware has a surprisingly large 32 ROPs (Render Output units) up against 16 on Durango. ROPs translate pixel and texel values into the final image sent to the display: on a very rough level, the more ROPs you have, the higher the resolution you can address (hardware anti-aliasing capability is also tied into the ROPs). 16 ROPs is sufficient to maintain 1080p, 32 comes across as overkill, but it could be useful for addressing stereoscopic 1080p for instance, or even 4K.

Teraflops difference is moot .

There's an argument that suggests that comparing Durango and Orbis on these terms is not realistic; that the platform holders have far more control over the design of the silicon than the raw specs suggest; that they can be adapted with manufacturer-specific 'secret sauce' customisations. The raw teraflop measurements being mooted - 1.23TF for Durango and 1.84TF for Orbis - have been dismissed as meaningless, and to a certain extent that is true.

Xbox Secret customisations

But here we do see some intriguing enhancements that are Durango-specific. Its 'Data Move Engines' carry out hardware compression as well as decompression (and support for JPEG too - perhaps to handle Kinect camera streams), while there is also support for texture swizzling. However, the main takeaway here is that core elements of the Move Engine functionality are apparently designed to extract the best performance from a RAM set-up that is much more complex (and slower) than its Orbis equivalent.

Pffff, Games doesn't need ultra fast bandwidth on PS4, if they did there is Esram!!!

However, while the disadvantages are obvious, this is not to say that the situation is anything like a complete disaster for Durango development. Speaking to game makers, the impression we come away with is that not every feature in a game actually requires ultra-fast memory. Systems will be developed on the DDR3, and if memory throughput becomes an issue, those features will be ported over to the ESRAM where there's enough bandwidth to provide the raw performance if needed.

The magic of DirectX lol.

Also mitigating the difference to a certain extent is the fact that Durango operates under an enhanced version of DirectX 11 - dubbed internally DirectX 11.x. It's highly likely that crucial rendering functions will automatically be optimised by Microsoft for use with the ESRAM.

He even added both DDR3 and Eram bandwidth together for it to match GDDR5 on PS4

Man, It was fun browsing that thread, hopefully you can get as much enjoyment of it as I did....until we get a legit leak for PS5 .

 

[Psajdak]

Putting aside exclusives, style, and design choices (for instance, to me some games like Chrono Trigger, or Final Fantasy IX visually look more pleasing than something released in recent years), do you think that powerwise, if someone simply wants the most insane graphics ever, they should just opt for PC now, instead of waiting for PS5 to come?

 

[TLZ]

Blast from the past:
DF: Orbis(PS4) vs Durango (X1) Specs analysis

Notice how Rich tried his hardest to mitigate the obvious power advantage of PS4.


Apparently more ROPs in PS4 in overkill and not needed, 16 in Xone is sufficient for 1080p gaming Lol



Teraflops difference is moot .



Xbox Secret customisations



Pffff, Games doesn't need ultra fast bandwidth on PS4, if they did there is Esram!!!



The magic of DirectX lol.



He even added both DDR3 and Eram bandwidth together for it to match GDDR5 on PS4

Man, It was fun browsing that thread, hopefully you can get as much enjoyment of it as I did....until we get a legit leak for PS5 .

Wow. To that extent.

 

[FranXico]

Blast from the past:
DF: Orbis(PS4) vs Durango (X1) Specs analysis

Notice how Rich tried his hardest to mitigate the obvious power advantage of PS4.


Apparently more ROPs in PS4 in overkill and not needed, 16 in Xone is sufficient for 1080p gaming Lol



Teraflops difference is moot .



Xbox Secret customisations



Pffff, Games doesn't need ultra fast bandwidth on PS4, if they did there is Esram!!!



The magic of DirectX lol.



He even added both DDR3 and Eram bandwidth together for it to match GDDR5 on PS4

Man, It was fun browsing that thread, hopefully you can get as much enjoyment of it as I did....until we get a legit leak for PS5 .

OMFG what a sleazy shill. I vaguely remembered the damage control, but it was far more ridiculous than I remember.
How could anyone take him seriously since? ROFL

 

[Aceofspades]

Wow. To that extent.

Even more than that, Richard Leadbetter is a known Xbox fanboy and he is the head of DF. It has been obvious for years.

 

[FranXico]

Putting aside exclusives, style, and design choices (for instance, to me some games like Chrono Trigger, or Final Fantasy IX visually look more pleasing than something released in recent years), do you think that powerwise, if someone simply wants the most insane graphics ever, they should just opt for PC now, instead of waiting for PS5 to come?

If you want insane graphics, and have the money, always PC.

 

[TLZ]

Even more than that, Richard Leadbetter is a known Xbox fanboy and he is the head of DF. It has been obvious for years.

I don't mind people having a preference. It's human nature. But at least be professional when delivering such content to the public. Tell it like it is.

 

[Aceofspades]

I don't mind people having a preference. It's human nature. But at least be professional when delivering such content to the public. Tell it like it is.

I agree with you completely, the issue magnifies even more since he is doing tech comparison with numbers. We know numbers don't lie so spinning them will just make him look like a clown .

 

[FranXico]

I agree with you completely, the issue magnifies even more since he is doing tech comparison with numbers. We know numbers don't lie so spinning them will just make him look like a clown .

Well, I'm pretty sure that he stopped putting spins on face-off numbers ever since the X1X came out...

 

[xool]

A bit late but :

I just realized... there are more different things... for example, memory frequency, memory size and memory width... he added those lines too.

yeah 4096 bit bus indicates HBM or HBM2 - that would be 16 chip(lets) at 2x 128 bit each - so for instance 4 chips each with 4GB (made of 4 1 GB dies stacked) - but that's 16GB not 18GB - there's no real sane way to get 18GB from 4096 bit bus - unless it's counting an additional 2GB DDR4 that isn't showing elsewhere..

I think the fakes are just intentionally trolling people now - a game of spot the mistake - that took a little effort - and tbh the half-double thing looks like it was made deliberately wrong..

 

[TLZ]

I agree with you completely, the issue magnifies even more since he is doing tech comparison with numbers. We know numbers don't lie so spinning them will just make him look like a clown .

Yea. With a platform like DF, with many people trusting you, you don't want to look silly.

 

[Aceofspades]

Well, I'm pretty sure that he stopped putting spins on face-off numbers ever since the X1X came out...

Actually I wish he simply stopped spinning, instead he started overblowing every miniscule difference touting how MS engineered the perfect console. He is just a MS shill, which now is accompanied with "Dictator" another MS shill at DF. Only natural person there is John who does "Retro" series and he usually does the best game analysis out of the three.

 

[ethomaz]

Microsoft Next Generation Systems - Pastebin.com

Pastebin.com is the number one paste tool since 2002. Pastebin is a website where you can store text online for a set period of time.

pastebin.com

 

[ethomaz]

BTW Zen 2 and Navi reveal are today... AMD leaker are really excited with what they saw about Zen2 but seems like they were asked to hold until the livestream.

 

[Fake]

Do you honestly think the PS5 can actually have PS3 BC by running games natively?

Lol. I never said that. You're saying Xbox One X have BC with Xbox/360 without messing with VRAM, but Xbox X never had BC. In his middle life was add a software BC with is different from native BC.
PS5 will have native hardware BC with PS4.

 

[LordOfChaos]

?

Ryzen R7 1800X do 220k in the benchmark... it is way over 16x over Jaguar lol

Where do you see a Jaguar result for Linpack?

The Athlon 5150 is Jaguar at 1.6GHz, I see no benchmarks showing beyond 16x gains? Or even close to them

ASUS All Series vs AMD Qogir - Geekbench

C'mon man, don't leave me hanging lol. Do you see a result that shows Ryzen over 16x faster than Jaguar single core or was that a mistaken assumption?

 

[Aceofspades]

Microsoft Next Generation Systems - Pastebin.com

Pastebin.com is the number one paste tool since 2002. Pastebin is a website where you can store text online for a set period of time.

pastebin.com

WTF is 2nd Generation Scorpio Engine?

So Anaconda around 10TF?

 

[Negotiator]

Interesting:

AMD Zen 2 6 Core CPU Benchmark Leaked, Faster Than Ryzen 7 2700X in Geekbench 4

Another day another Zen 2 Ryzen 3000 series CPU leak and to call this one a shocker would be an understatement! So let's get straight into it! A peculiar CPU entry has been spotted in the wild, specifically in Geekbench's database. The entry carries the same string we've seen on all other...

wccftech.com

Video games will have access to 6 x Zen 2 cores minimum.

 

[ethomaz]

C'mon man, don't leave me hanging lol. Do you see a result that shows Ryzen over 16x faster than Jaguar single core or was that a mistaken assumption?

You already made up your mind and choose a benchmark that didn’t work on Jaguar cores... it is a know issue... but if you take results from sucessora of Jaguar you will realize that in that benchmark the difference to Zen is over 10x... so I said To Jaguar cores de difference is even bigger.

But hey you are putting too much power in the increase IPC over the generations... Pentium 3 even today has a pretty good IPC because the increases are really small.

That gen is probably the biggest CPU jump we ever see in a console hardware.

 

[ethomaz]

You can't compare directly a SIMD oriented CPU with a MIMD oriented one just by the flop count.
Its like comparing a bullet train to a F1 car simply by saying one is a bit faster than the other.

Where I did that? I never even talked about flops count.

I talked about relative performance... what you need in PC at time to hold the same graphics and performance that the console I was comparing.

PS1 - Pentium MMX 200Mhz
PS2 - Pentium 4 1.6Ghz
PS3 - Phenom x3 2.5Gh
PS4 - Jaguar x8 1.6Ghz
PS5 - Zen2 8x 3.0Ghz?

 

[Ar¢tos]

Where I did that? I never even talked about flops count.

You said the ps4 cpu was barely an upgrade compared to the CELL.
I would consider it a sidegrade, because the most important change was in architecture, not overall power.

 

[LordOfChaos]

You already made up your mind and choose a benchmark that didn’t work on Jaguar cores... it is a know issue... but if you take results from sucessora of Jaguar you will realize that in that benchmark the difference to Zen is over 10x... so I said To Jaguar cores de difference is even bigger.

But hey you are putting too much power in the increase IPC over the generations... Pentium 3 even today has a pretty good IPC because the increases are really small.

That gen is probably the biggest CPU jump we ever see in a console hardware.

What's a sucessora (edit: oh, successor? So where's the Puma result?)? You've been saying this was the biggest jump in PS CPU history and compared each to specific CPUs, I found a benchmark that let me compare them and showed the MMX-P3 jump was over 16x, another that shows Jaguar to Ryzen 2 is ~5. You keep making these claims but haven't showed anything...I'm just trying to figure out if you have something or not, I'm not going to do all the work here.

If this was a bigger jump than an MMX-P3 I'd be thrilled, prove me wrong if I'm wrong, but I see zero benchmarks corroborating this.

 

[ethomaz]

You said the ps4 cpu was barely an upgrade compared to the CELL.
I would consider it a sidegrade, because the most important change was in architecture, not overall power.

I believe it is still a upgrade but it is a pretty small one in performance... Cell is hard to work so that alone like you said is already a big jump.

 

[Imtjnotu]

Microsoft Next Generation Systems - Pastebin.com

Pastebin.com is the number one paste tool since 2002. Pastebin is a website where you can store text online for a set period of time.

pastebin.com

Lol uh what 32 gigs of ram.

 

[ResilientBanana]

A bit late but :





yeah 4096 bit bus indicates HBM or HBM2 - that would be 16 chip(lets) at 2x 128 bit each - so for instance 4 chips each with 4GB (made of 4 1 GB dies stacked) - but that's 16GB not 18GB - there's no real sane way to get 18GB from 4096 bit bus - unless it's counting an additional 2GB DDR4 that isn't showing elsewhere..

I think the fakes are just intentionally trolling people now - a game of spot the mistake - that took a little effort - and tbh the half-double thing looks like it was made deliberately wrong..

Those look like they were typed in Notepad++

 

[Negotiator]

You said the ps4 cpu was barely an upgrade compared to the CELL.
I would consider it a sidegrade, because the most important change was in architecture, not overall power.

Definitely not true:



Keep in mind that the PS3 only has 1 CPU core, not 3.