WTH IGN?! They posted the spec analysis from Major Nelson!!

[PARANOiA]

But if the rest of the world is like Americans, we like to argue over baseball stats and the like. Game console numbers are no different. We may not know exactly what they mean, but gosh darn it we have an opinion!

That's the best wrap up of E3 and the console war as of yet.

 

[Proforma]

This PR stunt has me laughing and crying. I mean really, it smacks of desperation. After MS's bad start, the show floor was apparently sounding enthusiam for XB360, and then they along and offer a 'technical performance analysis' of hardware they know nothing about, incorporating bald-faced lies.

Has anything like this ever happened before, when a company has officially compared side-by-side their unlaunched product with their competitor's unlaunched product?

Who's idea was this? Earlier Allard seemed to be okay saying 'we're playing our game, the competitors are playing theres' but now they seem to be reacting (overreacting) in a really inprofessional...it's just crazy! I can't believe they actually wrote all that.

I guess this is the era of Next-Gen HDFUD

yeah like Sony didn't lie at all.

Sony put out enormous FUD about their system. 2 Tflops! Why didn't they just say 6! A lie is a lie

hahahahahahahahahahahahahahahahaha

Not only did Sony put out the most outlandish specs in the world, but they also put out movies instead of actual gameplay.

Talk about the ultimate FUD. Mindless drones thinking the PS3 can render everything on the planet in realtime and games that are all real time games (when in fact they are just pre-rendered CG movies).

 

[rabidrabbit]

Both of them lie, both of them take stabs at competitors.
It's stupid fanboyism to claim either of them lies more, or is using more dubious tactics to downplay others, or who said what first as if you really had tracked everything they said or showed.

They both do it and that's pretty much expected as they're direct competitors of a big entertainment business.
It's all a big entertainment show, very little of it is true or based on real facts.
Consoles are marketed to kids and young adult(male)s.
You can see similar exggarated marketing more or less everywhere, from cars to tampons.

Even Nintendo "lied", although many pretend they're honest as can be.
Maybe not the specs, but I remember the first GCN footage of for example Donkey Kong Racing that was shown as what the machine could do, now it's clear they were prerendered cgi... just like others.

Btw. did anyone read the Allard interview on http://www.eurogamer.net/article.php?article_id=59297 ?
He must be a reader of B3D console forums
That "Namco Girl PS2 demo was classic

Eurogamer: Was that not also a tech demo then?

J Allard: That was not tech demo. That was rendered.

Eurogamer: But didn't you use the same trick on Gotham 3 the other day?

J Allard: [Bounds up and down in chair, waves arms] We didn't, we didn't... First; I mean there's no trick because it was a combination of in-game footage and rendered footage, just like videogames are. So, I mean we had a combination of both those elements throughout. All the stuff that we did show that was game footage was all running on Alpha kits though, so I think it's coming along really really well.

You show the best that you can show. We tried to show things that were representative of what you're going to be able to buy. That was our part. In some cases it was lower than what we'd like people to think this system is capable of, but it's real, and we wanted to show people stuff that was real and say 'hey, launch is around the corner and you're going to buy games that look like this, you're not going to be disappointed'.
We didn't want to set an expectation, and I think there was an expectation, just talking as a gamer, of Namco Girl. Remember Namco Girl on PS2, I remember saying 'I can't wait to get that game'. That game doesn't exist. That game is going to exist; the fidelity of the graphics of Namco Girl for PS2 will first ship on Xbox 360 called Dead Or Alive 4. Dead Or Alive 4 will have that kind of fidelity, but that's years after Namco Girl.

I'd say that "Namco Girl" was already done on xbox DOA, DOA Volleyball... heck, even Tekken TAG the 1st gen PS2 title did her.
Unless Namco really are making a "Reiko Nagasi" game on xbox360, in which case that statement would at least have some sense.

 

[Shifty Geezer]

yeah like Sony didn't lie at all.

Sony put out enormous FUD about their system. 2 Tflops! Why didn't they just say 6! A lie is a lie

That was a direct response to MS's claims of 1 teraflop for XB360 months earlier.

Not only did Sony put out the most outlandish specs in the world, but they also put out movies instead of actual gameplay.

Which they didn't say were real. They ALSO put out very impressive realtime tech demos, if you watched the whole thing!

What Sony haven't done is mispresent their technology to create a totally inane number like "system aggregate bandwidth".

Sony hype. MS hype. But MS have escalated the war to the next level of atomic hype-of-mass-confusion. It's got out of hand.

 

[MrWibble]

And in fact the PS2 can do that.

I dont recall any ps2 games having more than 1-2M vertices per second. In contrast both nintendo gamecube and xbox have better models and higher poligon counts than the ps2, the difference is they claymed way lower numbers.

How would you know? I mean seriously, other than running a title through a PA and looking at stats, how would you know the difference between a title running 2M v/s and 20M v/s?

For that matter, would you know the difference between a game using 2 TFlop/s and 1Tf/s? or 6Gf/s?

I think the point being raised here, and it's a very good one, is that people generally don't *understand* what the numbers mean anyway, they just like comparing them to try to guess which machine is going to be "better". And on that assumption, the companies don't need to get "real world" numbers, they just need to be using roughly equivalent calculations to give peak numbers and hopefully the numbers will be useful enough for comparison purposes.

In the PS2 and XBox world I think most XB "peak" numbers are about double the PS2 numbers. Personally I think that although XB titles do mostly look better than PS2 ones, the difference isn't quite 2x. However I think Sony do more to help people get close to the theoretical performance.

If anything, Sony are a lot more honest about the performance than the opposition sometimes are. Sony are the only current company who have released full details of their hardware to the public, including complete manuals and presentations detailing how to optimise for it.

This: http://www.technology.scee.net/sceesite/files/presentations/gdce2002/HardTuningPS2.pdf shows a lot of details including a scan of a title (and lets be honest here, anyone that's use a PA can probably spot a Jak+Daxter scan a mile away) running at 10M polys/s or more, and that translates to between 10 and 30 million verts/s (probably nearer the lower bound of that if the stripping is good). Ok, even if you assume 15M v/s as an average that's only 25% of the theoretical, but it's certainly in a "game" situation with a lot of other stuff going on (lighting, animation, skinning, AI, light physics, etc.)

Accusing companies of lying because you don't understand what they're telling you, is a bit harsh. The PS2 manuals are "out there" so if you feel that one of their claims is not actually possible with the hardware they released, feel free to make a specific point.

 

[nAo]

and lets be honest here, anyone that's use a PA can probably spot a Jak+Daxter scan a mile away

Good quote the most analyzed game on the PS2

 

[DeanoC]

It's most definately not representative (at least on consoles). Besides if you want to go just by "one developer" examples, I've once heard of an app that used 15MB of main memory for just strings...
...
on a PS2...
:?

Sudeki allocated originally 32Mb to graphics, went upto 40MB IIRC.
When we first thought about HS X360 we started by allocating 128Mb for graphics (out of the original 256Mb).

We've got 256Mb RAM of GDDR on PS3 which sound like a nice start.

I personally start with an opening bid of 50/50 on a new platform and then see which part of the studio bribes me the most ;-)

BTW By graphics mem I include framebuffers, DMA chains etc.

 

[[maven]]

Deano,
so you think you'll end up with another 256mb of game and sound data? I find that a bit excessive.
For a PC game I've done ages ago (ignoring DX memory-backed surfaces), the usual case was approximately 64mb graphics, about 25 "game" and ~10 sound. And that was running a fair bit of simulation, but no physics.

 

[AlNom]

We've got 256Mb RAM of GDDR on PS3 which sound like a nice start.

So what we saw at E3... that was using 256MB for the graphics ?

 

[DeanoC]

So what we saw at E3... that was using 256MB for the graphics ?

The demo is a bad example, lots of uncompressed stuff with things higher res than they need to be and we had restricted visuals (no real level shown). But saying that we used about 200Mb, a hefty portion of that was because we don't any normal map compression yet.

 

[Gubbi]

....I've once heard of an app that used 15MB of main memory for just strings...
...
on a PS2...
:?

Zork I, II & III for PS2 ?

Cheers
Gubbi

 

[nAo]

if there aren't big penalties having the GPU working on the xdr ram and the CPU working on the gddr ram I would split workload between both memory types to exploit more bandwith.

 

[Fafalada]

Sudeki allocated originally 32Mb to graphics, went upto 40MB IIRC. When we first thought about HS X360 we started by allocating 128Mb for graphics (out of the original 256Mb).

Well my point was that it really varies on application, I don't believe in a rule of thumb with this. Speaking of - on 360 that means you planned on using 58% for graphics memory from the start (or was that before you were told how much kernel will occupy?).
On PS2 we've been using 60-70% of free memory for graphics, but I'm only counting main-mem allocation. Could add eDram into counting(since you said we should count framebuffers etc.) to skew the allocation further.

I personally start with an opening bid of 50/50 on a new platform and then see which part of the studio bribes me the most

Heheh, so your studio works on bribes? Got any openings?
Anyway, with the little we know of PS3 so far, I'm leaning towards thinking that exclusively using one memory pool for graphics may not be the most optimal approach.

Zork I, II & III for PS2 ?

I'm not sure even all 3 would eat that much. Especially since they are trivially compressable in a game like Zork.

 

[DeanoC]

]Deano,
so you think you'll end up with another 256mb of game and sound data? I find that a bit excessive.
For a PC game I've done ages ago (ignoring DX memory-backed surfaces), the usual case was approximately 64mb graphics, about 25 "game" and ~10 sound. And that was running a fair bit of simulation, but no physics.

Hope we don't use that much, but I don't want the last few months being a memory hunt Until were closer to release those figures will do.

Its a case of lets see what systems need what, we use lots of animation data (like 20Mb+ (tightly packed!) for the heroine), usually we take that out of the graphics pool but we may choose to use the other pool if that looks more reasonable.

 

[DeanoC]

Well my point was that it really varies on application, I don't believe in a rule of thumb with this. Speaking of - on 360 that means you planned on using 58% for graphics memory from the start (or was that before you were told how much kernel will occupy?).
On PS2 we've been using 60-70% of free memory for graphics, but I'm only counting main-mem allocation. Could add eDram into counting(since you said we should count framebuffers etc.) to skew the allocation further.

Well I always take OS kernel out of my non graphics pools, so yep we probably are closer to 60-70% of 'useable' RAM being graphics...

 

[Fafalada]

Well I always take OS kernel out of my non graphics pools, so yep we probably are closer to 60-70% of 'useable' RAM being graphics...

Yeah I do the same. Anyway, on to the the milion $ question, what's the Kernel on... you know?

 

[wazoo]

....I've once heard of an app that used 15MB of main memory for just strings...
...
on a PS2...
:?

Zork I, II & III for PS2 ?

Cheers
Gubbi

Max Payne

 

[marconelly!]

Well I always take OS kernel out of my non graphics pools, so yep we probably are closer to 60-70% of 'useable' RAM being graphics...

What?

You are using 200MB for graphics, and that's 60-70% of total usable RAM? Does that mean that the total usable RAM on PS3 is like 300-350MB, and the rest is OS?

 

[DeanoC]

Well I always take OS kernel out of my non graphics pools, so yep we probably are closer to 60-70% of 'useable' RAM being graphics...

What?

You are using 200MB for graphics, and that's 60-70% of total usable RAM? Does that mean that the total usable RAM on PS3 is like 300-350MB, and the rest is OS?

I'm not talking actual current use but future allocations.

If we say have 480Mb of useable RAM (BTW made up figure, just in case somebody jumps on it), I'd still allocate 256Mb for GFX which is closer to 60% of useable RAM than 50%.

 

[cthellis42]

....I've once heard of an app that used 15MB of main memory for just strings...
...
on a PS2...
:?

Zork I, II & III for PS2 ?

Nuh... Cat's Cradle: World Party.