If you consider leaking = announcing then have fun with it.
Is the PS3 on track to deliver full specs?
- Thread starter Rash'
- Start date
If you consider leaking = announcing then have fun with it.
To be fair to MS at was at the conference where that diagram was first shown outside of MS, and at that point MS was saying 3.0Ghz for the CPU's, possibly higher.
It was also leaked by a third party and not MS.
If your palnning on any sort of predictable scaling between clockspeed and actualy performance you're likely screwed anyway.
The smart thing to do when building a launch game is to build it on the hardware you have and don't bet on hardware bailing out your performance issues.
The file on my computer has a date of april 26, 2004 - more than 1.5 years before the console's release.
And it's interesting how many of those specs weren't properly understood by the online communities. The actual computational power of the GPU, the EDRAM bandwith... And of course the leak had no info on the architecture of the CPU. So, many haven't really believed it to be accurate at that time.
I think he actually sold his soul.
No, no, no! Only f@nbois sell their souls. Professionals rent theirs.
Angelcurio
Regular
On a side note, you guys remember that ****** that posted a lot of ps3 tidbits, one of them being Resistance being 20GB big, etc, etc, etc.
Well, you remember the PSP pack that was also part of that info, well check this link that was posted at neogaf:
http://www.buy.com/prod/Sony_PSP_Entertainment_Pack/q/loc/108/202945637.html
Well, you remember the PSP pack that was also part of that info, well check this link that was posted at neogaf:
http://www.buy.com/prod/Sony_PSP_Entertainment_Pack/q/loc/108/202945637.html
I still have my doubts about this guy , but he has been already correct about 3 things i believe .
Angelcurio
Regular
I was actually one of the first persons to call that Article BS, but today, as someone posted at gaf, he has been right in a couple of things:
- PSP bundle
- Resistance weapons
- Resistance audio
- Resistance disc size
- Virtua Tennis 3 textures
I am really starting to consider that article. The psp bundle really surprised me.
Mefisutoferesu
Regular
Ugh... at least here it's not going crazy like it did over there... guys just remember the best way to hide a lie is to wrap it up in some truth. More than likely the poster went to some presser and got some tidbits of info and then expanded that info with lies like KILLZONE AM REAL TOTAL!! It's easy for him to get a bunch of things right, but still be BSing from here to Kuala lumpur and back. BE CAUTIOUS.
I know it's fun to try and eek info from devs, and they're happy to play the game, but may I remind people that trying to get such info can land them in serious trouble. Basically you're asking them to put their jobs and potentially livelihoods on the line to feed your curiosity. It only takes a slip of a remark for a dev to have said something they shouldn't for them to break NDA.so your ARE basicly CONFIRMING the gpu speed is dropped from the initial 550mhz figure...
eg. If Barbarian had said something he shouldn't that confirmed what was only rumour, he's in breach of NDA. And if someone hears (news travels quickly through the internet) and can pin his identity, he's scuppered.
I think it's only fair to let devs say what they feel comfortable saying, rather than pester them for more info. They're not just keeping secrets because they get cheap thrills seeing the rest of us suffer from interminable curiosity and like to watch us squirm (although there's nothing they like better than sharing stories about the torments they have inflicted with their enigmatic comments), but protecting their jobs. Asking them for info is like asking anyone to break their contract; asking a doctor to break medical confidentiality, asking an architect to show you the plans for a proposition, etc. They have nothing to gain and everything to lose, and the game of 'Baiting the Devs' shouldn't be pushed too hard.
Yeah..PS3 is a quantum computer, now the world is ready to finally know the truth!
Kidding, using an half bit is quite simple once one starts to reinterpet bits using floating point math.
Let say you want to encode real numbers in the interval [0,n) using m bits, where n is a real number < 2^m, and m is a natural number: you can simply divide your real number by n, convert it to a m bit integer fixed point representation using zero bits for the integer part and m for the fractional part (what about m = 8 using our dear RGBA8 frame buffers?
) and that's it, you used log2 bits to encode the integer part and log2((2^m)/n) bits to encode the fractional part.[edit] before someone starts to complain I know you can't really represent a real number using a floating point representation, it's just a crappy approximation!
Marco
Last edited:
Yeah..PS3 is a quantum computer, now the world is ready to finally know the truth!
Kidding, using an half bit is quite simple once one starts to reinterpet bits using floating point math.
Let say you want to encode real numbers in the interval [0,n) using m bits, where n is a real number < 2^m, and m is a natural number: you can simply divide your real number by n, convert it to a m bit integer fixed point representation using zero bits for the integer part and m for the fractional part (what about m = 8 using our dear RGBA8 frame buffers?
[edit] before someone starts to complain I know you can't really represent a real number using a floating point representation, it's just a crappy approximation!
two remarks : )
1. you really need a variable exponet to have the floating aspect. the above scheme is normalised yet fixed.
2. the crappy approximation part of the computerized reals sorta ruins the numerical significance of the 'fine-grained' bitness measures you gave. consider the following simple example:
n = 5, m = 3
x = 3.5
x / n = 0.7, encoded in ours scheme = ~.101b, or = ~.110b (depending on how we want our approximation).
.101b * 5 = 3.125
.110b * 5 = 3.75
let's consider the second variant (.110b) - how many bits were used for the integer and how many for the fractional part?
according to your estimate, it should be log2
= ~2.322 bits for the integer and log2((2^m) / n) = 0.678 bits for the fractional parts. which i'm not convinced is the actual case ; )Right, but when I was talking about floating point math I was not referring to this scheme, I was referring to the fact that I used floating point math to encode/decode it (it runs on a pixel shader)
That measure is just a theoretical one, does not obviously take in account quantization errors, I did not even care about the truncation/rounding mode
(and obviously your completely valid case is a kind of very bad case since n and 2^m are very close.. )The point was to reasonably show that it makes sense to use fractional bits to represent quantities, there's nothing exotic about it.
ah. oks then.
That measure is just a theoretical one, does not obviously take in account quantization errors, I did not even care about the truncation/rounding mode
The point was to reasonably show that it makes sense to use fractional bits to represent quantities, there's nothing exotic about it.
well, as much as i suspect you're theoretically right about it, i've always had issues with putting such concepts into practice. like the theoretical knowledge that the highest-informational efficiency numeral system would be of base .. e.
Well..believe me or not, when I was writing the post I was just thinking about that paper, lolwell, as much as i suspect you're theoretically right about it, i've always had issues with putting such concepts into practice. like the theoretical knowledge that the highest-informational efficiency numeral system would be of base .. e.
OK, we got some indications of a change in the operation frequencies of the RSX and the GDDR3 memory. However, there is no official confirmation about it, so I leave it at that.
Let us look at history, the official clock frequency of PSP games was downclocked from previously announced clock speeds quite close to the launch of the PSP in order to save the batteries. There is however no hardware limitations that prevent Sony from at some time in the future rise this frequency to free more power to the developers, that will of course shorten the the battery life, but maybe they could introduce better batteries for old console while new PSP models may be more energy efficient. I think this is plausable scenario because it would help Sony extend the life of the PSP platform by allowing developers to keep improving the games over a longer time than if the max frequency was available from day one. Keeping long life cycles for console platforms is good economy for both the hw manufacturer and the game developers.
What about if Sony applies a similar tactic for the PS3 they downclock the RSX at launch to keep thermal level low and thereby also keeping the noise level lower. Maybe the RSX is capable of running at 600 MHz with whith higher fan speeds already today, but Sony choose to apply the PSP scenario I described above. That is, restrict the developers from taking full advantage of the hardware at launch, but 3-4 years down the road you give them access to an extra over-drive that has been waiting there right from the start, just to give an extra boost when the improvements in the looks of the games are starting to level out.
Could this be a possible scenario? Just some food for thought.
Let us look at history, the official clock frequency of PSP games was downclocked from previously announced clock speeds quite close to the launch of the PSP in order to save the batteries. There is however no hardware limitations that prevent Sony from at some time in the future rise this frequency to free more power to the developers, that will of course shorten the the battery life, but maybe they could introduce better batteries for old console while new PSP models may be more energy efficient. I think this is plausable scenario because it would help Sony extend the life of the PSP platform by allowing developers to keep improving the games over a longer time than if the max frequency was available from day one. Keeping long life cycles for console platforms is good economy for both the hw manufacturer and the game developers.
What about if Sony applies a similar tactic for the PS3 they downclock the RSX at launch to keep thermal level low and thereby also keeping the noise level lower. Maybe the RSX is capable of running at 600 MHz with whith higher fan speeds already today, but Sony choose to apply the PSP scenario I described above. That is, restrict the developers from taking full advantage of the hardware at launch, but 3-4 years down the road you give them access to an extra over-drive that has been waiting there right from the start, just to give an extra boost when the improvements in the looks of the games are starting to level out.
Could this be a possible scenario? Just some food for thought.
No, because it would prevent existing users using the boost. The PSP's 'overdrive' mode can be applied to first-gen PSPs, just it would gobble the battery quickly. A PS3 overdrive would need a better cooling method. Or do you mean in 3-4 years time you can run a game on your first gen PS3 and it will up the fan speed, giving the better performance but with more noise?
