Cell mass production plan for 2nd half of 2005

[London Geezer]

good boy. cookie?

 

[chaphack]

Watch the column that says 1.03-1.23 Trillions Operations Per Second

Ah, I see. Well, as it says, those are not even floating operations, programmable or fixed. Those are not even nvFLOPS. They probably count even the most stupid things like memory reads and such into these OPS.

Please scroll up. i just posted the PR numbers regarding Xbox.

Pardon me, but IIRC, in good ol' PS2 vs Xbox topics, MS/Nvidia got toasted because of that "fake" 1 trillion number to take the heat off EE 6.2(?)GFLOPS ratings.

Thus, i would assume the 1 trillion thingie IS the fabled NVFLOPS, no?

 

[cthellis42]

You are also going "yeehaw" over the exact opposite, and quite frankly harping on it a lot more. It's already been stated (and not contradicted by pretty much EVERYONE here) that we "don't know what it means" until we see it all in perspective. It could do 1 trillion "Sonyflops", or it could do a sum of 1 trillion "marketing flops," or it could do completely VALID flops that are limited elsewhere in the architecture, or hell, it COULD deliver far beyond what we cynics are contemplating.

Regardless, you bring this up quite often, perpetuate it in confrontational instead of conversational manner, and then act all wounded and pious, as if it's just your valiant self trying to fight off the horde of horrible Sony fanbois. How did performance get INTO a thread about fabbing and chip production and contemplating launch timings...? Well, it was a throwaway comment by jvd and not really responded to, but brought up in lively accusatory manner by guess who...? Oh, you're right! You! Also brought up tied to a comment about nVidia in a manner than you think anyone clarifying "nvflops" is rushing to Sony's rescue, as opposed to being able to comment on an architecture that EXISTS, is right in front of us, and CAN be put into context.

The reason no one buys your "Why get so tensed up...cant you leave some room for a kid who wants to understand more?" line is because you NEVER bring things up in a conversational or information-gathering manner. Instead the vast majority of your questions seem irrevocably tied to making a dig at something simultaneously.

So far the comments have been shallow and are easily ignored as background noise, such as jvd's "Not by the hair of my chinny-chin-chin shall PS3 hit 1 teraflops!" and Paul's "HAHA, you silly fools, it shall!" The only thing I am sure of is that no matter WHAT the PS3 delivers, there will be a lot of arguement about it.

Meanwhile, if you want to argue about nVidia performance numbers, remember that these CAN be seen in their own light. If you want to try some kind of "proof" to the PS3, keep in mind that we HAVE no full definition right now, and you can't possibly attach anything physically to it until we know more about the chip, so there's nothing substantial to keep arguing that route WITH.

One can make speculation without accusation, and in fact people do that a LOT around here. It evokes discussion and commentary and people introducing "loaded" statements do nothing but tear things down and open the door to let in more.

We're ALL used to "number claims" from companies right now, so what the hell is the point of arguing in life-or-death manner in regards to "IS!" or "IS NOT!" right now? We won't know the context of the 1TFLOP claim until Sony makes it clear, we won't know how to put it in perspective until we know the rest of the architecture, we won't know what that translates to in games and other applications until they are OUT and in front of us, and we probably won't even be able to see much of the "full picture" until a year or two AFTER launch, when developers have more time to bring out its capabilities. It's what we've seen before, what we see now, and is the ONLY think we can really "depend on happening" in the future.

 

[chaphack]

*yawn* cthelis, no offense but can we get back to this OPS FLOPS TOPS stuff? Anyone? Any hardware 3D guys care to share their knowledge?

 

[London Geezer]

it's just..... i can't.......... he...................... URRRGGGGGGHHHHHHHH....

 

[rabidrabbit]

awww that love-hate relationship I feel between those two is just so hearty.
Wonder if it is mutual, or is it gonna end up tragically unrequited

 

[Phil]

We're ALL used to "number claims" from companies right now, so what the hell is the point of arguing in life-or-death manner in regards to "IS!" or "IS NOT!" right now? We won't know the context of the 1TFLOP claim until Sony makes it clear, we won't know how to put it in perspective until we know the rest of the architecture, we won't know what that translates to in games and other applications until they are OUT and in front of us, and we probably won't even be able to see much of the "full picture" until a year or two AFTER launch, when developers have more time to bring out its capabilities. It's what we've seen before, what we see now, and is the ONLY think we can really "depend on happening" in the future.

Nice post cthellis42! Though we might not know what the context of a tere-flop per second is, if Sony stays with a PS2 similar approach in being through out programmable (which I think is pretty sure), it's going to be a lot of performance coppled with lots of freedom for many different approaches that may lead to a similar situation as found today when comparing Xbox and PS2 games. If launches are pretty much within the same timeframe by both, I would think that technically, the one with more freedom should pull away in complexity, though on average being on the same level or worse. Perhaps thanks to the raw power, developers may invest that given power in stuff like physics more than just graphics, which in turn may give worse looking screenshots, but still graphics that are simply on another level. Not better, just different. A trend that for many, is already so evident just by looking at this generation.

Thus, i would assume the 1 trillion thingie IS the fabled NVFLOPS, no?

I answered what NVFLOPS are back on page 5.

 

[London Geezer]

awww that love-hate relationship I feel between those two is just so hearty.
Wonder if it is mutual, or is it gonna end up tragically unrequited

well i'm sorry, i tried to explain, Cthellis tried to explain, Phil is still trying to explain... all we get is a "YAWN CAN WE GET BACK TO THE TOPIC OF HEYAHAHA NVFLOPSTHINGY PR NUMBERS TRILLIONS OF OPSTHINGY!!"
don't know what more we can do.

 

[ChryZ]

don't know what more we can do.

Ignore or filter the noise, that will raise the signal. Feeding the noise will degrade the signal ...

 

[London Geezer]

don't know what more we can do.

Ignore or filter the noise, that will raise the signal. Feeding the noise will degrade the signal ...

ok.... can i remind u, take less coffee... it's not good for your health....

 

[zidane1strife]

The only thing I am sure of is that no matter WHAT the PS3 delivers, there will be a lot of arguement about it.

Indeed.

 

[chaphack]

Phil,

not that i want to insist anything, but lets just say the 200GFLOPS IS NVFLOPS. Any ideas on how they arrived at teh number? Is there any reason for Nvidia to use "faked" FLOPS? I mean doesnt that "fools" the developers? Is it even "legal"? Whats the "comparable" FLOPS ratings? Do you know the "actual" FLOPS ratings of say, GF1 -> GFFX cards, so as we can try to establish a trend? Hell, is there even a "right" way to calculate 3D rendering FLOPS?

Any good place to read more in details about this NVFLOPS? Of course any recommended sites to know more about FLOPS in current PC 3D hardware?

 

[London Geezer]

not that i want to insist anything, but lets just say the 200GFLOPS IS NVFLOPS. Any ideas on how they arrived at teh number?

it was explained earlier on. they take into consideration all the hardwired effects. the Sony numbers are "programmable" Flops.

Is there any reason for Nvidia to use "faked" FLOPS?

yes. PR.

I mean doesnt that "fools" the developers?

no, developers, when developing a game, dont particularly care about Flops ratings. more about poly throughput, bandwidth, memory amount...

Is it even "legal"?

it's not "illegal" since, from a certain point of view, it is correct.

Whats the "comparable" FLOPS ratings?

how would we know?

Do you know the "actual" FLOPS ratings of say, GF1 -> GFFX cards, so as we can try to establish a trend?

u try.....

Hell, is there even a "right" way to calculate 3D rendering FLOPS?

yes.

Any good place to read more in details about this NVFLOPS? Of course any recommended sites to know more about FLOPS in current PC 3D hardware?

arstechnica, gamasutra, anandtech... beyond3d of course..

 

[chaphack]

hoho! I found an old topic here on FLOPS HOPS TOPS!

FLOPS is short for FLoating point Operations Per Second, and hence is used to measure the rate of computanional capacity.

Usually an 'op' is either an addition or a multiplication. Divisions are not counted as they are vastly more expensive to do fast, so much so that divisions are typically done by multiplying with the reciprocal ( a/b -> a* 1/b).

The numbers normally quoted in sales material is the theoretical maximum (the 'guaranteed not to exceed' number), which is often a poor measure of the true throughput of the chip/solution.

For nvidia to reach a number of 200GFLOPS they probably count every single floating point unit on the chip, assumne that each is active every single cycle and multiplies with the number of cycles per second. Looks good... But isn't very informative.

Take the R300 with 8 pixel shaders and 4 vertex shaders each of which can do 8 FP ops per cycle for a total of 31GFLOPS of shading power. One would assume that the NV3x would wipe the floor with R300 in FP. - it doesn't.

That's because alot of the FP power in the Nvidia number, goes into "hidden" stuff like triangle setup, FP<->int conversions, iterators etc. Whereas the 31 GFLOPS for the R300 is the raw shader power.

As for general purpose CPUs:

P4 can do 4 FP ops per cycle for +12GLOPS @ +3GHz, Athlon/Opteron can do a similar amount of ops per cycle but is clocked (alot) lower (But has fewer restrictions on issuing FP ops so probably has slightly higher throughput per cycle than the P4).

Gecko can do 2 FPMADDS (2FP adds and 2 FP muls) per cycle, so just under 2GFLOPS.

Cheers
Gubbi

As Gubbi pointed out FLOPs are somewhat of a generic measurement. To determine the flop rating of an IC, usually its fastest fp operation's throughput per second is measured. This makes flops one of the most meaningless performance metrics there is (even worse then OPs counts.). For an in-depth easy read on the various reasons (besides the obvious difference in arithmetic complexety between a fully IEEE-compliant 128 bit fp division and an arbitrarly rounded 16 bit FMAC, where the first one is regarded a single op while the second counts as 2...) for this i'd really recommend the beginning chapters of Computer Architecture by Hennessy & Patterson. Another good example is that when fp-coprocessor became add-in options for low-end desktop pcs, your flop rating actually shrinked when adding those while your performance went up, as emulating fp operations with int arithmetic usually yielded higher inst. throughput.

Damn! Guessed the forums moved too fast and i missed it!

 

[chaphack]

thanks to gubbi and pinky! so 1tflops, assuming it is done the "non"-hidden way is pretty impressive.


to the good guys, what does this sentence meant.....in a more basic understandable exampled terms?

That's because alot of the FP power in the Nvidia number, goes into "hidden" stuff like triangle setup, FP<->int conversions, iterators etc.

 

[chaphack]

That be the book to read? Hmmm....say for someone young, with minimal knowledge of computers, is it digestable/fun if yeay interested in 3d graphiX?

 

[Phil]

chaphack:

Is there any reason for Nvidia to use "faked" FLOPS? I mean doesnt that "fools" the developers? Is it even "legal"?

Good question. Ask Microsoft! Given that they hyped Xbox of being able to do 4 GPixels/sec on their specsheets handed over to retailers, I'd say they can.

Whats the "comparable" FLOPS ratings?

I think Vince or someone posted an ATi slide back a few pages (page 4?) with the accurate floatingpoint performance...

As to how comparable they are - they aren't, since the GPU in a Xbox 2 will likely be used for much different things than the processing unit of the PS3. Personally, I think PS3 games will be more physics driven (effects) compared to what software on the evolved PC architecture will be doing. No doubt, they'll be many debates over which is better or worse. It's all subjective if you ask me, though I addmitedly prefer the first approach.

 

[Phil]

That be the book to read? Hmmm....say for someone young, with minimal knowledge of computers, is it digestable/fun if yeay interested in 3d graphiX?

No idea, but just out of curiousity, how old are you anyway? If anything, I think programming would be a good start, as it could give you some perception as to what 'performance' is, the validity of memory and what potentially ultimate freedom can give you when developing something.

 

[jvd]

Chap there is a great book at barnes and nobles that goes through all the dx generations . What the specs mean , what the limitations are , what the hardware is capable of . Its 40$ and I had to buy it in the summer as part of one of my programing class's .

 

[chaphack]

So as what Gubbi and Pinky have said, FLOPS ratings are "quirky" to say the least? The better way to judge a cool performance is by the ram/bandwidth/latency/clockspeed/etc?

Not that i have anything hohohaha with that 1TFLOPS number in the first place, but with the commotion going on, i thought it was THE cool factor. Thus i wanted to see more about that.

No idea, but just out of curiousity, how old are you anyway? If anything, I think programming would be a good start, as it could give you some perception as to what 'performance' is, the validity of memory and what potentially ultimate freedom can give you when developing something

well, if anything, it is just for fun reading with 3D graphiX. Hell no am i going to write some codes or develop a game!
any easy yet cool 3D hardware books/sites to read up on?