Carl Lewis also took illegal stimulants, and should per the rules have been banned from competing on at least one occasion (perhaps he had a better marketing/legal team than even intelOriginally Posted by Drwho?
), so your analogy could have been more wisely chosen. Perhaps something involving cars would work better? Oh, wait...
Obsolescence be thy name
here is a better idea of what larrabee is/is not.I don't know the tech specs of it but it still strikes me as a rather inefficient thing to take a bunch of x86 CPUs and have them do graphics. I mean, that's why the graphics card was created, because it was designed specifically for graphics and did a much better job than a CPU
nvidia has 240SP,AMD 160SP's, and intel 32.
I take vitamine C and Cafeine every morning ... I am guilty ... lolCarl Lewis also took illegal stimulants, and should per the rules have been banned from competing on at least one occasion (perhaps he had a better marketing/legal team than even intel
DrWho, The last of the time lords, setting up the Clock.
you are on the right track ... at the end, when all DirectX 11 stuffs are adopted, it will end up to a race to instruction per clock ... because you will start having "special" processing for each pixels , and when doing so, you ll have to start reusing all the tricks of the CPU ... load units that don't have issue with alignment, fast branching ... SIMD and all the usual gadgets.here is a better idea of what larrabee is/is not.
nvidia has 240SP,AMD 160SP's, and intel 32.
by 2015 to 2020, you ll be back to CPU cores, that is my prediction.
(It is ok to dissagree, don't have to beat me up verbally :-P)
DrWho, The last of the time lords, setting up the Clock.
I thought that was the same thing....I take vitamine C and Cafeine every morning ... I am guilty ... lolMe sipping my cup of coffee on the computer in the morning and my wife would glare at me and ask..."What the hell do you think your doing?" And I would tell her I am taking my vitamins!
Last edited by Sadasius; 09-24-2009 at 04:15 PM.
I wouldn't talk that way, if you (Your Company) weren't so arrogant in the last years with Larrabee. Your Marketing was more in the way, "We'll nerver made a gpu, but what's the problem. We make one now, revolutionize the whole market and crush everything else without a problem. It's just a gpu".With all due respect, I don't know if you ever tried to work on a puzzle that have more than 1 billion parts ... this is what GPU are ... so, it is always easy from your screen point of view to write those kind of statements, never the less, I am sure you never even completed a Lego with 100 000 parts ;-)
so, send us the picture of a 100 000 parts lego before you are "that" agressive.
so, put your ego back in your pocket, and show some respect for the people who are actually trying to build one of the most complexe machine ever build.
The point of the demo was to show that it is alive, I did not want to performance of it to be visible, neither the Big boss, so, that was perfect to allow no comparaison yet. This is like a young kid, it just standed up, don't ask it to run the fastest 100 meter yet. Carl Lewis crawled before he walk, and he walk before he smoke you at the 100 meter.
Thanks for understand that this kind of archivement don't get done over night.
Francois
PS: I only used your ton to make you feel how we feel when you write this kind of thing.
And now it's suddenly not that easy. I respect the people working for it, but i don't respect the attitude of your company.
PS: Never made such a big lego, but if i would, you still couldn't see it. My Legoparts are much smaller than yours
I agree with you, some people talk high of their projects in high words, but it is legitimate, just think about it:I wouldn't talk that way, if you (Your Company) weren't so arrogant in the last years with Larrabee. Your Marketing was more in the way, "We'll nerver made a gpu, but what's the problem. We make one now, revolutionize the whole market and crush everything else without a problem. It's just a gpu".
And now it's suddenly not that easy. I respect the people working for it, but i don't respect the attitude of your company.
PS: Never made such a big lego, but if i would, you still couldn't see it. My Legoparts are much smaller than yours
This is their dream come true, they are like me and you, Geeks that enjoy technology. if they would not talk high about their project, they would not be putting much energy into it, they are very enthousiat about it, they and I believe into it, we spent a lot of energy on it, and we will not stop before we get the best out of it. This is intel at its best, this is what we are good at ... take a problem, analyst it, and bang your head on the wall as long as it is not the best ... Conroe is born like this, Pentium is born like this, Atom is born like this.
They forgot that before running, you got to be walking ... we just showed crawling ... we showed prototypes, it was running ... now, they are putting even more energy to make it awesome, they are motivated people, the best kinds, young and smart.
Give them a chance, and don't take their enthousiams as an issue, but as a strenth ... they will surprise you, and at the tuning, I got my big surprises coming too ... so, just relax and wait ...
we just wanted to show that the baby is born ... that's it.
we are enthousiat about it, not arrogant. make sense ?
Francois
DrWho, The last of the time lords, setting up the Clock.
...but from time to time you can give birth to the Itanium
Adobe is working on Flash Player support for 64-bit platforms as part of our ongoing commitment to the cross-platform compatibility of Flash Player. We expect to provide native support for 64-bit platforms in an upcoming release of Flash Player following the release of Flash Player 10.1.
I am pretty anxious to see what it can do myself. I can understand the many frustrations from both sides of the fence. People have heard about it for so long with nothing to hold on to and Intel very enthusiastic about their work but cannot run with the big boys because they are crawling. Did I get that right? Well anyway I understand what you mean and I am hoping that sometime in the near future we have a little more to hold on to and talk about.
Just look at our track record for the last 4 years ... and you ll get the taste of the sauce we are preparing.
Francois
(Sorry, can't say more)
DrWho, The last of the time lords, setting up the Clock.
Sounds like a gay :banana::banana::banana::banana: teaser line....
That's because you only have a demo running at 0.004FPS.
I am excited for the future of Intel graphics, though.
For my part I know nothing with any certainty, but the sight of the stars makes me dream.
.
.
OMGSounds like a gay :banana::banana::banana::banana: teaser line....
DrWho, The last of the time lords, setting up the Clock.
Francois, would you make some comments on this, please?
What does that mean exactly? The LRB prototype showed in the demo isn't the most powerful version Intel currently have, or it will get better in the future?
Larrabee rendered that image above using raytracing, it's not running anywhere near full performance.
.
I imagine it means that the hardware-software relationship is very one-dimensional at this stage, and, in turn, the hardware is being held back from its true potential. I definitely could be wrong, though.
For my part I know nothing with any certainty, but the sight of the stars makes me dream.
that picture is very misleading...here is a better idea of what larrabee is/is not.
nvidia has 240SP,AMD 160SP's, and intel 32.
here, this is much better:
Larrabee
GT200
RV770
somebody correct me if im wrong, but this is how i understood it from the siggraph papers:
LRB = 32+ "16way" processors, 2flops/clock
RV770 = 10 "16way" processors, 10flops/clock
GT200 = 30 "8way" processors, 3flops/clock
RV870 = 20 "16 way" processors, 20flops/clock
GT300 = 60(?) "8way"(?) processors, 6flops/clock
EDIT: corrected the infos
http://graphics.stanford.edu/~kayvon...g/diagrams.pdf
Last edited by saaya; 09-25-2009 at 08:41 AM.
You're mixing different things. 4way for LRB means 4 threads per core. It has 512-bit vector FPU, so each core can execute 16 32-bit ops per cycle. Considering FMAC capability it has in total 32 FLOP/cycle.
it is simple, we don't want to give away the performance of the part too early. we still have few issues, and fix are found.
When you play chess, you don't want to give your next 5 move to your opponent ... it is all about this.
Francois
DrWho, The last of the time lords, setting up the Clock.
There are many more parameters to concidere, like the speed of branching, the speed of loading aligned and unaligned ... you are back to the instruction per clock race ... hehehehe ...that picture is very misleading...
here, this is much better:
Larrabee
GT200
RV770
somebody correct me if im wrong, but this is how i understood it from the siggraph papers:
LRB = 32+ "16way" processors
RV770 = 16 blocks of 10 "5way" processors = 160 "5way" processors
GT200 = 8 blocks of 3 8way processors = 24 "8way" processors
RV870 = 32 blocks of 10 "5 way" processors = 320 "5way" processors
GT300 = 16(?) blocks of 3(?) 8way(?) processors = 48 "8way" processors
so if you just look at processors, afaik lrb is 32+ (probably 40-48), gt300 probably 40-48 too and rv870 is a whooping 320. lrb and gt300 have fast and very beefy processors while rv870 has rather simple processors, clocked at only half of lrb and gt300, but makes up for it with the sheer number of them.
DX11 and other OpenCL will open the door to this again, IPC is the key ... actually IPC/watts.
check out here: http://www.realworldtech.com/page.cf...WT090909050230
you ll figure out that Silverthome (Atom) is as efficent as the RV700 , and it is without the 512 bits execution units of Lrb , and withoout its texture sampler...
Engineers understood where it is going, I wish that you guys put your fanboys hate in your pocket, and look at the technology itself.
Being able to compensiate for the overhead of x86 was the challenge, this graph shows that it is done.
now, the next challenge is to make Many of those cores x86 to work together ...
Then , it comes down to who is able to make the large transistor dices, and we all know who has the best fabs on the planete.
May the Core be with you!
Francois
Last edited by Drwho?; 09-25-2009 at 08:46 AM.
DrWho, The last of the time lords, setting up the Clock.
ha! i found the same mistake and edited my post at 12:30 while you posted at 12:31 ^^ hehehe
thanks for the headsup tho!
i made a bunch of mistakes cause i only watched the pics and didnt read the actual presentation of the guy who made them
i think i got it right now...
so basically:
if intel wants to beat rv870 in Gflops they will need
32 cores at more than 2.70Ghz ()
or
40 cores at more than 2.10Ghz
or
48 cores at more than 1.77Ghz
2.7ghz is out of question i think... especially within a 250W tdp envelope...
so intel needs at least 48 cores to beat them in gflops, something that they seem to put a lot of effort in...
gt300 will be out by the time they launch though, so they will probably need 48 cores or even more to beat THAT... probably more cause nvidia is also focussing a lot on gflops and will probably improve that in gt300 over gt200, so itll probably have more flops per cycle...
i really dont think intel can beat ati and nvidia in gflops... not in 45nm... im pretty sure they need 32nm not only for power but also for die size...
but overall... i dont get what the hype about gflops is?
thats not a very good benchmark to compare performance...
francois, but why is it comparing die size to dp flops?
that makes no sense... shouldnt it be transistors or die size/mfc node or something like that?
interesting graph though... thanks
Last edited by saaya; 09-25-2009 at 08:50 AM.
I wish it was that simple ...ha! i found the same mistake and edited my post at 12:30 while you posted at 12:31 ^^ hehehe
thanks for the headsup tho!
i made a bunch of mistakes cause i only watched the pics and didnt read the actual presentation of the guy who made them
i think i got it right now...
so basically:
if intel wants to beat rv870 in Gflops they will need
32 cores at more than 2.70Ghz (
or
40 cores at more than 2.10Ghz
or
48 cores at more than 1.77Ghz
2.7ghz is out of question i think... especially within a 250W tdp envelope...
so intel needs at least 48 cores to beat them in gflops, something that they seem to put a lot of effort in...
gt300 will be out by the time they launch though, so they will probably need 48 cores or even more to beat THAT... probably more cause nvidia is also focussing a lot on gflops and will probably improve that in gt300 over gt200, so itll probably have more flops per cycle...
i really dont think intel can beat ati and nvidia in gflops... not in 45nm... im pretty sure they need 32nm not only for power but also for die size...
francois, but why is it comparing die size to dp flops?
that makes no sense... shouldnt it be transistors or die size/mfc node or something like that?
interesting graph though... thanks
you concidere each efficency of NV, ATI and Lrb cores equal ... that is a mistake, the speed of branching of a Lrb core is sooo much faster than what a stream engine can do, the speed of loading of unligned data is infinitly faster than a stream engine, because it only can do it ... so, every data in 20th century GPU is aligned, increasing their memory foot print by adding "Zeros" bobles into the memory blocks ...
The problem is fortunatly not as simple as you simplistic maths.
I am not even talking about the inter-connect speed, you probably understood by now that Intel has a "very good" memory controler ... fast interconnect (look at the xeon scaling on spec_fp_rate).
we need to tune to thing, at compute, it is already really nice, now the challenge is to make it a good GPU ...
We have a lot of new talented people, let s see ...
For competitive reason, I can't give away a lot of information, I am very sorry for this, but definitively, your maths does not work.
Francois
DrWho, The last of the time lords, setting up the Clock.
based on that chart, the ATOM may be as powerful as 870 for flops/watts, but takes 6-7x the chip size to do so? that turns into costs and perf/$ will get hit hard.
OK now I'm struggling to wrap my head around all this
The Cardboard Master Crunch with us, the XS WCG team
Intel Core i7 2600k @ 4.5GHz, 16GB DDR3-1600, Radeon 7950 @ 1000/1250, Win 10 Pro x64
Sounds like alot of marketing speak to me. :P: And no, just because Intel has executed well over the last 3 years, doesn`t earn them the right to benefit of the doubt AFAIC. There`s no such thing. And they haven`t executed to their roadmap perfectly for that matter either. Tick tock has slipped some, and LRB has had huge delays. If you want to go down that road though, let`s see how this thing competes against Fusion, as it will probably surface within a reasonbly close timeframe.
You sound hurt and angry.
Posting Permissions
Reply With Quote
Bookmarks