nightsurge said: Ascended_Saiyan3 said: It's more thinking/planning for GREAT results!  Devs HAVE to learn how to parallelize their code and tasks to go forward on PC and PS3.  That type of talk lends itself to the public perception of lazy devs.  That's what developers get paid for.  It's...their job!  Occasionally, your job throws you a curve ball.  At that point, your choice is to adapt or get out.  This applies to everyone.  Why should devs be excluded from this sort of "natural selection" process?  Code parallelization is the devs' curve ball.  Now, they're faced with their choice.  It will never get easier if devs don't suck it up and learn it.  The tools are here to make it easier on devs and more tools are being developed.  It's been explained at several conferences that the knowledge gained from working on the Cell travels over to the PC.  Then, devs can finally make a game that properly uses the Intel Core i7 965 XE's 8 cores (hell, even the 4 core version) to bring new forms of gameplay, A.I., physics, and graphics. All core i7s currently only have 4 cores.  Seriously what are you smoking?  You keep coming up with randomly wrong stuff.  They have 8 threads, but only 4 cores.

I was just following Intel's roadmap.  I assumed that the $1,000 version would be 8 cores.  However, you are correct.  Intel's Core i7 965 XE is probably not near equal to Cell.  This would seem to put the Cell at a decent amount of performance above Intel's current implementation.  The stuff I am wrong about, I admit.  So far, that's 2 things.  This one (I was ahead of the curve of) and the one below the I will admit to.  That's not randomly wrong stuff. To determine a pattern or lack thereof, you need more that 2 examples.

 

Rpruett said: nightsurge said: Ascended_Saiyan3 said: nightsurge said: @Ascended I already read the whole deal about the Pub Fund and everything. Bandwidth costs are not covered. Sony is not a "safer" investment since you have to pay fees for every time someone actually buys your game. Xbox Live, if your game is successful and a good one like Burn Zombie Burn, will be much better for you overall when it comes to profits. Oh please!  It's an advertising fee of $0.16 per Gb.  Plus, that's only for the first 60 days. Plus, it's ONLY for demos, hence being labeled an ad fee.  Right now, Sony gets charged for publishers' demo content.  This charge is for publishers and NOT developers.  You have been poisoned.   Uh noooo.  It is "bandwidth fees" and applies to all downloads, all the time.  You have been fooled if you really think it is some "ad fee".   You're both wrong. It is a bandwidth fee for all content on the PSN.  It only applies for 60 days though on non-paid content (Such as Demos) but it is a permanent fee for paid content (Like expansion packs or extra costumes or whatever).    It's a way to keep file sizes down on the network.  Microsoft limits file sizes (Which has been a big annoyance for developers as well).  The PSN gives them the flexibility of doing essentially whatever they want (Size wise) (But now,  they are going to pay for it).  Really, it's almost a non-issue entirely that has yet to show adverse effects.    For most large publishers the fees are negligible.    The concern is for the smaller sized ones.   Which Sony gave back to them by matching their budgets.

You are correct.  I read that some time ago and mis-remembered.  Thanks.  I just found the articles.  Of course, they haven't been confirmed, but I'm inclined to believe it.  We all know Sony is doing what they need to keep PSN free to PS3 owners.  Personally, I think that's the way it should be.

 

Ascended_Saiyan3 said: nightsurge said: Ascended_Saiyan3 said: It's more thinking/planning for GREAT results!  Devs HAVE to learn how to parallelize their code and tasks to go forward on PC and PS3.  That type of talk lends itself to the public perception of lazy devs.  That's what developers get paid for.  It's...their job!  Occasionally, your job throws you a curve ball.  At that point, your choice is to adapt or get out.  This applies to everyone.  Why should devs be excluded from this sort of "natural selection" process?  Code parallelization is the devs' curve ball.  Now, they're faced with their choice.  It will never get easier if devs don't suck it up and learn it.  The tools are here to make it easier on devs and more tools are being developed.  It's been explained at several conferences that the knowledge gained from working on the Cell travels over to the PC.  Then, devs can finally make a game that properly uses the Intel Core i7 965 XE's 8 cores (hell, even the 4 core version) to bring new forms of gameplay, A.I., physics, and graphics. All core i7s currently only have 4 cores.  Seriously what are you smoking?  You keep coming up with randomly wrong stuff.  They have 8 threads, but only 4 cores. I was just following Intel's roadmap.  I assumed that the $1,000 version would be 8 cores.  However, you are correct.  Intel's Core i7 965 XE is probably not near equal to Cell.  This would seem to put the Cell at a decent amount of performance above Intel's current implementation.  The stuff I am wrong about, I admit.  So far, that's 2 things.  This one (I was ahead of the curve of) and the one below the I will admit to.  That's not randomly wrong stuff. To determine a pattern or lack thereof, you need more that 2 examples.

I agree in all counts except in saying that the Core i7 is not near equal to the Cell.  They are completely different architectures designed to be good at completely different things.  The Core i7 would demolish the Cell when it comes to running operating systems and the various software involved in that.  The Cell would demolish in cluster computing and complex simulations.  Gaming... I would have to say that Core i7 is probably better simply because developers know it and can get it done easier, but I'm sure performance and theoreticals they are about the same as far as gaming.

 

So after 19 pages of these e-expert analysis, what have we learned? Is it true or false that the one person in question can handle one side better than the other? Who wins? And what will the result generate?

nightsurge said: Ascended_Saiyan3 said: nightsurge said: Ascended_Saiyan3 said: It's more thinking/planning for GREAT results!  Devs HAVE to learn how to parallelize their code and tasks to go forward on PC and PS3.  That type of talk lends itself to the public perception of lazy devs.  That's what developers get paid for.  It's...their job!  Occasionally, your job throws you a curve ball.  At that point, your choice is to adapt or get out.  This applies to everyone.  Why should devs be excluded from this sort of "natural selection" process?  Code parallelization is the devs' curve ball.  Now, they're faced with their choice.  It will never get easier if devs don't suck it up and learn it.  The tools are here to make it easier on devs and more tools are being developed.  It's been explained at several conferences that the knowledge gained from working on the Cell travels over to the PC.  Then, devs can finally make a game that properly uses the Intel Core i7 965 XE's 8 cores (hell, even the 4 core version) to bring new forms of gameplay, A.I., physics, and graphics. All core i7s currently only have 4 cores.  Seriously what are you smoking?  You keep coming up with randomly wrong stuff.  They have 8 threads, but only 4 cores. I was just following Intel's roadmap.  I assumed that the $1,000 version would be 8 cores.  However, you are correct.  Intel's Core i7 965 XE is probably not near equal to Cell.  This would seem to put the Cell at a decent amount of performance above Intel's current implementation.  The stuff I am wrong about, I admit.  So far, that's 2 things.  This one (I was ahead of the curve of) and the one below the I will admit to.  That's not randomly wrong stuff. To determine a pattern or lack thereof, you need more that 2 examples.   I agree in all counts except in saying that the Core i7 is not near equal to the Cell.  They are completely different architectures designed to be good at completely different things.  The Core i7 would demolish the Cell when it comes to running operating systems and the various software involved in that.  The Cell would demolish in cluster computing and complex simulations.  Gaming... I would have to say that Core i7 is probably better simply because developers know it and can get it done easier, but I'm sure performance and theoreticals they are about the same as far as gaming.

It's already been shown, though many tests, that 1 SPU can equal or surpass Intel's top single core processor in many tasks. The inter-core bandwidth has been proven at 197GB/s. The link from my percious post shows Intel Core i7 965XE's TOTAL bandwidth (cache, memory, and inter-core bandwidth) is 106GB/s. Plus, it only has 4 cores at 2 ops per clock. Also, I can't believe you talked about running code/apps not written for it's architecture! That's a bit like saying 2.4GHz PowerPPC processor would demolish an Intel Core i7 965XE at Linux written specifically for PPCs. LOL. It's just that the code is written to take advantage of different architectures.

"most ps3 exclusives will continue to suck"

This guy has been living under a rock ! He's crazy !

Ascended_Saiyan3 said: It's already been shown, though many tests, that 1 SPU can equal or surpass Intel's top single core processor in many tasks. The inter-core bandwidth has been proven at 197GB/s. The link from my percious post shows Intel Core i7 965XE's TOTAL bandwidth (cache, memory, and inter-core bandwidth) is 106GB/s. Plus, it only has 4 cores at 2 ops per clock. Also, I can't believe you talked about running code/apps not written for it's architecture! That's a bit like saying 2.4GHz PowerPPC processor would demolish an Intel Core i7 965XE at Linux written specifically for PPCs. LOL. It's just that the code is written to take advantage of different architectures.

I would like to see those tests. If you get me ps3, I can also make program for it which runs fine on Xbox(original) and really doesn't run well on ps3 even if you find any expert to optimate that program in anyway but still making it function like original(Of course).

 

Btw, parallizing is very easy on homogeneous platform.

@ Deneidez

How?

Using 2 SPUs, they should well be able to perform more than twice as well (or many times for many) as what is running on one Xenon core with optimised code (for both architectures) regarding just about any function.



According to the experts it's not that different nor hard to do on the Cell.

everybody look!

its the obvious!

Deneidez said: Ascended_Saiyan3 said: It's already been shown, though many tests, that 1 SPU can equal or surpass Intel's top single core processor in many tasks. The inter-core bandwidth has been proven at 197GB/s. The link from my percious post shows Intel Core i7 965XE's TOTAL bandwidth (cache, memory, and inter-core bandwidth) is 106GB/s. Plus, it only has 4 cores at 2 ops per clock. Also, I can't believe you talked about running code/apps not written for it's architecture! That's a bit like saying 2.4GHz PowerPPC processor would demolish an Intel Core i7 965XE at Linux written specifically for PPCs. LOL. It's just that the code is written to take advantage of different architectures. I would like to see those tests. If you get me ps3, I can also make program for it which runs fine on Xbox(original) and really doesn't run well on ps3 even if you find any expert to optimate that program in anyway but still making it function like original(Of course).   Btw, parallizing is very easy on homogeneous platform.

http://www.mc.com/uploadedFiles/Cell-Perf-Simple.pdf

Against the fastest single cores (like Intel Xeon)...1 SPE can perform up to 7 times better than a single core 3.6GHz Intel Xeon.

http://www.research.ibm.com/cell/whitepapers/alias_cloth.pdf

Cloth Physics on 2.4GHz Cell vs single core 3.6GHz Intel Xeon

http://www.simbiosys.ca/science/white_papers/eHiTS_on_the_Cell.pdf

Performance comparison between Intel/AMD dual-core processors versus Cell...also partially explains that realism iin games is not limited by the GPU anymore.

http://www.cs.berkeley.edu/~samw/research/papers/ipdps08.pdf

Cell beats Intel Quad-Core (Clovertown) at DP GFLOPS, which it's terrible at compared to it's SP (just divide the chart number in half for Cell due to Cell blade being used).

http://www.ibm.com/developerworks/power/library/pa-cellperf/

IBM document on Cell and per SPE performance

http://www.tomshardware.com/news/ibm-lead-architect-cell-cpu-ps3-gaming,1336.html

SPEs capable of "running single core scalar programs in their entirety.

http://lzhan.wikispaces.com/Cell+Programming?f=print

Individual programmer experiments on PS3...shows per SPE output at bottom.

http://arxiv.org/PS_cache/physics/pdf/0611/0611201v2.pdf

Cell almost 20x (2 orders of magnitude) faster than Opteron processor at Molecule Dynamics

http://www.sintef.no/upload/IKT/9011/SimOslo/eVITA/2008/hagen2.pdf

Cell versus Intel Core2 Duo in power consumption and theoretical performance.

http://www.power.org/resources/devcorner/cellcorner/hpcspe.pdf

Cell versus Intel Quad Core (Clovertown) theoretical GFLOPS...and real world Multigrid Finite Element solver running at an unprecedented 52GFLOPS sustained performance.

http://www.power.org/devcon/07/Session_Downloads/PADC07_Bergmann_Sourcery_VSIPL.pdf

Cell almost 14x the sustained performance of single core 3.6GHz Intel Xeon at VSIPL++ fast convolution.

http://www.cis.udel.edu/~cavazos/cisc879/papers/cellFMwhitepaper.pdf

Cell (1GB RAM) 11x faster than Intel Quad-Core (Clovertown) w/16GB RAM at SP for financial market applications.

http://74.125.45.104/search?q=cache:lLEILf-tpPAJ:gametomorrow.com/blog/index.php/2005/07/26/beyond-polygons/+SPE+GFLOPS&hl=en&ct=clnk&cd=30&gl=us&client=opera

Cell ray-casts complex 720p scenes at greater than 30fps.

http://64.233.169.132/search?q=cache:XcMos60ZImsJ:gametomorrow.com/blog/index.php/2005/11/30/+gametomorrow+gpus-vs-cell&hl=en&ct=clnk&cd=6&gl=us&client=firefox-a

Cell beats Nvidia 7800 GT OC at ray-tracing quaternion Julia fractals by a factor of over 5x.

http://64.233.169.132/search?q=cache:F7MZrviYvQwJ:gametomorrow.com/blog/index.php/%3Fp%3D187+gametomorrow+Cell+vs+g80&hl=en&ct=clnk&cd=1&gl=us&client=firefox-a

Cell over 4x faster at ray-tracing Stanford Bunny than Nvidia 8800GTX.

 

The last 3 links were to GameTomorrow.  That site is down, but a search will reveal that information on other site.