Re: [agi] NVIDIA GPU's
BTW, the CUDA toolkit for programming the GPU's is developing rapidly (and is still in beta). here are memory bandwidths actually measured on my machine: CUDA version 0.8: Host to Device Bandwidth for Pinned memory Transfer Size (Bytes) Bandwidth(MB/s) 33554432 1647.6 Device to Host Bandwidth for Pinned memory Transfer Size (Bytes) Bandwidth(MB/s) 33554432 1654.7 Device to Device Bandwidth Transfer Size (Bytes) Bandwidth(MB/s) 33554432 3332.1 CUDA version 0.9: Host to Device Bandwidth for Pinned memory Transfer Size (Bytes) Bandwidth(MB/s) 33554432 2700.0 Device to Host Bandwidth for Pinned memory Transfer Size (Bytes) Bandwidth(MB/s) 33554432 2693.3 Device to Device Bandwidth Transfer Size (Bytes) Bandwidth(MB/s) 33554432 53768.0 In other words, once uploaded to the GPU, you can afford to reorder your data any way you want as often as you need to to take advantage of parallel ops. Or to put it another way, the GPU has about the same bandwidth to its memory as your brain does to its on a per-byte basis, in my ballpark estimates. Somewhere on the order of 100 GPUs is a brain-equivalent. We're getting damn close... Josh - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
Re: [agi] NVIDIA GPU's
That's 53.8 GB/s for a load of 33.6 MB? Is there a burst cache effect going on here or do you think that's sustainable for multiple seconds? Bo On Fri, 22 Jun 2007, J Storrs Hall, PhD wrote: ) BTW, the CUDA toolkit for programming the GPU's is developing rapidly (and is ) still in beta). here are memory bandwidths actually measured on my machine: ) ) CUDA version 0.8: ) ) Host to Device Bandwidth for Pinned memory ) Transfer Size (Bytes) Bandwidth(MB/s) ) 33554432 1647.6 ) ) Device to Host Bandwidth for Pinned memory ) Transfer Size (Bytes) Bandwidth(MB/s) ) 33554432 1654.7 ) ) Device to Device Bandwidth ) Transfer Size (Bytes) Bandwidth(MB/s) ) 33554432 3332.1 ) ) CUDA version 0.9: ) ) Host to Device Bandwidth for Pinned memory ) Transfer Size (Bytes) Bandwidth(MB/s) ) 33554432 2700.0 ) ) Device to Host Bandwidth for Pinned memory ) Transfer Size (Bytes) Bandwidth(MB/s) ) 33554432 2693.3 ) ) Device to Device Bandwidth ) Transfer Size (Bytes) Bandwidth(MB/s) ) 33554432 53768.0 ) ) In other words, once uploaded to the GPU, you can afford to reorder your data ) any way you want as often as you need to to take advantage of parallel ops. ) ) Or to put it another way, the GPU has about the same bandwidth to its memory ) as your brain does to its on a per-byte basis, in my ballpark estimates. ) Somewhere on the order of 100 GPUs is a brain-equivalent. ) ) We're getting damn close... ) ) Josh ) ) - ) This list is sponsored by AGIRI: http://www.agiri.org/email ) To unsubscribe or change your options, please go to: ) http://v2.listbox.com/member/?&; ) - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
Re: [agi] NVIDIA GPU's
On Friday 22 June 2007 11:22:19 am Bo Morgan wrote: > > That's 53.8 GB/s for a load of 33.6 MB? Is there a burst cache effect > going on here or do you think that's sustainable for multiple seconds? > > Bo ./bandwidthTest --mode=range --dtod --start=1000 --end=2 --increment=1000 Range Mode Device to Device Bandwidth Transfer Size (Bytes) Bandwidth(MB/s) 1000 47935.4 2000 47845.2 3000 48923.1 4000 49836.0 5000 52602.0 6000 53740.7 7000 54015.1 8000 53912.3 9000 53952.7 1 54050.9 11000 54001.9 12000 53965.0 13000 54010.2 14000 54089.5 15000 54026.4 16000 54019.1 17000 54067.7 18000 54110.9 19000 54004.4 2 54075.4 If I understand the specs right, there's a 320 bit wide data bus to on-card memory. The major hangup is that there just isn't that much (640MB on my GTS cards, 1.5GB on the Tesla) memory to start with. Josh - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
[agi] AGI introduction
Hi, I put a brief introduction to AGI at http://nars.wang.googlepages.com/AGI-Intro.htm , including an "AGI Overview" followed by "Representative AGI Projects". It is basically a bunch of links and quotations organized according to my opinion. Hopefully it can help some newcomers to get a big picture of the idea and the field. Pei - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
Re: [agi] AGI introduction
On 6/22/07, Pei Wang <[EMAIL PROTECTED]> wrote: Hi, I put a brief introduction to AGI at http://nars.wang.googlepages.com/AGI-Intro.htm , including an "AGI Overview" followed by "Representative AGI Projects". Thanks! As a first note, SAIL seems to me a better replacement for Cog, because SAIL has much generality and some theoretical accomplishment where Cog is (AFAIK) hand-crafted engineering. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
Re: [agi] AGI introduction
Pei:> I put a brief introduction to AGI at http://nars.wang.googlepages.com/AGI-Intro.htm , including an "AGI Overview" followed by "Representative AGI Projects". Very helpful. Thankyou. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
[agi] Association for Uncertainty in Artificial Intelligence
Looking through Wikipedia articles I stumbled upon a probably very interesting place: http://www.auai.org/ Association for Uncertainty in Artificial Intelligence - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
[agi] "Reinforcement Learning: An Introduction" Richard S. Sutton and Andrew G. Barto
Obligatory reading: http://www.cs.ualberta.ca/~sutton/book/ebook/the-book.html Cheers. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
Re: [agi] AGI introduction
On 6/22/07, Lukasz Stafiniak <[EMAIL PROTECTED]> wrote: As a first note, SAIL seems to me a better replacement for Cog, because SAIL has much generality and some theoretical accomplishment where Cog is (AFAIK) hand-crafted engineering. In many aspects, I agree that SAIL is more interesting than Cog. I include Cog in the list, because it is explicitly based on a theory about intelligence as a whole (see http://groups.csail.mit.edu/lbr/hrg/1998/group-AAAI-98.pdf), while in SAIL such a theory is not very clear. Of course, this boundary is fuzzy, so I may include SAIL in a future version of the list, depending on the development of the project. Pei - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
Re: [agi] "Reinforcement Learning: An Introduction" Richard S. Sutton and Andrew G. Barto
You make AGI sound like a "members only" club by this "obligatory" comment. ;) Reinforcement learning is a simple theory that only solves problems for which we can design value functions. We need some good readings about how to organize better programs. Books on how to program large complicated systems with many interconnected parts. Bo On Fri, 22 Jun 2007, Lukasz Stafiniak wrote: ) Obligatory reading: ) http://www.cs.ualberta.ca/~sutton/book/ebook/the-book.html ) ) Cheers. ) ) - ) This list is sponsored by AGIRI: http://www.agiri.org/email ) To unsubscribe or change your options, please go to: ) http://v2.listbox.com/member/?&; ) - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
Re: [agi] "Reinforcement Learning: An Introduction" Richard S. Sutton and Andrew G. Barto
On 6/23/07, Bo Morgan <[EMAIL PROTECTED]> wrote: Reinforcement learning is a simple theory that only solves problems for which we can design value functions. But it is good for AGI newbies like me to start with :-) - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e