Anonymity breeds distrust in public communication. Whether it's trolling for fun or misinforming for profit, the upshot is a building general distrust of the communications channel itself. It is literally communications breakdown.
The only solution to this is full authentication of every user on every computer throughout the net, with some government controlled centralized database. In other words, DRM on steroids. And the total end of anonymous political dissent.
I wuz mocking with ya until the quip about dipping that laptop into a vat of spotted owl feathers. I gotta say, were I rich I'd want to buy that. I mean, nothing says 'I've got more money than you' like covering your overpriced depreciating asset in material from the death of an endangered species.
DMA has latency, and requires a response from the PPU on an interrupt. I realize the DMA controller can move arbitrary pages mapped from the page table into LS. It's - what - broken up into 16KB blocks. Right? I think that's right. But within LS the SPU doesn't have real memory protection. Arguably, it doesn't need it.
WRT your last paragraph: you are absolutely right. I'm looking at this from the perspective of one who wants to solve a single problem, and who isn't a professional Cell dev. I have a problem I want to solve and am hoping that the PS3 will be a cheap alternative to PCs for my lab.
I just recently bought a PS3 and stuck linux on it. I've got the source to the tools I need. Now I need to figure out the best way to get the thing ported right. And... I'm doing it on my own off hours. Mostly because nobody at work really believes that a toy could be cost effective for scientific compute. I disagree.
BTW: your two posts are by far the most informative I've seen about Cell 'round here. Thank you very much for taking time to correct me. Would you be willing to drop me an email and give advice? My addy is public on my/. homepage. Feel free to send from a throw away address if you need to remain anonymous.
Yes. I was absolutely wrong. I've been corrected numerous times, but because that post is improperly modded up it's getting undue attention. Thanks for the detailed correction.
I was already corrected on the issue of floats. I appear to have been misinformed. However, on many of these issues you appear to be splitting hairs. For example, you appear to argue that an SPU does not lack a supervisor bit - but instead go on a tangent about how SPUs don't have local memory access, but must instead perform a DMA transfer. Well, as it turns out, the SPU does NOT have a supervisor bit for memory protection.
Your other arguments boil down to: we don't program Cell like that. In particular, you're discussing how to double buffer and/or stream many code/data segments on Cell with low performance penalty. I'm sure you're right for your particular task (which I assume is game programming). However, that is not the best choice for coding up a Monte Carlo sim. Particularly since it might actually be possible to cram the algorithm and data segment straight into LS. And regarding your userspace threading argument: when a simulation that just spent CPU days (or weeks) in compute goes tits up, bad things do happen. Heh.
Single and double precision is entirely dependent on the hardware configuration. However, as noted in a prior post, It's absolutely true that Cell is currently designed to a subset of the IEEE floating point standard. So... 32bit single precision floats is right; I was wrong.
Single precision floating point computation is geared for throughput of media and 3D graphics objects. In this vein, the decision to support only a subset of IEEE floating point arithmetic and sacrifice full IEEE compliance was driven by the target applications. Thus, multiple rounding modes and IEEE-compliant exceptions are typically unimportant for these workloads, and are not supported. This design decision is based the real time nature of game workloads and other media applications: most often, saturation is mathematically the right solution. Also, occasional small display glitches caused by saturation in a display frame is tolerable. On the other hand, incomplete rendering of a display frame, missing objects or tearing video due to long exception handling is objectionable.
Cell is very optimized toward one data type for calculation: 64bit floats. If you want to efficiently use the PS3 in a cluster, just be aware that your code must:
a) use primarily 64bit floating point
b) either:
- fit code and data segments within 256K for each SPU
- crunch long enough between streamed data blocks such that DMA latency doesn't kill performance
c) have the entire calculation broken down into no more than six parts for streaming (one per SPU)
There are SPU userspace threading models that run cooperatively (similar to the old userspace pthreads, I guess), but the thread manager consumes valuable SPU RAM. Also, SPUs don't support a supervisor bit for memory protection... so... bad things happen when threaded code running on SPU goes tits up.
If you want to calculate 128bit floats, ints, or have lots of branch logic... buy a quad core2duo; cell don't do you any good.
Sorry, that goes to the preface, not Dennis' 'anti-forward'. It's not available in html online, but it is in the pdf I posted in my parent comment. Apologies.
Dunno how Simpson Garfinkle feels about Plan 9. But the author of Plan 9 did write the preface to The UNIX Haters Handbook. You can read Dennis Ritchie's remarks here.
Except Simpson didn't go to MIT back in the day, and never hacked ITS. At least, I don't think he did. There were lots of UNIX hackers at Symbolics (as well as LMI). Though it's true that the place, like BBN, was a revolving door for MIT dropouts and recent graduates.
"And finally the old unix guys will flame about how none of these vulnerabilites would have happened if we would have stayed away from GUIs."
No. Old UNIX hackers will instead berate UNIX for being a total piece of shit and then endlessly whine about the downfall of Symbolics and its old dedicated LISP machines. And they'd be right.
Actually, one of the biggest reasons I bought the PS3 was to try to port a Cell optimized GEANT and BLAST to Yellowdog. These are popular Monte Carlo engines. My lab is broke and could use a cheap alternative to quad-opterons. That said, few professors here are optimistic, so I'm doing this on my own.
IMO: if you have a 64bit floating point problem that either:
a) fits data and object code within the paltry 256K of an SPU, you're golden.
b) breaks down into five or less parts and computes slower than DMA latency so you can stream, you're golden.
If you need 128bit floats, or ints, or have lots of branch logic... forget it; a quad opteron or quad core2duo is the way to go.
Well... you're right. That Samsung BDP-1000 is $.23 cheaper than a 20GB PS3. I wouldn't buy it though. I think you'll find that the AVS Forum thread on that unit has a good number of negative comments about it:
But yeah, it's great to finally see these things hit below $500 finally. Hopefully, when Sony releases the BDP-S300 we'll see the street price fall down to ~$350-$400.
No. The PS3 outputs 1080i to my PJ, which scales it down to 720p internally. But you're right that the PS3 lacks an internal scaler. However, while there are plenty of 1080i only sets out there, I've never seen a 720p set that doesn't also take 1080i input. And I'll tell you, the visual difference between SD-DVD and Blu-Ray is *astonishing*. It's so freak'n good I nearly creamed my pants. And I'm nearly 40, well over the hill and closing in on my geriatric viagra days...
We're on opposite sides of the pond, so there are some differences. Terrestrial HD broadcast ATSC is still MPEG2 and will likely remain that for a good while. Cable and satellite are transitioning to MPEG4 someday, though it's unclear just when. For now they're still broadcasting MPEG2 over QAM. So, for us USians, BD and HD-DVD would appear to offer the best quality available at the moment. So I guess you UKians are lucky if you can get the satellite service.
All of your bullet points are correct, but I want to respond to a couple of them:
1) Yes. The PS3 does not output 720p for blu-ray movies. I suspect this is the same problem as 720p games being unable to scale to 1080i - the unit lacks a hardware scaler. My 720p projector takes 1080i as input and that's what I used. Yes, I'm sure scaling from 1080i to 720p in the projector isn't the best option. It still looks freak'n great compared to cable HD and OTA HD.
8) Oh God yes. The selection even for BD is nothing like what I want. However, netflix does have over 250 titles available right now, and if you check upcoming title availability BD has a bunch of releases in the works. Here's a list:
Uhhhh... that would be an HD-DVD player, and not a Blu-Ray player. Can you find me a Blu-Ray player that's cheaper than a 20GB PS3? I don't think so. But, within six months, you will be able to.
I also own a 360. The PS3 is connected to a Sony HS-20 720p digital projector via HDMI and to a standard dd5.1 sound system via optical out. Here are my impressions:
Setup on the PS3 is very buggy and filled with poorly translated instructions.
- when first initially turning the unit on, it will auto-detect HDMI and display 480p. The second or third question it then asks is: "Do you want video and sound to be output via HDMI"? Since I wanted to output sound via optical, I thought this was asking if I wanted to split audio off the HDMI connection, so I selected "No". What then happened was that it spit 480i video out the composite cable.
- When initializing the network via wireless it will search and then display a list of available wireless SSIDs. Select one and DHCP an ip address. Now test the network. It succeeds in grabbing an IP address, but the network test always reports a failure even though the network is live and updates can be downloaded.
- Everyone knows about the lack of background downloading. Blech.
All that said, once I stuck a Blu-Ray copy of Casino Royale... whoah. I gotta say, the image is stunning. BD is definitely much much much better than HBO-HD, SHO-HD and OTA HD material. NO pixelation whatsoever. Extremely fast video plays without a hiccup. I'm IMPRESSED.
I don't have an HD-DVD player, so I can't compare the two. I suspect they're about the same in quality. But Blu-Ray has the movies. So... my impression is that if you want a BD player, the PS3 is the cheapest option out there and it's VERY GOOD.
As for games.... well, frankly, I'm much more impressed by Gears of War than I am by Resistance. JMO.
graphics call access to RSX is blocked by a hypervisor which has control over the kernel. I'm asking if Sony plans to create a hypervisor interface to RSX, as well as kernel patches and optimized GL support.
I don't have time to worry about internal Debian politics. Perhaps it is a clusterfuck. Beats me. But Debian Stable (Woody) may run old software, may lack some desirable features, and may not have the latest Gnome interface... but so what. It is stable. I have a cluster of machines running Stable that serve AFS to hundreds of clients. With those machines, my problems are almost all hardware related.
That's all I care about. Is it stable? Yes. Is it secure? Yes. Does it perform a function I need? Yes. Then deploy.
Slashdot Mirror
Anonymity breeds distrust in public communication. Whether it's trolling for fun or misinforming for profit, the upshot is a building general distrust of the communications channel itself. It is literally communications breakdown.
The only solution to this is full authentication of every user on every computer throughout the net, with some government controlled centralized database. In other words, DRM on steroids. And the total end of anonymous political dissent.
Which is worse? I have my opinion.
Oh. Of course not. Such things have never happened in the past.
The Pinkertons specialized in such corporate sponsored actions.
I wuz mocking with ya until the quip about dipping that laptop into a vat of spotted owl feathers. I gotta say, were I rich I'd want to buy that. I mean, nothing says 'I've got more money than you' like covering your overpriced depreciating asset in material from the death of an endangered species.
I say we need to add elephant tusk too!
Yeah. Two points:
/. homepage. Feel free to send from a throw away address if you need to remain anonymous.
DMA has latency, and requires a response from the PPU on an interrupt. I realize the DMA controller can move arbitrary pages mapped from the page table into LS. It's - what - broken up into 16KB blocks. Right? I think that's right. But within LS the SPU doesn't have real memory protection. Arguably, it doesn't need it.
WRT your last paragraph: you are absolutely right. I'm looking at this from the perspective of one who wants to solve a single problem, and who isn't a professional Cell dev. I have a problem I want to solve and am hoping that the PS3 will be a cheap alternative to PCs for my lab.
I just recently bought a PS3 and stuck linux on it. I've got the source to the tools I need. Now I need to figure out the best way to get the thing ported right. And... I'm doing it on my own off hours. Mostly because nobody at work really believes that a toy could be cost effective for scientific compute. I disagree.
BTW: your two posts are by far the most informative I've seen about Cell 'round here. Thank you very much for taking time to correct me. Would you be willing to drop me an email and give advice? My addy is public on my
Yes. I was absolutely wrong. I've been corrected numerous times, but because that post is improperly modded up it's getting undue attention. Thanks for the detailed correction.
I was already corrected on the issue of floats. I appear to have been misinformed. However, on many of these issues you appear to be splitting hairs. For example, you appear to argue that an SPU does not lack a supervisor bit - but instead go on a tangent about how SPUs don't have local memory access, but must instead perform a DMA transfer. Well, as it turns out, the SPU does NOT have a supervisor bit for memory protection.
Your other arguments boil down to: we don't program Cell like that. In particular, you're discussing how to double buffer and/or stream many code/data segments on Cell with low performance penalty. I'm sure you're right for your particular task (which I assume is game programming). However, that is not the best choice for coding up a Monte Carlo sim. Particularly since it might actually be possible to cram the algorithm and data segment straight into LS. And regarding your userspace threading argument: when a simulation that just spent CPU days (or weeks) in compute goes tits up, bad things do happen. Heh.
Single and double precision is entirely dependent on the hardware configuration. However, as noted in a prior post, It's absolutely true that Cell is currently designed to a subset of the IEEE floating point standard. So... 32bit single precision floats is right; I was wrong.
http://www.research.ibm.com/cell/SPU.html
No. It's my understanding that Cell uses 64bits for single precision and 128bits for double precision.
Cell is very optimized toward one data type for calculation: 64bit floats. If you want to efficiently use the PS3 in a cluster, just be aware that your code must:
a) use primarily 64bit floating point
b) either:
- fit code and data segments within 256K for each SPU
- crunch long enough between streamed data blocks such that DMA latency doesn't kill performance
c) have the entire calculation broken down into no more than six parts for streaming (one per SPU)
There are SPU userspace threading models that run cooperatively (similar to the old userspace pthreads, I guess), but the thread manager consumes valuable SPU RAM. Also, SPUs don't support a supervisor bit for memory protection... so... bad things happen when threaded code running on SPU goes tits up.
If you want to calculate 128bit floats, ints, or have lots of branch logic... buy a quad core2duo; cell don't do you any good.
BTW: Anyone here hacking GEANT or BLAST for Cell?
Sorry, that goes to the preface, not Dennis' 'anti-forward'. It's not available in html online, but it is in the pdf I posted in my parent comment. Apologies.
Dunno how Simpson Garfinkle feels about Plan 9. But the author of Plan 9 did write the preface to The UNIX Haters Handbook. You can read Dennis Ritchie's remarks here.
Except Simpson didn't go to MIT back in the day, and never hacked ITS. At least, I don't think he did. There were lots of UNIX hackers at Symbolics (as well as LMI). Though it's true that the place, like BBN, was a revolving door for MIT dropouts and recent graduates.
BTW: Are you affiliated with the CDF project?
"And finally the old unix guys will flame about how none of these vulnerabilites would have happened if we would have stayed away from GUIs."
No. Old UNIX hackers will instead berate UNIX for being a total piece of shit and then endlessly whine about the downfall of Symbolics and its old dedicated LISP machines. And they'd be right.
Actually, one of the biggest reasons I bought the PS3 was to try to port a Cell optimized GEANT and BLAST to Yellowdog. These are popular Monte Carlo engines. My lab is broke and could use a cheap alternative to quad-opterons. That said, few professors here are optimistic, so I'm doing this on my own.
IMO: if you have a 64bit floating point problem that either:
a) fits data and object code within the paltry 256K of an SPU, you're golden.
b) breaks down into five or less parts and computes slower than DMA latency so you can stream, you're golden.
If you need 128bit floats, or ints, or have lots of branch logic... forget it; a quad opteron or quad core2duo is the way to go.
Well... you're right. That Samsung BDP-1000 is $.23 cheaper than a 20GB PS3. I wouldn't buy it though. I think you'll find that the AVS Forum thread on that unit has a good number of negative comments about it:
8 8793
http://www.avsforum.com/avs-vb/showthread.php?t=6
But yeah, it's great to finally see these things hit below $500 finally. Hopefully, when Sony releases the BDP-S300 we'll see the street price fall down to ~$350-$400.
No. The PS3 outputs 1080i to my PJ, which scales it down to 720p internally. But you're right that the PS3 lacks an internal scaler. However, while there are plenty of 1080i only sets out there, I've never seen a 720p set that doesn't also take 1080i input. And I'll tell you, the visual difference between SD-DVD and Blu-Ray is *astonishing*. It's so freak'n good I nearly creamed my pants. And I'm nearly 40, well over the hill and closing in on my geriatric viagra days...
(so you should be amazed)
Excellent post.
8 2217
We're on opposite sides of the pond, so there are some differences. Terrestrial HD broadcast ATSC is still MPEG2 and will likely remain that for a good while. Cable and satellite are transitioning to MPEG4 someday, though it's unclear just when. For now they're still broadcasting MPEG2 over QAM. So, for us USians, BD and HD-DVD would appear to offer the best quality available at the moment. So I guess you UKians are lucky if you can get the satellite service.
All of your bullet points are correct, but I want to respond to a couple of them:
1) Yes. The PS3 does not output 720p for blu-ray movies. I suspect this is the same problem as 720p games being unable to scale to 1080i - the unit lacks a hardware scaler. My 720p projector takes 1080i as input and that's what I used. Yes, I'm sure scaling from 1080i to 720p in the projector isn't the best option. It still looks freak'n great compared to cable HD and OTA HD.
8) Oh God yes. The selection even for BD is nothing like what I want. However, netflix does have over 250 titles available right now, and if you check upcoming title availability BD has a bunch of releases in the works. Here's a list:
http://www.avsforum.com/avs-vb/showthread.php?t=7
Uhhhh... that would be an HD-DVD player, and not a Blu-Ray player. Can you find me a Blu-Ray player that's cheaper than a 20GB PS3? I don't think so. But, within six months, you will be able to.
I also own a 360. The PS3 is connected to a Sony HS-20 720p digital projector via HDMI and to a standard dd5.1 sound system via optical out. Here are my impressions:
Setup on the PS3 is very buggy and filled with poorly translated instructions.
- when first initially turning the unit on, it will auto-detect HDMI and display 480p. The second or third question it then asks is: "Do you want video and sound to be output via HDMI"? Since I wanted to output sound via optical, I thought this was asking if I wanted to split audio off the HDMI connection, so I selected "No". What then happened was that it spit 480i video out the composite cable.
- When initializing the network via wireless it will search and then display a list of available wireless SSIDs. Select one and DHCP an ip address. Now test the network. It succeeds in grabbing an IP address, but the network test always reports a failure even though the network is live and updates can be downloaded.
- Everyone knows about the lack of background downloading. Blech.
All that said, once I stuck a Blu-Ray copy of Casino Royale... whoah. I gotta say, the image is stunning. BD is definitely much much much better than HBO-HD, SHO-HD and OTA HD material. NO pixelation whatsoever. Extremely fast video plays without a hiccup. I'm IMPRESSED.
I don't have an HD-DVD player, so I can't compare the two. I suspect they're about the same in quality. But Blu-Ray has the movies. So... my impression is that if you want a BD player, the PS3 is the cheapest option out there and it's VERY GOOD.
As for games.... well, frankly, I'm much more impressed by Gears of War than I am by Resistance. JMO.
You could do this too, you don't have to be a federal agent.
Only if you want to go to jail. Impersonating a federal agent is a felony.
The PS3 is not a PC. You might want to read this:
/ LinuxKernelOverview.html
http://moss.csc.ncsu.edu/~mueller/cluster/ps3/doc
graphics call access to RSX is blocked by a hypervisor which has control over the kernel. I'm asking if Sony plans to create a hypervisor interface to RSX, as well as kernel patches and optimized GL support.
Sir,
Will Sony be offering up kernel patches and an RSX optimized OpenGL library for PS3/Linux?
You're right. Sarge is what's deployed. Woody is deprecated, but I believe they're still providing security updates for it. My error.
I don't have time to worry about internal Debian politics. Perhaps it is a clusterfuck. Beats me. But Debian Stable (Woody) may run old software, may lack some desirable features, and may not have the latest Gnome interface... but so what. It is stable. I have a cluster of machines running Stable that serve AFS to hundreds of clients. With those machines, my problems are almost all hardware related.
That's all I care about. Is it stable? Yes. Is it secure? Yes. Does it perform a function I need? Yes. Then deploy.