He was the guy who slandered the mozilla foundation and called them names and then quit in a hissy fit.
That's a gross misinterpretation. He created mozilla.org quite literally when he registered the domain. Read the commentary that's been on his site for ages (You'll find a direct link in the article summary). He seems justifiably proud of initiating an open source browser and the ancillary tools created to develop it. The grandparent was referring to xscreensaver anyway.
You're right. I did some poking around and stumbled on an essay critical of the autotools on freshmeat. The (to me, educational) commentary on that article included that very same point. There seemed to be some consensus that autoconf was solid and worth appreciating. Automake on the other hand was fingered as a likely culprit when things go south.
AMD actually beats Intel's power specs too on the desktop, but I still don't see that making them a match (though it should). Cost seems a mute point. Dell cranks consumer computers out like fast food without AMD and Apple as a brand can set higher prices than them. It's been noted above that Apple doesn't want to have supply problems that might come from pairing with AMD. Surely Intel is a no-brainer in this department, but their roadmap may have as much to do with it. Apple's timing aligns their move away from PPC with Intel's move beyond the P4. That's Yonah, Merom, Dothan-on-the-desktop country. Ignore the ISA and you've got something very harmonious with Apple design.
If Apple really works on shining up Wine (or buys out some other Wine based company - Crossover I believe?), then they can offer Windows compatibility with a certain number of apps, perhaps a solid list such as Photoshop, Office, etc (and grow the list as necessary).
It might become an even cooler deal than that if the hypervisor tech Intel and AMD have on their roadmaps could enhance WINE. HL2 might then run just like the real thing (after all, what's dual-proc for?). Make an Aqua analog of WINE and the superfreaks might be able to run Final Cut Pro and Call of Duty 2 on the same machine without rebooting. Quad-core A64s + hypervisor + cracked OSXi + Cedega + VMware + MOL + UML + L4/Linux = many trippy and annoying screenshots.
That is an interesting point, but we don't really know much about what AMD will do to respond to Intel when they finally drop the P4. Everything I've been able to gather about the Pentium-M shows enormous potential on the desktop (Surely that's what Apple will be buying). A64 rules that roost at the moment, but I haven't heard much about AMDs plans beyond a new socket and hyper-visor-type technology. We know that Intel will eventually have a desktop chip that is dual-core, 64-bit, with SMT, SSE3, 2MB L1, and Vanderpool. It will be based on something they have right now that can be cooled like a PIII. It will be able to run on Intel's chipsets. Is AMD going to stay on top of that or just in the game? Does anyone know more?
Just to hammer home some agreement with another reply, here's a quote from Ryan Gordon I found from another/. story: "SSE3 is way more flexible and feature-rich than Altivec. I don't care what the haters say, that's the truth. In terms of vectorization, SSE will be a better instruction set. There are lots of places were you get the feeling that Altivec should be a good optimization, but you can't find a way to coerce it to do what you want without a lot of effort (and frequently, without making the program _slower_). SSE has wider application. It just does, even without names like "Velocity Engine"."
If you want to program in assembly, it's easier for a beginner to pickup risc because of its regularity, but I highly doubt most that the people complianing about the ISA actually used it and were just complaining that it seemed to them to be unelegant...
Yeah, I don't buy this complaint about the ISA. PPC and x86 SIMD instructions are mostly the same. "Legacy" instructions add impurity to the x86 ISA but that's it. They don't force a PIII to run at 33 or 66 or 133 MHz. The only thing I'd knock is the endianness, but where is that a show stopper? Even segmented memory has certain benefits. The number of PPC opcodes has grown to equal that of the x86 world anyway and most code shouldn't be written any closer to the metal than C.
I can certainly see your point about Speedstep being a feature and acknowledge the spin that would wrap an announcement like this, but I still disagree with your comparative assessment of 970. I tried to track down the numbers you cited earlier but didn't find them on the ibm link and top500.com sends me to one of those acursed domain-parking-search-portal-things. Being able to see how the G5 can run at 24 Watts would bolster your argument, but still beg an explanation as to why there are no G5 laptops and why the Powermac has a liquid filled heatsink. It may be that the 970s power management is less granular or dramatic in comparison to Speedstep, but is it less sophisticated than the G4?
Googling for G5 power consumption specs isn't leading me to numbers like the ones you've posted.
This pdf
from IBM has the 970 running at between 65 and 75 Watts. I'm just not finding the numbers that can buttress your claim that the 970 does the most work for the least amount of current. What I'm finding is that 970 is a touch better than A64 with the amps (at least for pre-Winchester), A64 is a touch better than 970 with IPC, and that Dothan beats them soundly on both accounts but is unscaled and chokes on certain applications. Dothan apparently can run at 2.6+ GHz right now with a P3 class heatsink if you want to break pins on your $250 processor and take a soldering iron to a $300 motherboard. Is that not a harbinger of the future? A fivefold difference in bogomips per watt may sound extreme but it isn't beyond plausibility. I wouldn't argue that the G5 sucks. I just wouldn't claim it's flat out superior to any other consumer CPU or that it's impossible for Yonah/Merom to leave it behind.
I can't claim to understand all the issues involved in accounting for portability, but finding that someone else thinks configure is unreadable makes me feel better about it. It's mind bending to check something out of anonymous cvs, find that its autotooling breaks on your system, and then try to debug the problem. The files generated can be byzantine and inscrutable. Checks are done more than once. The configure script itself can be twice the size of all the.c and.h files combined. Surely there is room for a cleaner, simpler functional equivalent to the autotools.
Re:Here's why Jobs likes Intel and not AMD
on
Apple Switching to Intel
·
· Score: 2, Interesting
I've been doing a bit of googling lately on A64 and the Pentium-M. I'm not sure it's quite that clear. Venice versus Dothan is a tough call. Overclock the Dothan and it can outperform Venice by a bit with less power to boot. But it still gets hung up on memory bandwith and floating-point stuff. I found your link to the lostcircuits article very interesting. I knew that recent 939s (from Winchester onward) were very good in the power consumption department, but I didn't know it was that good. However, I haven't found any decent evaluations of the Turion that could back up your claim for its performance and efficiency.
In any event, it doesn't matter because Apple is getting in on the future Intel CPUs which will be derived from Banias/Dothan. Here is an interesting thread on
overclocking Dothan. The results make me wonder if Intel has kept it's FSB hobbled to avoid direct competition with the P4. We have to imagine Yonah or Merom will be like one of these little screamers plus dual core and a better FSB. Of course, AMD will have improved too between now and then, but if tweakers can make Dothan competetive with a cooling solution that belongs on a northbridge then I would imagine that Intel can flat out prove to Jobs just what they have coming down the pike.
They passed by AMD because they want what Intel is cooking up to replace the P4. Right now A64 is the shit hands down. But Dothan is the IPC and bogomips/Watt king. That's the line Apple is jumping in on.
Steve Jobs summed it up at the conference by noting that IBM's PowerPC road map would only deliver about a fifth the performace per watt as a comparable Intel chip. How is that a weaker product?
The laptop issue is about dynamic power features not inherent power consumption where the PPC whoops X86.
That's not true. The 970 runs at 66 Watts which makes it directly comparable to an A64. Intel's most power thrifty mobile runs at a bit more than 9 Watts (the figure is Intel TDP) and can drop down to 1 Watt. Dothan (which has a higher IPC than 970) tops out in the low 30s.
As far as the performance goes, the article posted just the other day shows that the 970 isn't king of the hill. It lost 7 out of 8 floating point benchmarks to an Opteron machine with a slower clock. It won Lightwave Raytrace by 0.5 seconds. It lost Queens, Povray, Lightwave Radiosity, and Cinema 4D. What exactly is so great about this chip? It isn't exceptionally fast or efficient or miserly with energy.
As far as your thesis goes, lets drop the conspiracy theory. Apple is not a poor company and they're switching to Yonah or Merom for the same reason Intel is, and for the same reason I might when the time comes. Its a better chip.
To add a tangent to your observations, I've never been able to buy the RISC camp's charge that x86/CISC is broken because the chips are RISC on the inside -- as if that were some sort of vindication. The argument is semantic. One could just as easily characterize RISC chips as lacking ISA compression which (as you noted) effectively increases memory bandwidth.
Yeah, I don't understand all the bitching about what this guy did or didn't compare. A third or more of the article was devoted to chip comparison with very specific, low level metrics. Kernel performance was benchmarked. The hardware was compared. The software was compared.
Incorrect. In many cases, such jokes are playing off the fact that these ridiculous stereotypes exist in the first place. Transcendant post-modern humor.
Exactly. It's that and the fact that Macs are so Gay...
You know, when I noticed this man's name in the summary I felt an instinctive, atypical desire to submit a slanderous post about him without even reading the article. It's amazing how intuition can save time. I don't know who this guy is or why I remember him as a total asshole, but the two sentences you've quoted reassure my faith in my own foggy notions. Dear Lord, what a shithead.
To provide another strange example consider the life of Ezra Pound. One of the distinguishing marks of his prose was his somewhat justified, almost always arrogant posture of intelectual superiority. His verse however was informed by bigotry, structured by imitation, and almost entirely recanted at the end of his life over a moral awakening that came far too late for pity. Being able to write in 18 languages doesn't mean a man has anything worth writting about.
Yeah, I know. Some of the posts on this topic were so irksome that I must have skimped on clairity when I chimed in. There is a natural tendency for people to take things for granted and treat privileges like rights. I wasn't saying the cafe is wrong to turn it off. I was saying they turn it off instead of asking people to leave because people are assholes. The staff have more important things to do than teach adults manners. They don't owe anyone a WiFi equiped livingroom. Yet some people apparently think they do. So why is it so great to note their lack of a "manners enforcement" policy to correct the problem? It's naive and at best a kind of backhanded form of sympathy.
You might be surprised at how far the presumptions extend when the social contract and the service industry intersect. Do you really not know what the response to that challenge will be? "I'm taking my business elsewhere."
It's twisted.
In fact, it's a bit like your post with further reflection. The important thing isn't to appreciate the way the cafe threw a cool bonus into the environment. What's really important is to note what chumps/morons they are for simply turning it off periodically to keep the place the way they want it to be instead of roaming the place like cops.
Slashdot Mirror
That's a gross misinterpretation. He created mozilla.org quite literally when he registered the domain. Read the commentary that's been on his site for ages (You'll find a direct link in the article summary). He seems justifiably proud of initiating an open source browser and the ancillary tools created to develop it. The grandparent was referring to xscreensaver anyway.
You're right. I did some poking around and stumbled on an essay critical of the autotools on freshmeat. The (to me, educational) commentary on that article included that very same point. There seemed to be some consensus that autoconf was solid and worth appreciating. Automake on the other hand was fingered as a likely culprit when things go south.
AMD actually beats Intel's power specs too on the desktop, but I still don't see that making them a match (though it should). Cost seems a mute point. Dell cranks consumer computers out like fast food without AMD and Apple as a brand can set higher prices than them. It's been noted above that Apple doesn't want to have supply problems that might come from pairing with AMD. Surely Intel is a no-brainer in this department, but their roadmap may have as much to do with it. Apple's timing aligns their move away from PPC with Intel's move beyond the P4. That's Yonah, Merom, Dothan-on-the-desktop country. Ignore the ISA and you've got something very harmonious with Apple design.
It might become an even cooler deal than that if the hypervisor tech Intel and AMD have on their roadmaps could enhance WINE. HL2 might then run just like the real thing (after all, what's dual-proc for?). Make an Aqua analog of WINE and the superfreaks might be able to run Final Cut Pro and Call of Duty 2 on the same machine without rebooting. Quad-core A64s + hypervisor + cracked OSXi + Cedega + VMware + MOL + UML + L4/Linux = many trippy and annoying screenshots.
After all we are speculating here...
That is an interesting point, but we don't really know much about what AMD will do to respond to Intel when they finally drop the P4. Everything I've been able to gather about the Pentium-M shows enormous potential on the desktop (Surely that's what Apple will be buying). A64 rules that roost at the moment, but I haven't heard much about AMDs plans beyond a new socket and hyper-visor-type technology. We know that Intel will eventually have a desktop chip that is dual-core, 64-bit, with SMT, SSE3, 2MB L1, and Vanderpool. It will be based on something they have right now that can be cooled like a PIII. It will be able to run on Intel's chipsets. Is AMD going to stay on top of that or just in the game? Does anyone know more?
Just to hammer home some agreement with another reply, here's a quote from Ryan Gordon I found from another /. story: "SSE3 is way more flexible and feature-rich than Altivec. I don't care what the haters say, that's the truth. In terms of vectorization, SSE will be a better instruction set. There are lots of places were you get the feeling that Altivec should be a good optimization, but you can't find a way to coerce it to do what you want without a lot of effort (and frequently, without making the program _slower_). SSE has wider application. It just does, even without names like "Velocity Engine"."
Yeah, I don't buy this complaint about the ISA. PPC and x86 SIMD instructions are mostly the same. "Legacy" instructions add impurity to the x86 ISA but that's it. They don't force a PIII to run at 33 or 66 or 133 MHz. The only thing I'd knock is the endianness, but where is that a show stopper? Even segmented memory has certain benefits. The number of PPC opcodes has grown to equal that of the x86 world anyway and most code shouldn't be written any closer to the metal than C.
Googling for G5 power consumption specs isn't leading me to numbers like the ones you've posted. This pdf from IBM has the 970 running at between 65 and 75 Watts. I'm just not finding the numbers that can buttress your claim that the 970 does the most work for the least amount of current. What I'm finding is that 970 is a touch better than A64 with the amps (at least for pre-Winchester), A64 is a touch better than 970 with IPC, and that Dothan beats them soundly on both accounts but is unscaled and chokes on certain applications. Dothan apparently can run at 2.6+ GHz right now with a P3 class heatsink if you want to break pins on your $250 processor and take a soldering iron to a $300 motherboard. Is that not a harbinger of the future? A fivefold difference in bogomips per watt may sound extreme but it isn't beyond plausibility. I wouldn't argue that the G5 sucks. I just wouldn't claim it's flat out superior to any other consumer CPU or that it's impossible for Yonah/Merom to leave it behind.
I just found this: "Stop the Autoconf Insanity".
I can't claim to understand all the issues involved in accounting for portability, but finding that someone else thinks configure is unreadable makes me feel better about it. It's mind bending to check something out of anonymous cvs, find that its autotooling breaks on your system, and then try to debug the problem. The files generated can be byzantine and inscrutable. Checks are done more than once. The configure script itself can be twice the size of all the .c and .h files combined. Surely there is room for a cleaner, simpler functional equivalent to the autotools.
In any event, it doesn't matter because Apple is getting in on the future Intel CPUs which will be derived from Banias/Dothan. Here is an interesting thread on overclocking Dothan. The results make me wonder if Intel has kept it's FSB hobbled to avoid direct competition with the P4. We have to imagine Yonah or Merom will be like one of these little screamers plus dual core and a better FSB. Of course, AMD will have improved too between now and then, but if tweakers can make Dothan competetive with a cooling solution that belongs on a northbridge then I would imagine that Intel can flat out prove to Jobs just what they have coming down the pike.
They passed by AMD because they want what Intel is cooking up to replace the P4. Right now A64 is the shit hands down. But Dothan is the IPC and bogomips/Watt king. That's the line Apple is jumping in on.
Steve Jobs summed it up at the conference by noting that IBM's PowerPC road map would only deliver about a fifth the performace per watt as a comparable Intel chip. How is that a weaker product?
That's not true. The 970 runs at 66 Watts which makes it directly comparable to an A64. Intel's most power thrifty mobile runs at a bit more than 9 Watts (the figure is Intel TDP) and can drop down to 1 Watt. Dothan (which has a higher IPC than 970) tops out in the low 30s.
As far as the performance goes, the article posted just the other day shows that the 970 isn't king of the hill. It lost 7 out of 8 floating point benchmarks to an Opteron machine with a slower clock. It won Lightwave Raytrace by 0.5 seconds. It lost Queens, Povray, Lightwave Radiosity, and Cinema 4D. What exactly is so great about this chip? It isn't exceptionally fast or efficient or miserly with energy.
As far as your thesis goes, lets drop the conspiracy theory. Apple is not a poor company and they're switching to Yonah or Merom for the same reason Intel is, and for the same reason I might when the time comes. Its a better chip.
To add a tangent to your observations, I've never been able to buy the RISC camp's charge that x86/CISC is broken because the chips are RISC on the inside -- as if that were some sort of vindication. The argument is semantic. One could just as easily characterize RISC chips as lacking ISA compression which (as you noted) effectively increases memory bandwidth.
Why should they? If you want to know how Linux handles an increasing amount of threads the numbers are right there.
Yeah, I don't understand all the bitching about what this guy did or didn't compare. A third or more of the article was devoted to chip comparison with very specific, low level metrics. Kernel performance was benchmarked. The hardware was compared. The software was compared.
Exactly. It's that and the fact that Macs are so Gay...
You know, when I noticed this man's name in the summary I felt an instinctive, atypical desire to submit a slanderous post about him without even reading the article. It's amazing how intuition can save time. I don't know who this guy is or why I remember him as a total asshole, but the two sentences you've quoted reassure my faith in my own foggy notions. Dear Lord, what a shithead.
I just saved that image.
To provide another strange example consider the life of Ezra Pound. One of the distinguishing marks of his prose was his somewhat justified, almost always arrogant posture of intelectual superiority. His verse however was informed by bigotry, structured by imitation, and almost entirely recanted at the end of his life over a moral awakening that came far too late for pity. Being able to write in 18 languages doesn't mean a man has anything worth writting about.
Yeah, I know. Some of the posts on this topic were so irksome that I must have skimped on clairity when I chimed in. There is a natural tendency for people to take things for granted and treat privileges like rights. I wasn't saying the cafe is wrong to turn it off. I was saying they turn it off instead of asking people to leave because people are assholes. The staff have more important things to do than teach adults manners. They don't owe anyone a WiFi equiped livingroom. Yet some people apparently think they do. So why is it so great to note their lack of a "manners enforcement" policy to correct the problem? It's naive and at best a kind of backhanded form of sympathy.
Play the tape either way. If lumps of powder appear on your coffee you've got it backwards. If it turns creamier it's alright.
It's twisted.
In fact, it's a bit like your post with further reflection. The important thing isn't to appreciate the way the cafe threw a cool bonus into the environment. What's really important is to note what chumps/morons they are for simply turning it off periodically to keep the place the way they want it to be instead of roaming the place like cops.
Oh, really?