mootness n. Usage Note: The adjective moot is originally a legal term going back to the mid-16th century. It derives from the noun moot, in its sense of a hypothetical case argued as an exercise by law students. Consequently, a moot question is one that is arguable or open to debate. But in the mid-19th century people also began to look at the hypothetical side of moot as its essential meaning, and they started to use the word to mean "of no significance or relevance." Thus, a moot point, however debatable, is one that has no practical value. A number of critics have objected to this use, but 59 percent of the Usage Panel accepts it in the sentence The nominee himself chastised the White House for failing to do more to support him, but his concerns became moot when a number of Republicans announced that they, too, would oppose the nomination. When using moot one should be sure that the context makes clear which sense is meant.
but the data isn't getting onto or off the chips at anything like the speed required to keep up. >>>>>>>>> First, we're talking about single CPUs here, not whole systems. Second, your comments haven't been true for a couple of years. The memory subsystems of current PC chips are very good. Good enough that they're entirely competitive with Sun or SGI machines in the same class (2-4 processors). Beyond that, the Sun or SGI machines might scale better, but the overall performance of the PC in its "weight class" (small SMP machines) is unmatched. Nobody thinks that a PC architecture machine will match up to a Power4 in the Power4's normal configuration (16 or 32 way) but it is true that a P4 is on the heels of a single Power4 CPU.
Your explanation is dead on, your nomenclature is reversed. Current CD drives are CAV (Constant Angular Velocity) while before they were CLV (Constant Linear Velocity).
Do you understand the concept of hyperbole? It was precisely because of Motorola's focus on the embedded market that I called the G4 an embedded processor. Note that I called Banias an embedded processor as well, which it clearly is not.
Next, the 2 and 3 GHz marks are not useless. The thing PowerPC people don't understand is that Performance = ClockRate * Efficiency. Note that both are linear factors. Raising the effifiency does not improve the overall performance any more than raising clock rate. However, raising ClockRate is a lot easier than raising efficiency, because most code (outside of a few problem domains) does not lend itself to extensive parallization. Intel, with the P4 architecture, made an engineering decision to emphasise clock rate over efficiency. As the benchmarks clearly show, Intel achieved their goal of having absolutely incredible performance, to the point where a P4 provides more than 70% the floating point performance of a Power4 in CPU-bound benchmarks. PowerPC-heads can wave their arms and shout "efficiency" as much as they want, all that matters is the end result performance.
Lastly, the G4 is a sucky chip. Its clock rates are lower than its x86 competitors. It's IPC is lower than chips like the Alpha or POWER. It's overall performance isn't very good. In an age when x86 chips (which were famous for bad I/O) have 4.2GB/sec-6.4GB/sec of memory bandwidth, the G4 is stuck in the '90s, with only 1.3GB/sec of memory bandwidth. On top of all that, the desktop version is expensive, and the high clock rate models are only available in an Apple. The G4 is great for a number of uses. XServe, for example, is a good use of the G4 because of its good heat dissapation characteristics. So is the Apple laptop line. But as a general purpose desktop CPU, it blows chunks.
There is a difference between putting an embedded processor in a mobile, small form factor machine, and putting it in a supposedly powerful desktop computer. Apple sucks because it does the latter. Intel rules because it's doing the former.
The game probably looks bad because coders aren't artists. Somebody had to do model those graphics, and I bet you they weren't a professional 3D modeler.
Many high-end laptops can be purchased with a UXGA 15" LCD. Dell ships one (branded UltraSharp) and IBM ships one (branded FlexView). I think the panels are made by Sharp and Hitachi. You can find some info on Alienware's Website.
Actually, in P6-class processors, the stack-based FPU is there only in the interface. Top of stack changes are free. In the Pentium 4, you should be using SSE2 (flat register file) anyway, given you give a damn about the interface at the ASM-level.
What matters, is that a P4 is performance competitive with CPUs costing an order of magnitude more. The P4 is the fastest processor for integer code. It's within 70% the performance of a Power4 in FPU. It has 4.2GB/sec of memory bandwidth. It costs less than $1000. MHz might not matter, but that kind of performance at that kind of price point certainly does!
Only if the data set fits in the L3 cache of the G4. The G4 extremely limited memory bandwidth, so while the AltiVec unit is theoretically 2-3 times as fast at a given clock speed (which makes it comparable to maybe 1.5 Xeons overall) it's totally demolished by the meager 1.3 GB/sec of main memory bandwidth. Data sets like these are extremely bandwidth limited. You know, SSE stands for "Streaming" SIMD Extensions for a reason.
Um, that's exactly what the Coyote Gulch benchmarks say! The whole point was that GCC's code generation was competitive with Intel's. In my mind, this is offset by the fact that it's slightly more conformant. Nobody claims that it's anywhere near competitive in terms of compilation speed.
1) Take a look at the polygon counts of a PS2 game. Even brand new games like Battlefield 1942 still have really blocky (polygon-wise) graphics. Until the average polygon is about the size of a pixel onscreen (taking into account LOD meshes) then the poly counnt will be high enough. 2) Try doing a realistic lighting engine. Even Doom III cheats. The best algorithms available right now for doing realistic shadows requires re-rendering the scene once from the perspective of each light. In a cathedral-type scene (with as many light sources as windows) would absolutely crawl.
Ugh. You use VC++ for template code? I feel for you. But I second the Intel error message thing. The EDG frontend has the best error messages around. GCC's has improved in 3.2, but I still use ICC to compile when I can't find out what's wrong.
I wouldn't accuse ICC of questionable shortcuts. It's generated code works just great. On the other hand, those benchmarks are quite out of date. GCC 3.2.x is *much* faster than GCC 2.95.x (or GCC 3.1.x for that matter). Instead look at these benchmarks which tests GCC 3.2.1 vs Intel C++ 7.0
There's no such thing as gcc 3.2.1-r6. The 'r6' is a Gentoo ebuild revision. GCC 3.2.1 works just fine. I'm on Gentoo, and I was using 3.2.1-r7 without any problems.
If you'd take a look at the Coyote Gulch GCC vs ICC benchmarks, you'd realize, that for P6 class processors (P3 and Athlon) GCC is right up there with ICC. It has some issues with P7 processors (Pentium 4) but it's only a matter of time before those are fixed. Now I don't know how well it works for non-x86 architectures (which Watcom and ICC don't support) but it's x86 code is damn good.
My friend doesn't waste time and energy investigating people who steal his money. Instead, he just earns more money >>>>>>>>>> Where does he live? Specific street address please. Maybe some landmarks nearby, you know, so I can find it. Thanks...
Slashdot Mirror
Have you seen most of America lately? It's Joe Beergut, more than anything else.
From dictionary.com
mootness n.
Usage Note: The adjective moot is originally a legal term going back to the mid-16th century. It derives from the noun moot, in its sense of a hypothetical case argued as an exercise by law students. Consequently, a moot question is one that is arguable or open to debate. But in the mid-19th century people also began to look at the hypothetical side of moot as its essential meaning, and they started to use the word to mean "of no significance or relevance." Thus, a moot point, however debatable, is one that has no practical value. A number of critics have objected to this use, but 59 percent of the Usage Panel accepts it in the sentence The nominee himself chastised the White House for failing to do more to support him, but his concerns became moot when a number of Republicans announced that they, too, would oppose the nomination. When using moot one should be sure that the context makes clear which sense is meant.
but the data isn't getting onto or off the chips at anything like the speed required to keep up.
>>>>>>>>>
First, we're talking about single CPUs here, not whole systems.
Second, your comments haven't been true for a couple of years. The memory subsystems of current PC chips are very good. Good enough that they're entirely competitive with Sun or SGI machines in the same class (2-4 processors). Beyond that, the Sun or SGI machines might scale better, but the overall performance of the PC in its "weight class" (small SMP machines) is unmatched. Nobody thinks that a PC architecture machine will match up to a Power4 in the Power4's normal configuration (16 or 32 way) but it is true that a P4 is on the heels of a single Power4 CPU.
mute point :)
>>>>>>>>>
That's 'moot' point. But given the topic of discussion, that's a very interesting play on words
Your explanation is dead on, your nomenclature is reversed. Current CD drives are CAV (Constant Angular Velocity) while before they were CLV (Constant Linear Velocity).
These disks spin at constant angular velocity. That means the oustide spins faster than the inside.
Do you understand the concept of hyperbole? It was precisely because of Motorola's focus on the embedded market that I called the G4 an embedded processor. Note that I called Banias an embedded processor as well, which it clearly is not.
Next, the 2 and 3 GHz marks are not useless. The thing PowerPC people don't understand is that Performance = ClockRate * Efficiency. Note that both are linear factors. Raising the effifiency does not improve the overall performance any more than raising clock rate. However, raising ClockRate is a lot easier than raising efficiency, because most code (outside of a few problem domains) does not lend itself to extensive parallization. Intel, with the P4 architecture, made an engineering decision to emphasise clock rate over efficiency. As the benchmarks clearly show, Intel achieved their goal of having absolutely incredible performance, to the point where a P4 provides more than 70% the floating point performance of a Power4 in CPU-bound benchmarks. PowerPC-heads can wave their arms and shout "efficiency" as much as they want, all that matters is the end result performance.
Lastly, the G4 is a sucky chip. Its clock rates are lower than its x86 competitors. It's IPC is lower than chips like the Alpha or POWER. It's overall performance isn't very good. In an age when x86 chips (which were famous for bad I/O) have 4.2GB/sec-6.4GB/sec of memory bandwidth, the G4 is stuck in the '90s, with only 1.3GB/sec of memory bandwidth. On top of all that, the desktop version is expensive, and the high clock rate models are only available in an Apple. The G4 is great for a number of uses. XServe, for example, is a good use of the G4 because of its good heat dissapation characteristics. So is the Apple laptop line. But as a general purpose desktop CPU, it blows chunks.
There is a difference between putting an embedded processor in a mobile, small form factor machine, and putting it in a supposedly powerful desktop computer. Apple sucks because it does the latter. Intel rules because it's doing the former.
The game probably looks bad because coders aren't artists. Somebody had to do model those graphics, and I bet you they weren't a professional 3D modeler.
Many high-end laptops can be purchased with a UXGA 15" LCD. Dell ships one (branded UltraSharp) and IBM ships one (branded FlexView). I think the panels are made by Sharp and Hitachi. You can find some info on Alienware's Website.
The P4 can handle 64-bit SIMD math just fine.
Actually, in P6-class processors, the stack-based FPU is there only in the interface. Top of stack changes are free. In the Pentium 4, you should be using SSE2 (flat register file) anyway, given you give a damn about the interface at the ASM-level.
What matters, is that a P4 is performance competitive with CPUs costing an order of magnitude more. The P4 is the fastest processor for integer code. It's within 70% the performance of a Power4 in FPU. It has 4.2GB/sec of memory bandwidth. It costs less than $1000. MHz might not matter, but that kind of performance at that kind of price point certainly does!
Only if the data set fits in the L3 cache of the G4. The G4 extremely limited memory bandwidth, so while the AltiVec unit is theoretically 2-3 times as fast at a given clock speed (which makes it comparable to maybe 1.5 Xeons overall) it's totally demolished by the meager 1.3 GB/sec of main memory bandwidth. Data sets like these are extremely bandwidth limited. You know, SSE stands for "Streaming" SIMD Extensions for a reason.
Um, that's exactly what the Coyote Gulch benchmarks say! The whole point was that GCC's code generation was competitive with Intel's. In my mind, this is offset by the fact that it's slightly more conformant. Nobody claims that it's anywhere near competitive in terms of compilation speed.
1) Take a look at the polygon counts of a PS2 game. Even brand new games like Battlefield 1942 still have really blocky (polygon-wise) graphics. Until the average polygon is about the size of a pixel onscreen (taking into account LOD meshes) then the poly counnt will be high enough.
2) Try doing a realistic lighting engine. Even Doom III cheats. The best algorithms available right now for doing realistic shadows requires re-rendering the scene once from the perspective of each light. In a cathedral-type scene (with as many light sources as windows) would absolutely crawl.
Dude. I run 1600x1200 on a 15" LCD, and I can still see the pixels. Maybe when things hit 200-300dpi...
Gentoo Portage Directory -- 157 megabytes.
FreeBSD 5.0 ISO - 650 MB.
Remote Internet Backup -- 2GB.
An ISP that doesn't cap usage -- Priceless.
Ugh. You use VC++ for template code? I feel for you. But I second the Intel error message thing. The EDG frontend has the best error messages around. GCC's has improved in 3.2, but I still use ICC to compile when I can't find out what's wrong.
You mean this link? The one that shows GCC matching Intel C++ 7.0 on everything except the P4 FPU benchmarks?
I meant these!
I wouldn't accuse ICC of questionable shortcuts. It's generated code works just great. On the other hand, those benchmarks are quite out of date. GCC 3.2.x is *much* faster than GCC 2.95.x (or GCC 3.1.x for that matter). Instead look at these benchmarks which tests GCC 3.2.1 vs Intel C++ 7.0
There's no such thing as gcc 3.2.1-r6. The 'r6' is a Gentoo ebuild revision. GCC 3.2.1 works just fine. I'm on Gentoo, and I was using 3.2.1-r7 without any problems.
If you'd take a look at the Coyote Gulch GCC vs ICC benchmarks, you'd realize, that for P6 class processors (P3 and Athlon) GCC is right up there with ICC. It has some issues with P7 processors (Pentium 4) but it's only a matter of time before those are fixed. Now I don't know how well it works for non-x86 architectures (which Watcom and ICC don't support) but it's x86 code is damn good.
My friend doesn't waste time and energy investigating people who steal his money. Instead, he just earns more money
>>>>>>>>>>
Where does he live? Specific street address please. Maybe some landmarks nearby, you know, so I can find it. Thanks...