All is well.
As I said, I ended up ordering two of the dual G4s, in part to have a spare, and in part because they were cheap and the shipping was about the same for two as for one. Tonight I transplanted the "spare" one into a 400MHz AGP tower running X.3 Server. I ran XBench on the tower before and after the processor swap.
In one sense, benchmarks are meaningless because who knows which of the things they measure are being called upon in ordinary usage. However, so far as it goes, here is what I came up with:
A 500MHz processor is 25% faster than a 400MHz processor. XBench measures performance across a range of components, from the CPU to memory to Graphics. The Graphics card is a basic Rage 128 Pro AGP in this case. The machine has 1.25GB RAM.
The results were occasionally predictable, and occasionally more surprising. The CPU was anything from 26% to 41% faster, depending on the individual test in question; threads were from 67% to 145% faster; even memory was from 2% to 28% faster across various tests in which it was accessed; the graphics were from 24% to 33% faster, which surprised me to no end; and on what XBench calls the User Interface Test, the dual G4 at 500MHz was 61% faster than the single G4 400MHz machine. I am not sure what that involved but it sure looks good.
All of this means only as much as an XBench test is worth, I suppose, but these are tentative results.
When I get a few spare hours, I will be popping the other dual G4 card into my Cube and will try to run a wider range of benchmarking software to gauge the effect. The Cube requires serious surgery for this work, which involves some modest hackery. Hopefully I will get a smoother user experience; a 61% improvement would suit me just fine! (I won't get it, of course, but we live in hope.)
The real proof of the pudding has to be gained from actual use, and that will need a bit of time spent actually using the machine on long, complex documents and the like. I'll report back one of these days on the results.
As I said, I ended up ordering two of the dual G4s, in part to have a spare, and in part because they were cheap and the shipping was about the same for two as for one. Tonight I transplanted the "spare" one into a 400MHz AGP tower running X.3 Server. I ran XBench on the tower before and after the processor swap.
In one sense, benchmarks are meaningless because who knows which of the things they measure are being called upon in ordinary usage. However, so far as it goes, here is what I came up with:
A 500MHz processor is 25% faster than a 400MHz processor. XBench measures performance across a range of components, from the CPU to memory to Graphics. The Graphics card is a basic Rage 128 Pro AGP in this case. The machine has 1.25GB RAM.
The results were occasionally predictable, and occasionally more surprising. The CPU was anything from 26% to 41% faster, depending on the individual test in question; threads were from 67% to 145% faster; even memory was from 2% to 28% faster across various tests in which it was accessed; the graphics were from 24% to 33% faster, which surprised me to no end; and on what XBench calls the User Interface Test, the dual G4 at 500MHz was 61% faster than the single G4 400MHz machine. I am not sure what that involved but it sure looks good.
All of this means only as much as an XBench test is worth, I suppose, but these are tentative results.
When I get a few spare hours, I will be popping the other dual G4 card into my Cube and will try to run a wider range of benchmarking software to gauge the effect. The Cube requires serious surgery for this work, which involves some modest hackery. Hopefully I will get a smoother user experience; a 61% improvement would suit me just fine! (I won't get it, of course, but we live in hope.)
The real proof of the pudding has to be gained from actual use, and that will need a bit of time spent actually using the machine on long, complex documents and the like. I'll report back one of these days on the results.