I came across a site where these guys wrote a CPU neutral benchmark called geekbench, and took the test on both my PowerBook G4 1.5Ghz, and my MacPro 3.0...
got 40 ont the PB, and 367 on the MacPro,,, can it really be that much faster than a 1.5 G4??
Riser A, Slots 1&2 - 1 GB (where the Apple RAM was when you got it)
Riser A, Slots 3&4 - 512 MB
Riser B, Slots 1&2 - 1 GB
Note that since my Mac Pro isn't here yet, I don't know which way is the "front" on the riser. It should be the slots the Apple RAM came in. you referred to it as "inboard" and "outboard" or something which is meaningless to me. Also, Riser A versus Riser B shouldn't matter, except I think that one is slightly cooler.
It seems hard to have a cross-platform CPU-only benchmark. I mean, some processors are better at some things than others, and there's also the compiler issue to worry about. not to mention favoring processors with higher cache over extra SIMD units or other measurements. I mean the best way to do it is to look at individual scores, figure out what your needs out of the processor are, and go with that.
by inboard, I meant closest to the mother board... but they're actually numbered 1 thru 4 so I assume 1 is the closest.
the harder part is determining which is Riser A and which is B.. I assume top is A and bottom is B..
Now, if I added the 2x512.. should I mix with the 1GB's or do the 512's on one riser and the 1GB's on the other?
The one the RAM came in is Riser A (the top one). that shouldn't matter much though, since the 256-bit bus is split half to each riser. Put the 3 GB in the cooler riser (although it's all well-sinked so it shouldn't matter)
When you move to 6 GBs, you want to remember the 2 Rules. Keep it even across the risers (so 2x1GB + 2x512 in each one) and keep the biggest RAM in the first pair on each riser.
The channels are independent in a sense. The RAM is split along each Riser until one riser is maxed out, then all the rest of the work is assigned to the other riser. So your orginal set-up would have had 2 GBs of RAM with the work split evenly, then the last 3 GBs would have been all along the first two channels (to the first riser), cutting into your bandwidth.
Also, the way FB-DIMMs work, they're sort of chained. Work headed to Riser A goes to FB-DIMMs 1 & 2 - if they're full, they pass it along to FB-DIMMs 3 & 4. The act of passing it along like that introduces more latency. You had 2 GBs of RAM that was being hit by that latency penalty, and now only 1 GB is.
got 40 ont the PB, and 367 on the MacPro,,, can it really be that much faster than a 1.5 G4??
Geekbench is a CPU and memory benchmark; the Mac Pro has a huge advantage over the PowerBook since it has a much more modern processor, has more processor cores, and a significantly faster memory bus than the PowerBook.
That said, your PowerBook score is surprisingly low. Your score should be at least 50% higher. Was your PowerBook under any sort of load when you ran Geekbench?
Comments
Processor - Core architecture kills G4s clock for clock, and you have 8x the clock, and at least 4 times the FSB.
HDDs - 10k SATA Raptor + 500GB 7200RPM RAID 0 versus some 2.5" 5400RPM PATA drive? Not even close.
Video Card - x1900xt versus anything mobile is a slaughter. The PB has what, a 9600?
RAM - your RAM is configured wrong, but even with that, you have 5 GBs to 2 GBs, and you have (partial) quad-channel as well as higher clock rates.
I think Geekbench is CPU only bench, but I haven't figured it out for sure...
I took your advise and re-did the RAM and it made a difference on Xbench... read my post and see if I did it correctly...
thanks again
Riser A, Slots 1&2 - 1 GB (where the Apple RAM was when you got it)
Riser A, Slots 3&4 - 512 MB
Riser B, Slots 1&2 - 1 GB
Note that since my Mac Pro isn't here yet, I don't know which way is the "front" on the riser. It should be the slots the Apple RAM came in. you referred to it as "inboard" and "outboard" or something which is meaningless to me. Also, Riser A versus Riser B shouldn't matter, except I think that one is slightly cooler.
It seems hard to have a cross-platform CPU-only benchmark. I mean, some processors are better at some things than others, and there's also the compiler issue to worry about. not to mention favoring processors with higher cache over extra SIMD units or other measurements. I mean the best way to do it is to look at individual scores, figure out what your needs out of the processor are, and go with that.
the harder part is determining which is Riser A and which is B.. I assume top is A and bottom is B..
Now, if I added the 2x512.. should I mix with the 1GB's or do the 512's on one riser and the 1GB's on the other?
by inboard, I meant closest to the mother board... but they're actually numbered 1 thru 4 so I assume 1 is the closest.
the harder part is determining which is Riser A and which is B.. I assume top is A and bottom is B..
Now, if I added the 2x512.. should I mix with the 1GB's or do the 512's on one riser and the 1GB's on the other?
The one the RAM came in is Riser A (the top one). that shouldn't matter much though, since the 256-bit bus is split half to each riser. Put the 3 GB in the cooler riser (although it's all well-sinked so it shouldn't matter)
When you move to 6 GBs, you want to remember the 2 Rules. Keep it even across the risers (so 2x1GB + 2x512 in each one) and keep the biggest RAM in the first pair on each riser.
The channels are independent in a sense. The RAM is split along each Riser until one riser is maxed out, then all the rest of the work is assigned to the other riser. So your orginal set-up would have had 2 GBs of RAM with the work split evenly, then the last 3 GBs would have been all along the first two channels (to the first riser), cutting into your bandwidth.
Also, the way FB-DIMMs work, they're sort of chained. Work headed to Riser A goes to FB-DIMMs 1 & 2 - if they're full, they pass it along to FB-DIMMs 3 & 4. The act of passing it along like that introduces more latency. You had 2 GBs of RAM that was being hit by that latency penalty, and now only 1 GB is.
I'll keep passing it on to others.. I'm now in a search for two more 512's.. seems like all they sell is 1GB's everywhere..
http://www.crucial.com/store/MPartsp...&WSPN=CT579441
got 40 ont the PB, and 367 on the MacPro,,, can it really be that much faster than a 1.5 G4??
Geekbench is a CPU and memory benchmark; the Mac Pro has a huge advantage over the PowerBook since it has a much more modern processor, has more processor cores, and a significantly faster memory bus than the PowerBook.
That said, your PowerBook score is surprisingly low. Your score should be at least 50% higher. Was your PowerBook under any sort of load when you ran Geekbench?