Ableton Forum

pure conjecture here folks but...

Would you buy a G5 1.6ghz Powerbook if it was released rather than the latest G4 "speed-bumps" ?

bare with me here (entering mumbo-jumbo mode),
by the Live 4 Performance test , the G5 single 1.6ghz iMac performs a 37% to 39%. Now, no 1.67ghz G4 powerbooks have yet tested, but the 1.5ghz has tested with an average of 43 to 48%. Hypothetically, the latest G4 1.67ghz Powerbooks should perform nearly identically as effecient as the G5 1.6ghz iMac. you gotta be wondering WHY? the G5 has a 533mhz FSB, and the G4 only a 167mhz FSB. Well, did you know that the L2 cache on the G4 runs at full cpu speed, where as the L2 cache on the G5 runs only at 1.15ghz? Neither did i, till i dug a little deeper.

This makes the latest G4 1.67ghz Powerbook as fast as a would be G5 1.6ghz Powerbook. Yet so many people are miffed that its a G4.

Now, continuing on the hypothetical jive... lets say that Ableton (with regard to our endless requests for it), optimized Live 4 for altivec instructions. The altivec engine on the G4 generates upward of 6 Gflop/sec , where as the G5 generates 3.2 Gflop/sec (averages). This throughput would make the G4 FASTER in altivec optimized applications (that aren't also 64 bit optimized, see: Adobe CS) than the G5, clock for clock.

aint that something?
too bad that bastard starts at $2299

This is exactly why even with macs it pays to do your homework before buying a machine for audio. The best example I can give is the 1st generation G4 powerbooks, they did really well in Logic tests with reverb stacking as a CPU stress indicator. In fact the 400mhz did much better than the 550mhz that came out a year later. Tests indicated that the 400 was performing at roughly the same level as the 667mhz? No third level cache, and a CPU design that didn't work well for audio in that years model.

What's cool about this to me is that if you are right. and the G5 powerbook is mediocre in Live 4 tests, then a used or refurbished 1.6 will be cheap. I did exactly that with the 800mhz I own, waited until the next speed bump, saw some lackluster improvements in audio, and bought a refurbished machine for almost half what people were paying for the 867mhz.

The other consideration is that the iMac is a consumer model machine, though it might be a G5, it shouldn't compare to the power mac line, or powerbook line, though sometimes they do slip a bit.

AdamJay wrote:pure conjecture here folks but...

Would you buy a G5 1.6ghz Powerbook if it was released rather than the latest G4 "speed-bumps" ?

bare with me here (entering mumbo-jumbo mode),
by the Live 4 Performance test , the G5 single 1.6ghz iMac performs a 37% to 39%. Now, no 1.67ghz G4 powerbooks have yet tested, but the 1.5ghz has tested with an average of 43 to 48%. Hypothetically, the latest G4 1.67ghz Powerbooks should perform nearly identically as effecient as the G5 1.6ghz iMac. you gotta be wondering WHY? the G5 has a 533mhz FSB, and the G4 only a 167mhz FSB. Well, did you know that the L2 cache on the G4 runs at full cpu speed, where as the L2 cache on the G4 runs only at 1.15ghz? Neither did i, till i dug a little deeper.

This makes the latest G4 1.67ghz Powerbook as fast as a would be G5 1.6ghz Powerbook. Yet so many people are miffed that its a G4.

Now, continuing on the hypothetical jive... lets say that Ableton (with regard to our endless requests for it), optimized Live 4 for altivec instructions. The altivec engine on the G4 generates upward of 6 Gflop/sec , where as the G5 generates 3.2 Gflop/sec (averages). This throughput would make the G4 FASTER in altivec optimized applications (that aren't also 64 bit optimized, see: Adobe CS) than the G5, clock for clock.

aint that something?
too bad that bastard starts at $2299

Very interesting points, AdamJay, Thanks a lot.

R.

AdamJay wrote:did you know that the L2 cache on the G4 runs at full cpu speed, where as the L2 cache on the G4 runs only at 1.15ghz?

Hi Adam, sorry to be a picky bastard, I just want to get my facts straight, the second G4 in the above sentence should read G5, right?

sorry to be a pedant, mate. Very very interesting nonetheless

-paddy

Pitch Black wrote:

AdamJay wrote:did you know that the L2 cache on the G4 runs at full cpu speed, where as the L2 cache on the G4 runs only at 1.15ghz?

Hi Adam, sorry to be a picky bastard, I just want to get my facts straight, the second G4 in the above sentence should read G5, right?

sorry to be a pedant, mate. Very very interesting nonetheless

-paddy

correct, i'll edit my above post. thanks for the proofread paddy!

i have resisted this whole time as i think your benchmark file is designed against the ableton's own performace recommendations, but out of curiousity i am going to try it out on my ibook in a minute. i have never maxxed it out with what i do so the test is useless to me.

just as I thought, getting rid of live's reverb and using a free audio unit reverb on the send channel shaved off a whopping 30% on my cpu usage.

ejectorset wrote:i have resisted this whole time as i think your benchmark file is designed against the ableton's own performace recommendations, but out of curiousity i am going to try it out on my ibook in a minute. i have never maxxed it out with what i do so the test is useless to me.

against their own performance recommendations?
how do you mean?

1 Simpler Instrument with Filter & LFO activated
8 sounds in Impulse, all with Filters activated, sent to 8 individual channels for processing.
8 EQ4's, 1 CompressorI, 1 EQ3, 1 Chrous, 1 Redux, 1 Erosion
in the FX Returns - 1 Reverb (Comfort Setting), 1 Simple Delay
in the Master - 1 Compressor 2

that is a pretty standard .als there. designed after the release of version 4 to utilize the two new ableton instruments with an EQ, and at least 1 of various other FX on each channel. All ableton instruments and fx, no third party plug-ins.

you may be misinterpretting the test. The test isn't to see if you could use ableton with your computer effeciently (which obviously you can). The test is a baseline for comparison of multiple machines, simple as that. Please try not to read into any more than that.

ejectorset wrote:just as I thought, getting rid of live's reverb and using a free audio unit reverb on the send channel shaved off a whopping 30% on my cpu usage.

yep, i also recommended that IN the Live 4 Performance test, back in September.

http://www.ableton.com/forum/viewtopic. ... verb#66920

also guys, as a mac tech, from a design stand point, if you want a stable powerful mac, these are these last G4s are the ones to go for. the end of Apples production runs have always historically been the most stable useable problem free models apple makes.

interesting that in a recent bench mark test the 12" PB has taken a bigger performance jump then it's other two brethren.

bencodec wrote:interesting that in a recent bench mark test the 12" PB has taken a bigger performance jump then it's other two brethren.

any idea why that would be?

[quote="AdamJay"]pure conjecture here folks but...

bear with me here (entering mumbo-jumbo mode),
by the Live 4 Performance test , the G5 single 1.6ghz iMac performs a 37% to 39%. Now, no 1.67ghz G4 powerbooks have yet tested, but the 1.5ghz has tested with an average of 43 to 48%. Hypothetically, the latest G4 1.67ghz Powerbooks should perform nearly identically as effecient as the G5 1.6ghz iMac. you gotta be wondering WHY? the G5 has a 533mhz FSB, and the G4 only a 167mhz FSB. Well, did you know that the L2 cache on the G4 runs at full cpu speed, where as the L2 cache on the G5 runs only at 1.15ghz? Neither did i, till i dug a little deeper.

This makes the latest G4 1.67ghz Powerbook as fast as a would be G5 1.6ghz Powerbook. Yet so many people are miffed that its a G4.

Now, continuing on the hypothetical jive... lets say that Ableton (with regard to our endless requests for it), optimized Live 4 for altivec instructions. The altivec engine on the G4 generates upward of 6 Gflop/sec , where as the G5 generates 3.2 Gflop/sec (averages). This throughput would make the G4 FASTER in altivec optimized applications (that aren't also 64 bit optimized, see: Adobe CS) than the G5, clock for clock.

I am curious - where did you get your performance data for the bus speeds and the altivec performance? I can't find it on the Apple or the MacSpeedZone sites.

Do you know the width of the memory busses? I have noticed in the past that the PowerMacs and PowerBooks tend to have wider busses (e.g. 128 bits instead of 64) than the ibooks and imacs. The true measure of the memory bus performance is not the clock rate (e.g. 533 MHz), but the bandwidth (e.g. 4 G bytes/sec).

Also - AFAIK, there is not that much difference between the G4 and the G5 - the biggest diff is that the G5 is a 64-bit machine - so for the most part, I'd *expect* the G4s and G5s to run at about the same speed (all else - memory busses and cache sizes - being equal). Of course, one would expect to see some performance improvement on G5 machines when running apps that are written to take advantage of the wider registers and bigger address space.

xbench.com
i simply looked at the trends of various g4 1.67ghz vs. g5 1.6ghz submitted benchmarks for altivec performance.
cache speed is there too.

Great post AdamJay!! Thanks!!

any one heard of the Musicxpc ?? www.musicxpc.com
check it! specially made for music.