183,346 users have contributed to 42,293 threads and 255,053 posts.
In the past 24 hours, we have 4 new thread(s), 15 new post(s) and 52 new user(s).
the benchmark and bandwidth tests do surprise me to say the least ... i'll try to perform a VI loading test tomorrow with the i7 X58 board (2 vs. 3 channels)
btw: has anybody read about the used chipset?
and what just came into my mind ... is there something like hyperthreading existing in OS X? because interestingly the i7 with active hyperthreading turned up to be overall more performant than without ... this might be related to the design of the calculating units ont the processor ... obviously intel did their homework this time ...
christian
@julian said:
When I purchased my MacPro the top clock speed option (3.2) was around an extra £500 but with the new MacPro choose the highest speed and the price ramps more than the cost of a complete entry level MacPro. That's a bit steep isn't it?
Mine is 3.0 the same as yours only slightly slower clock. I guess they are thinking that the new processors are way above the old ones in terms of performance and that may very well be so. But for musical applications I don't think you would even notice it. To me, it's always been a ram and 64 bit thing. Therefore Julian, I would agree - that's a bit steep.
Stephane,
Are you saying that a 2.66 Quad Core I7 will provide more performance than the last generation (Harpertown) Dual Quad Core Xeon 2.66?
cm,
I am not sure what you are confirming. You mean that x58 1 x 4 x 3.2 GHz i7 12GB 1333 MHzi provides more performance than 5400 2 x 4 x 3.2 GHz XEON 16GB 1600 MHz ?
My understanding regarding the new Macs is:
If you have 3 DIMMs or 6 DIMMs, you get tri-channel modules (ca. 19GB/sec).
If you have 2 DIMMs, 4 DIMMs or 8 DIMMs, you get dual-channel (ca. 13GB/sec).
Maybe someone else can check on ECC, but I think at least some of these models are ECC (after all they are servers) - though no promises from me on this.
Do we actually care about the memory speed? I've never heard of that being a bottleneck, but is it one?
Good point. I saw a DAWBench performance graph comparing one new 2.66 I7 to a PAIR of last generation Xeon 2.66 chips (a 5350?), and they were comparable until you got to low latencies (64 and 128), where the I7 really blew the pair of Xeons away.
Sorry, I don't believe that memory speed has any noticeable effect whatsoever on the size of the buffer you can get away with. In my experience even the hard drives, e.g. 10K RPM Raptors, don't make much difference. Maybe if you're running MIR this kind of thing will make a difference, but for regular sampling I'm highly skeptical.<BR><BR>
The thing is, you can't run machines that close to the breaking point when you're using them for music. You have to leave some overhead just to operate the sequencer.
Hi Nick,
My only points about bringing the speed were indeed corollary to the discussion on the other VSL forum regarding the new Mac Pro's working for Mir, and replying to an earlier posting on this thread regarding the number of slots and the corresponding amount of RAM to be used.The numbers of slots and the amount do add up the way I mentioned and the speed of RAM differs accordingly - whether or not that relates directly to this discussion.
Yes, I certainly agree with your comments on RAM speed generally being adequate for sample playback - and the importance of high disk rotation speeds.
Best regards,
Jack
