It's nonsense, frankly.Macinposh said:What do you think about the rumours that a single socket Conroe thanks to it´s superior memory handling effiency (~70%?) compared to Xeons DB-Dimm´s lousy (~25?) might crush a dual socket Xeon in memory intesive tasks, like photoshop.
Have you heard seen any data on that one,exept the specuatlion on Anandtech?
It is speculation (or FUD) based on a single facet of the system design, without considering all of the factors that affect real application performance. (Remember when Apple was whining about the horrible "pipeline bubble" problem with the Pentium - meanwhile Pentiums and Xeons were benchmarking just fine against the Apples?)
Dell has submitted SPEC results for Woodcrest and Conroe systems, and they're virtually the same - with a very slight advantage to Conroe.
Code:
System CPU GHz Cores Sockets CINT CFP CINTRate CFPRate
PowerEdge 1950 Xeon 5110 1.60 4 2 1772 1719 72.0 60.9
PowerEdge 1950 Xeon 5120 1.86 4 2 1974 1979 82.1 65.0
PowerEdge 1950 Xeon 5130 2.00 4 2 2154 2130 88.6 71.2
PowerEdge 1950 Xeon 5140 2.33 4 2 2394 2451 101.0 75.9
PowerEdge 1950 Xeon 5150 2.66 4 2 2615 2761 111.0 79.8
PowerEdge 1950 Xeon 5160 3.00 4 2 2818 3061 120.0* 83.4
PrecWrkstn 390 Core 2 Duo E6300 1.86 2 1 1978 1939 39.5 35.0
PrecWrkstn 390 Core 2 Duo E6400 2.13 2 1 2168 2187 44.0 37.1
PrecWrkstn 390 Core 2 Duo E6600 2.40 2 1 2497 2588 53.7 43.9
PrecWrkstn 390 Core 2 Duo E6700 2.66 2 1 2673 2844 58.7 45.6
PrecWrkstn 390 Core 2 Extreme X6800 2.93 2 1 2872 3108 63.5 47.9
* PE1955
INT = "CINT2000" CFP = "CFP2000"
http://www.spec.org/cpu2000/results/res2006q3/
PE 1950 specs (5000X chipset, like the ProMacIntel)
PW 390 specs (975X chipset)
PE 1950 specs (5000X chipset, like the ProMacIntel)
PW 390 specs (975X chipset)
The Xeon 5160 is faster than any Core 2 Duo.
Note that CINT/CFP are single thread tests, and the two "Rate" tests are multi-threaded (so you see the quad core Woody at about double the dual core Conroe).
Also note that CFP is a *very* heavy memory test - you can't get a good FP score with a lousy memory system.
Clearly the faster bus speed that FB-DIMMs can use compensates for the "lousy efficiency".
One point with FB-DIMMs is to reduce the number of FB-DIMMs that you use. 8 GiB with four 2 GiB FB-DIMMs will be slightly faster than 8 GiB with eight 1 GiB FB-DIMMs. Very slightly faster.Macinposh said:Any idea if the upcoming products (CS3 for example) might find a way to utilize the FB-Dims more efficiently, or is the problems so prevalent,that it cant be overcome with anything?
There were a number of areas where the late G3 systems were faster than the new G4 systems - only on AltiVec apps was the G4 clearly better. The speed advantage of the Conroe is much less than the G3 had.Macinposh said:Apple definately wouldn want a cheaper/weaker product to equal or crush it´s workstations in any area. Let alone on one that is considered it is pride,DTP.
In addition, the Woody will crush the Conroe on any app that can use more than two cores.
Don't worry, the New Form-Factor Conroe Mini-Tower/Home-Theatre Mac® will be here soon.Macinposh said:Because that might be the deathblow to the Pizza-Mac.